Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

TL;DR

• Meta Platforms confirmed on July 13, 2026 that its Hyperion campus in Richland Parish, Louisiana will scale to 5 gigawatts across 3,200 acres, with total investment exceeding $50 billion. The project began as a $10 billion, roughly 4-million-square-foot facility at groundbreaking in December 2024 and is now Meta’s largest data center, with the first 2 GW expected online by 2030, the full 5 GW around 2032, and completion by 2036.

• Hyperion is financed through a joint venture with Blue Owl Capital — the largest private capital transaction on record for a single data center — structured to hold the development off Meta’s balance sheet. Blue Owl’s special purpose vehicle owns 80 percent and Meta 20 percent, on a roughly $28.8 billion committed capital base including $27.3 billion of SPV-issued debt arranged by Morgan Stanley as sole bookrunner, priced at roughly 225 basis points over Treasuries, rated A+ by S&P, and anchored by Pimco with BlackRock participating; Meta’s rent begins June 1, 2029, backstopped by a payment guaranty and a Residual Value Guarantee.

• Entergy Louisiana is building 10 natural gas-fired power plants delivering more than 7 gigawatts for Hyperion — an increase of more than 30 percent to Louisiana’s entire grid capacity. Meta is separately funding up to 2.5 GW of renewable generation, roughly 240 miles of transmission, grid-scale battery storage, and nuclear plant uprates, while approval from the Louisiana Public Service Commission for the gas plants remains pending and will determine delivery speed.

Meta Platforms confirmed on July 13, 2026 that its Hyperion data center campus in Richland Parish, Louisiana will scale to 5 gigawatts across 3,200 acres, with total investment exceeding $50 billion.

The expansion is financed through a joint venture with Blue Owl Capital structured to hold the development off Meta’s balance sheet.

The capital funds a supercluster in the Holly Ridge community that began as a $10 billion project at groundbreaking in December 2024.

Meta expects the first 2 GW online by 2030, the full 5 GW around 2032, and total project completion by 2036.

The Escalation From $10 Billion

Meta broke ground on Hyperion in December 2024 with an initial $10 billion budget for a roughly 4-million-square-foot facility.

In October 2025, Meta formed the Blue Owl joint venture targeting up to $27 to $28.8 billion in total development costs for a 2.06 GW facility.

By February 2026, Meta had acquired roughly 1,400 additional adjacent acres, signaling a phase-two expansion.

The July 2026 confirmation brought the campus to 5 GW and more than $50 billion, making it Meta’s largest data center.

President Trump previewed the $50 billion figure as early as August 2025.

The Blue Owl Joint Venture Structure

The transaction is the largest private capital transaction on record for a single data center.

Blue Owl’s special purpose vehicle owns an 80 percent interest in the joint venture, and Meta retains 20 percent.

The joint venture owns the landlord entity, and Meta operates as tenant.

Meta contributed $5.7 billion in equity against Blue Owl’s $23.0 billion, funded through $27.3 billion in SPV-issued debt, $2.5 billion in equity, and $1.2 billion in accrued interest income, a total committed capital base of roughly $28.8 billion.

Morgan Stanley arranged the debt as sole bookrunner, pricing 144A bonds maturing in 2049 at roughly 225 basis points over Treasuries.

S&P assigned an A+ rating, one notch below Meta’s AA- corporate rating.

Pimco anchored the debt as lead investor, with BlackRock and other institutional credit investors participating.

The Lease And Guarantee Terms

Meta’s rent obligations begin June 1, 2029 and must be paid in full even amid construction delays, with narrow abatements for force majeure or landlord default.

Creditors have no direct recourse to the underlying data center assets.

Meta backstops the arrangement with a payment guaranty and a Residual Value Guarantee designed to cover outstanding SPV debt at the end of each four-year renewal term.

The Power Buildout

Entergy Louisiana is building 10 natural gas-fired power plants delivering more than 7 gigawatts of capacity, a buildout representing more than a 30 percent increase to Louisiana’s entire grid capacity.

Meta is funding up to 2.5 GW of renewable generation, roughly 240 miles of new transmission lines, grid-scale battery storage, and nuclear plant uprates.

Meta states it is paying the full costs of the energy, water, and related infrastructure, and Entergy projects more than $2 billion in customer savings over 20 years.

Regulatory approval from the Louisiana Public Service Commission for the additional gas plants remains pending and will determine how quickly Entergy delivers the capacity Hyperion requires.

Join Global Data Center Hub

Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

In this week’s issue:

First, the builders started selling. Bankers are marketing majority stakes in DataBank, EdgeCore, Netrality, and Edged to private equity, with the DataBank process alone reaching up to ~$25 billion, reportedly.

When the operators closest to build economics are the ones cashing out, what does their exit price assume that your entry price does not?

Then, the public market opened its bid. Switch tapped Goldman Sachs and JPMorgan for an IPO valuing it near $80 billion including debt, seven times its 2022 take-private mark, while Brookfield-backed Csquare debuted on the NYSE at $3.2 billion.

Finally, power became equity. Blackstone, Apollo, and KKR committed $5.3 billion for 49 percent of five behind-the-meter plants Williams is building for data centers, the clearest sign yet that secured megawatts are bought, not contracted.

Let us get into it.

THIS WEEK BY REGION

The week’s biggest moves — what happened and what it signals.

North America

This week, the market tested data center valuations across both private and public markets. Bankers launched sales of DataBank, EdgeCore, Netrality, and Edged, with the DataBank process reportedly reaching ~$25 billion, while Switch pursued an IPO valuing it at ~$80 billion, including debt.

On the power side, Blackstone, Apollo, and KKR invested $5.3 billion in behind-the-meter generation, Meta expanded its 5 GW Hyperion campus, and New York became the first U.S. state to impose a statewide data center moratorium.

For infrastructure funds and public-markets investors, the entry multiple is now the whole trade, and the operator selling into strength has better information than the buyer stepping in.

Asia-Pacific

Capital is strengthening the infrastructure underpinning AI growth. Symphony Communication and Nokia are upgrading a key Southeast Asian subsea cable network, while India's AI infrastructure buildout is gaining momentum.

Across both markets, the opportunity lies in funding cables, land, and power ahead of demand.

For tech and infrastructure investors, the seabed and the interconnect are where position is won in Asia this year, well before the marquee campus announcements land.

Europe

Europe is increasingly being underwritten by hyperscaler demand before projects are built. Pure DC secured Microsoft as the anchor tenant for a new AI data center in Finland, making the project financeable before construction.

Meanwhile, Brookfield-backed Csquare debuted on the NYSE at a $3.2 billion valuation, reflecting investor demand for future capacity in the power-rich, low-cost Nordic market.

For real estate and infrastructure investors, the signed hyperscaler offtake is the instrument that turns a European development into a financeable asset, and the projects without one wait.

NOTABLE TRANSACTIONS

Key structures and capital moves from this week’s deal tape.

Blackstone, Apollo, and KKR: $5.3 billion for 49% of Williams’ behind-the-meter power portfolio

The consortium invested $5.3 billion for a 49% stake in five 2.6 GW behind-the-meter gas plants, treating onsite generation as an equity asset rather than a power supply contract.

For the financing landscape, the mechanism is power-as-equity, giving credit and insurance capital ownership of scarce generation while returning long-term control to the developer.

Switch: IPO valuing the operator near $80 billion including debt

Switch has engaged Goldman Sachs and JPMorgan for an IPO that could value the company at ~$80 billion, including debt, up from $11 billion in 2022. The valuation reflects expectations for both long-term growth and leverage.

For public-markets investors, this listing sets the reference multiple for every private data center platform now weighing an exit, and the print will either validate or puncture the marks those private processes are seeking.

DataBank: majority stake process reaching up to ~$25 billion

Bankers are marketing a majority stake in DataBank, reportedly worth up to ~$25 billion, alongside sales of EdgeCore, Netrality, and Edged. Buyers are paying platform valuations for immediate scale and secured growth.

For infrastructure funds, the signal is that the sophisticated developer-owners are electing to sell control into peak demand, and the buyer inherits both the platform and the build risk the seller is choosing to hand off.

When the Builders Start Selling

The people who understand data center build economics best are selling.

That is the signal this week.

Bankers are marketing majority equity stakes in DataBank, EdgeCore, Netrality, and Edged to private-equity buyers, with the DataBank process alone reaching up to ~$25 billion, reportedly.

These are not distressed sales. They are exits at the top of the market.

Read the timing against the cost curve.

Build costs are climbing, not falling.

Shortages of turbines, transformers, electricians, and memory have pushed the cost of a new gigawatt higher every quarter.

The operators selling equity are the ones who see those costs first.

That is what makes the sell-side compelling.

Builders typically sell for two reasons: to raise capital for the next phase or because they believe the current valuation exceeds future value.

Often, both are true. The buyer’s challenge is determining which motive is driving the transaction.

A controlling stake buys more than operating capacity it includes secured sites, customer relationships, and a development pipeline.

Its value depends on whether future demand continues to outpace new supply.

The entry multiple defines the investment.

If it assumes sustained demand growth, the buyer pays a premium for future build risk. If it prices in a slowdown, the seller may be exiting too early.

Switch illustrates the point: after being taken private at $11 billion in 2022, it is now pursuing an IPO valuing the company at roughly $80 billion, including debt a re-rating that reflects expectations far beyond current cash flows.

It is a repricing of what contracted compute capacity is worth, and the public market is about to vote on whether that number holds.

So here is the discipline.

When the sellers know more about the asset than you do, the entry price is the entire argument.

Do not underwrite the stake on demand you assume.

Underwrite it on the cost to replace what you are buying, and on the multiple you would still pay if demand grew slower than the seller’s ask implies.

The seller has priced their conviction. Price yours.

If you are new to Global Data Center Hub, visit our Start Here page for more information about the newsletter.

If you found this newsletter useful, share it with a colleague.

Have a great week.

— Obinna

Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

Ask ChatGPT a question and the reply appears on your screen in about a second. It feels like the answer comes from your phone, or from somewhere in the air. It does not.

It comes from a specific building, most likely hundreds of miles away, a long windowless box behind a fence with no sign out front.

Your bank runs on one of these buildings. So does Netflix, so does your email, so does the app that holds your photos.

For about twenty years, almost nobody outside the technology business gave these buildings a second thought.

They sat quietly at the edges of towns and did their job, and the rest of us used the internet without ever asking where it physically lived.

That changed almost overnight.

In the last two years, these buildings have shown up in election campaigns, in fights over electricity bills, in warnings from the people who run national power grids, and in the plans of the biggest companies on earth.

To understand why, you first need to know what a data center actually is. It is simpler than it sounds.

Two problems: power in, heat out

A data center is a building with one purpose: to keep a very large number of computers running, all the time, without ever switching off.

The computers themselves are called servers.

Picture a normal desktop computer, then strip away the screen and the keyboard, flatten it into a thin tray, and slide it into a tall metal shelf alongside dozens of others.

Those shelves are called racks, and a big data center holds thousands of them in long rows, like library stacks made of humming metal.

The servers are the part doing the actual work: storing your files, running the apps, answering the questions.

Everything else in the building exists to keep those servers alive, and it comes down to two problems.

The first problem is power.

A room full of running computers needs a constant, heavy supply of electricity, and it cannot afford to lose it even for a moment, because when these machines go dark, so do the services running on them.

So a data center pulls electricity straight from the power grid, and behind that it keeps banks of batteries and backup generators ready to take over the instant the grid hiccups. Keeping the power flowing is the whole game.

The second problem is heat.

Anyone who has felt the underside of a laptop knows that computers get hot, and heat is what destroys them. Multiply one warm laptop by thousands of servers packed into a room, and you have a furnace.

So a large part of every data center is a cooling system, essentially industrial air conditioning, and increasingly a system that runs liquid coolant right up against the chips to carry the heat away.

Add fiber-optic cables to connect the building to the wider internet, and fences and guards to keep people out, and that is the whole thing.

Power coming in, heat going out, and a small city of computers in between. For most of the internet’s history, that description was enough, and these buildings drew a modest amount of electricity and attracted no attention at all.

Then the thing running inside them changed, and the modest amount of electricity stopped being modest.

The shelf that draws a hundred homes’ worth of power

Here is the part that most people have not been told plainly.

You have probably absorbed the idea that artificial intelligence is a triumph of clever software.

That is true, but it hides something.

The kind of AI behind ChatGPT, Google’s Gemini, and the rest runs on NVIDIA chips that have helped make the company one of the world’s most valuable businesses. These chips consume enormous amounts of electricity.

A standard server rack uses about as much power as a few electric ovens. A rack filled with NVIDIA's latest AI chips can consume more than ten times that up to the equivalent of 100 homes.

The heat this produces is so intense that ordinary air conditioning cannot keep up, which is why the newest AI buildings pipe liquid coolant directly onto the chips.

Now multiply that across an entire building. A large AI data center can draw 100 megawatts or more enough electricity to power roughly 80,000 American homes for a year.

One building, drawing as much power as a small city, poured continuously into a windowless box so that a machine can answer questions and generate images for millions of people at once.

That is the reason everyone suddenly cares.

The AI story looks like a software story, and underneath it is a story about electricity, at a scale the world has never had to supply before.

The largest technology companies, Amazon, Microsoft, Google, and Meta, understood this early and began spending on a scale with no precedent.

The consulting firm McKinsey estimates the world will spend roughly $6.7 trillion building data centers by 2030.

And once you follow the electricity, you arrive at the thing that now controls the entire race.

Building it takes months. Powering it takes years.

You might assume the hard part of building one of these is the machines, or the money to buy them.

It is neither. NVIDIA can make more chips over time, and money, for these companies, is not the constraint.

The constraint is getting enough electricity to a specific patch of ground.

The surge in AI demand has broken old assumptions. The International Energy Agency estimates data centers used about 1.5% of global electricity in 2024 and expects that to nearly double by 2030 growing more than four times faster than overall electricity demand.

In the United States the pressure is sharper. Lawrence Berkeley National Laboratory, a US government research lab, found that data centers used about 4.4 percent of the country’s electricity in 2023 and could reach as much as 12 percent by 2028.

In the thirteen eastern states served by the grid operator known as PJM, data centers are expected to account for almost all of the growth in electricity demand between now and 2030.

The trouble is that a power grid cannot grow as fast as a construction crew.

Connecting a big new source of electricity to the American grid now takes a median of about five years, according to Lawrence Berkeley National Laboratory, and the line of projects waiting to connect is longer than the entire existing power supply of the country.

Building a data center, it turns out, is quick. Powering it is slow. Two real stories show what that gap does to the people building these things.

A supercomputer built in 122 days

In 2024, Elon Musk’s AI company, xAI, needed an enormous computer to train its chatbot, Grok, and it needed it immediately.

The company took over an abandoned appliance factory in Memphis, Tennessee, and set out to fill it with a hundred thousand of NVIDIA’s AI chips wired together into a single machine, which it named Colossus.

The industry estimated this build would take eighteen to twenty-four months.

xAI completed it in just 122 days, transforming an empty factory into a working supercomputer. NVIDIA's CEO called it the fastest build of its kind.

Then came the catch.

The local grid could supply only a fraction of the power Colossus needed. Rather than wait years for a grid connection, xAI installed dozens of portable natural gas turbines to generate power on-site.

That choice let the machine run, and it came at a cost to the surrounding neighborhood: the turbines drew objections from residents and environmental groups over the air pollution they added to a part of Memphis that already carried a heavy burden of it, and the dispute ended up in court.

The lesson sits underneath the whole AI boom.

You can build the building in a matter of months, but the power is so hard to come by that a company will burn its own fuel on the doorstep rather than wait in line for the grid.

Switching a dead nuclear reactor back on

Three-Mile Island is a nuclear plant in Pennsylvania, and to many Americans the name means only one thing: the site of the country’s most serious nuclear accident, in 1979. One of its two reactors was involved in that accident.

The other, Unit 1, kept running safely for decades afterward, until 2019, when its owner shut it down because it was losing money.

It sat cold and closed. That looked like the end of it.

In September 2024, Microsoft signed a twenty-year deal to buy the electricity from that plant, on the condition that its owner, Constellation Energy, bring the reactor back to life.

Constellation is spending about $1.6 billion to restart it, has renamed it the Crane Clean Energy Center, and expects it running again by 2027.

Microsoft is buying the plant’s entire output. Every watt that reactor produces is spoken for, to help power the data centers behind Microsoft’s AI services.

Sit with what that means. A shut-down nuclear reactor, a symbol of an accident older than most people reading this, is being switched back on, and the reason is the electricity appetite of AI.

Microsoft is not alone in this. Amazon has bought a data center site next to another nuclear plant. Google and others have signed their own deals for nuclear power.

When companies this large decide that the cheapest way to feed their buildings is to restart a nuclear reactor, you are watching the constraint of this era operate in the open.

Why this lands on your electricity bill

It is easy to see this as a problem for engineers and utility planners. That would be a mistake, because it is precisely why the story moved from the technology pages to the front page.

These buildings compete with homes, hospitals, factories, and schools for electricity.

As they cluster in one region, they can consume a growing share of local power, affecting both electricity prices and grid reliability.

The quiet buildings became loud the moment they started drawing enough power to change what electricity costs for the people living near them.

That is why a governor now holds a press conference about a data center, and why the people who keep the lights on for whole regions now issue public warnings about how fast these buildings are arriving.

There is a larger point underneath the electricity.

The digital life you already live, every message, payment, photo, and stream, has always run through a physical building somewhere, powered by real electricity drawn from a real grid. AI did not create that dependence.

It made it impossible to ignore, by demanding so much, so fast, that it ran headfirst into the limits of the power system the rest of us also rely on.

These buildings are not a footnote to the AI age. They are its foundation, and the foundation is where the strain now shows.

What is the one everyday thing you rely on, without ever thinking about the building it runs through, that surprised you most to learn about here?

Join Global Data Center Hub

Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

Tenant credit quality has become the primary determinant of whether a data center investment survives its own lease term.

The old paradigm underwrote concentration risk on asset quality, location, and power access, then treated the anchor tenant as a fixed input.

That paradigm no longer holds, and the cost of applying it is a facility that finances against a counterparty which cannot service the lease for its full duration.

The Intermediation Layer Nobody Priced

The AI buildout created a new intermediation layer that most underwriting models were never designed to price.

Neocloud providers lease long-dated capacity from operators, populate it with GPU clusters, and resell compute on short-duration contracts.

The capital is moving toward this layer at scale, and a widening share of new colocation demand now sits with tenants whose credit profile bears no resemblance to the hyperscalers who anchor the sector’s financing architecture.

The question is not whether these tenants are growing.

The question is whether they will be capable of paying rent for the full term of a fifteen-year lease, given every competing claim on their cash over that period.

How the Constraint Works

Here is how the constraint works.

A hyperscaler tenant carries an investment-grade rating, diversified revenue, and direct access to public debt markets.

A sub-hyperscaler tenant carries none of these.

It runs a duration mismatch at the center of its model: client compute contracts run four to five years, but the data center lease that houses them runs fifteen or more.

The tenant’s ability to service that lease depends on two things holding simultaneously. GPU hardware must retain residual value as collateral.

Compute rental rates must stay high enough to cover debt. Both are under pressure at once.

Peak GPU rental rates have compressed sharply as newer architectures cut per-token costs, and each hardware generation erodes the collateral value of the last.

A regional operator that accepted a fast-growing neocloud at premium pricing to lock in capacity ahead of a power moratorium now holds a fifteen-year obligation underwritten against a business whose margins compress faster than the lease resets.

A Lease Is Only As Long As Solvency

The industry has long assumed that a signed long-term lease is a credit-equivalent instrument.

That assumption held when the universe of tenants capable of signing a fifteen-year data center lease was small, investment-grade, and cash-generative.

It does not hold now, because the tenant pool has expanded downward into a tier of counterparties whose contractual commitment far outlasts their demonstrated ability to honor it.

A lease is only as long as the tenant’s solvency.

Concentration risk also propagates vertically.

A neocloud tenant may rely on a single customer for most of its revenue, meaning a landlord who appears diversified may still have indirect exposure to the same hyperscaler whose lease pause caused the original stress.

Default Is Not Clean

The supply side hardens the problem, and it does so through time.

Data centers are among the most capital-intensive real estate assets in existence.

A facility built to a specific tenant’s power density and cooling specification cannot be re-leased quickly when that tenant fails.

Backfilling a specialized asset commonly runs twelve to eighteen months in secondary markets, and debt service must be covered from reserves or sponsor support throughout.

The asset that looked like commodity real estate at underwriting reveals itself as a bespoke industrial facility at the moment of default. Power scarcity compounds the sequencing pressure.

Operators who might have waited for a tenant’s financials to season instead accepted the tenant early, because the capacity had to be committed before the interconnection window closed.

Where It Binds

For independent operators, the binding variable is tenant qualification.

A single-asset facility leased entirely to one below-investment-grade counterparty is not comparable to a multi-tenant building where diversification absorbs an individual default.

The operator underwrites the whole asset against one balance sheet.

When that balance sheet is speculative grade, the facility inherits the tenant’s credit as its own.

For private equity and infrastructure investors, the binding variable is the financing cost differential and its effect on returns.

A hyperscaler-anchored project finances at materially tighter spreads and higher loan-to-cost than a neocloud-anchored one.

The same asset, in the same location, with the same power contract, requires substantially more equity per megawatt when the anchor tenant is speculative grade.

That equity requirement compresses the internal rate of return directly and constrains how much the sponsor can deploy.

Poor tenant credit does not merely raise the cost of capital. It caps the velocity of the entire strategy.

For public equity investors, tenant concentration drives valuation.

Single-tenant vehicles often trade at persistent discounts because the market prices counterparty risk before fundamentals.

REITs heavily dependent on one tenant typically underperform those backed by diversified or investment-grade tenants, making valuation dispersion a lasting feature of the sector.

Emerging Markets Tighten Every Rule

In emerging markets, every one of these disciplines tightens.

The replacement tenant pool is thinner, so a default that a primary US market absorbs in months can strand an asset for far longer.

You anchor with an investment-grade cornerstone before financing, you build longer vacancy reserves, and you treat sovereign or telco offtake as a credit floor where private investment-grade tenancy does not exist.

The Direction Is Set

The rating agencies have already moved.

The direction of institutional underwriting standards is set, and it points toward progressively excluding the weakest tenants from institutional capital access.

Underwrite tenant credit for the full duration of the lease, at the portfolio level, and you convert a mispriced risk into a source of advantage.

Wait for further confirmation, and you will hold the assets the disciplined capital declined.

Join Global Data Center Hub

Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

The Risk Transfer Is The Signal, Not The $19 Billion

The structural signal in the Anthropic-TeraWulf lease is not the $19 billion headline.

It is the pass-through tariff filed with the Kentucky Public Service Commission, under which TeraWulf and Anthropic bear all MISO energy and capacity market risk while Big Rivers Electric and Kenergy carry none.

Mainstream coverage has read this deal as a revenue event and a crypto-to-AI pivot story.

The consequence of that read is a mispriced risk profile: the market is underwriting $950 million in average annual rent as if it were utility-grade cash flow, when the underlying power cost sits on the tenant and landlord as a floating wholesale exposure with no margin buffer.

Brownfield Interconnection Is The Scarce Asset Being Monetized

The first read is that the campus monetizes an interconnection, not a building.

The Justified Data site inherited grid access from a retired 600-job aluminum smelter a load already sized for utility-scale industrial draw.

Greenfield projects in this power category wait years for interconnection queue clearance.

TeraWulf calibrated its entire strategy around acquiring legacy high-power sites and building to an anchor tenant, and this lease validates that the interconnection is the value, not the concrete.

Over the next 12 to 24 months, expect competing developers to underwrite retired smelters, steel mills, and coal plants at a premium to greenfield, because the queue position is the moat.

The scarcity is regulatory, not physical.

The Pass-Through Structure Converts Landlord Yield Into Tenant Exposure

The second read is that the pass-through mechanism reallocates who absorbs the volatility that utility-grade cash flow is supposed to eliminate.

Power is delivered with no mark-up on wholesale cost, and the customer is responsible for all transmission and delivery costs and entitled to all MISO node revenues and credits.

This is what produces the premium unit economics approximately $2.37 million per critical IT megawatt-year against a prior industry assumption near $1.9 million but the premium is compensation for risk the tenant and landlord retain, not a free re-rate.

An investor benchmarking this yield against a triple-net utility lease is comparing two different instruments.

The MISO node exposure is the difference, and it does not disappear because the tenant is investment grade.

Rent Commencement Timing Exposes An 18-Month Funding Window

The third read is a liquidity structure hiding inside a revenue story.

Rent commences only as each phase is delivered and commissioned, which protects Anthropic from paying for unenergized capacity but leaves TeraWulf funding a $3.5 billion to $4.0 billion build across an approximately 18-month window with no offsetting lease cash flow.

The Abernathy divestiture $530 million in three installments timed to site preparation, long-lead procurement, and commissioning is not incidental.

It is the bridge across that window.

The proposed $3.5 billion debt package expected to be led by Morgan Stanley completes it.

The signal is that this deal’s execution risk is front-loaded into the pre-revenue period, and the capital stack has been engineered specifically to survive it.

Investor Action

Private Capital. Infrastructure and real estate funds evaluating brownfield conversion should re-benchmark interconnection queue position as a standalone underwritable asset, separate from the physical build.

The diligence action is to price the MISO pass-through exposure explicitly rather than accepting the headline per-megawatt yield as utility-equivalent.

The cost of waiting is that the premium sites retired smelters and coal plants with inherited transmission get bid away by developers who underwrite the queue position first.

Position now on legacy high-power industrial sites or accept greenfield timelines.

Public Markets. Equity investors in TeraWulf and comparable converted miners should separate the contracted-revenue narrative from the pre-revenue funding exposure.

The diligence action is to model the 18-month construction window against the disclosed debt raise and the Abernathy installment schedule, not against the $19 billion aggregate.

The cost of inaction is holding through a dilution or refinancing event mispriced as a pure re-rate. Verify the investment-grade credit reference against actual lease terms, because Anthropic is a private company and the language warrants confirmation.

Operators. Data center developers and hyperscalers acting as tenants should read the single-tenant direct-lease structure as the strategic move it is.

Anthropic secured dedicated, controlled 401 MW capacity outside the cloud metering model, at roughly 4% of its total planned capacity, as part of a deliberate multi-provider strategy.

The diligence action for operators competing for anchor tenants is to structure pass-through tariffs that transfer market risk cleanly, because that transfer is what makes the premium yield financeable.

The cost of failing to do so is losing anchor mandates to developers who can.

The Verdict

This lease is a template, not an outlier.

It decouples high-density compute from the public cloud and binds it directly to a utility-scale energy asset with the market risk pushed onto the counterparties best able to contract around it.

The inflection is the emergence of the brownfield-plus-pass-through structure as the repeatable unit of AI infrastructure development inherited interconnection, single-tenant anchor, wholesale risk transfer, phased rent commencement.

The open question for the next 24 months is whether the tenants signing these leases hold investment-grade credit that survives a training-economics downturn, because the entire structure rests on rent obligations that do not begin until capital has already been spent.

Join Global Data Center Hub

Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

Anthropic PBC has signed a 20-year lease for approximately 401 megawatts of critical IT load at the Justified Data campus in Hawesville, Kentucky, expected to generate approximately $19 billion in contracted revenue over the initial term.

TeraWulf Inc. subsidiary Raylan Data LLC acts as the landlord.

The lease funds a purpose-built, single-tenant high-performance computing facility for Anthropic’s Claude model training and inference.

TeraWulf announced the agreement on July 6, 2026, with phased capacity delivery beginning in the second half of 2027.

Lease Terms And Rent Commencement

The agreement guarantees approximately $19 billion in cumulative revenue over a 20-year initial term, an average annual run rate of approximately $950 million.

The lease includes two successive five-year renewal options, allowing Anthropic to extend to a total of 30 years.

Rent obligations commence incrementally as each phase of physical capacity is delivered, rather than at a fixed future date.

The lease is backed by an investment-grade credit profile referenced in the transaction filings.

The Hawesville Brownfield Site

The Justified Data campus sits at the site of the retired Century Hawesville aluminum smelter in Hancock County, which ceased operations in 2022 with the loss of 600 industrial jobs.

The facility comprises a gross site footprint of 482 MW, engineered to deliver 370 MW to 401 MW of critical IT load across two primary data halls, each supporting 185 MW of critical capacity.

The campus uses a closed-loop liquid cooling system designed by Motivair and Schneider Electric, using a water and propylene glycol mixture that recycles within a sealed system and draws no water from local reservoirs or river systems.

Grid Interconnection Structure

Electric service is governed by a long-term Agreement for Electric Service submitted to the Kentucky Public Service Commission on April 14, 2026, under Case Number 2026-00115.

The contract defines a 15-year initial term with five-year rolling extensions and operates as a pass-through structure tied to the Midcontinent Independent System Operator energy and capacity markets.

Power flows from MISO through Big Rivers Electric Corporation and Kenergy Corp to Raylan Data on a pass-through basis with no mark-up on wholesale cost.

The customer bears all MISO market energy and capacity risk and is entitled to all MISO revenues and credits attributable to the Hawesville Commercial Pricing Node.

Concurrent Abernathy Divestiture

Concurrently with the lease, TeraWulf subsidiary Big Country Wulf LLC divested its 50.1% interest in the Abernathy Joint Venture in Texas to an investor group led by Fluidstack CS I Inc. for approximately $530 million, an 18% premium over TeraWulf’s original $450 million investment.

The consideration is structured in three installments: $250 million within 14 days of execution, $150 million on or before December 31, 2026, and approximately $130 million on or before April 30, 2027.

Total site development capital for the Justified Data campus is estimated at $3.5 billion to $4.0 billion, excluding tenant hardware.

First capacity is scheduled for the second half of 2027, with full buildout following.

Join Global Data Center Hub

Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

In this week’s issue:

First, the lease became the asset. Anthropic committed to a ~$19 billion, 20-year lease at TeraWulf’s 401MW Kentucky campus, backed by investment-grade credit, and the market repriced a former bitcoin miner on the strength of one counterparty.

If a 20-year offtake can revalue the operator more than the building ever could, how should you price every lease that crosses your desk?

Then, private capital moved in behind the trade. Ares took a strategic stake in Sabey, CPP Investments committed ~$1.75 billion to EQT and EdgeConneX, and BlackRock said private credit will take a larger role funding the buildout.

Finally, the platforms went public. Brookfield’s Csquare filed for a ~$1.35 billion data center IPO, testing whether listed markets will pay for what private owners built.

Let us get into it.

When the Biggest Customer Becomes a Supplier

The contract now reprices the operator faster than the building ever could.

TeraWulf proved that this week.

Anthropic signed a 20-year lease at the Justified Data campus in Hawesville, Kentucky.

The lease is expected to generate approximately $19 billion in contracted revenue.

TeraWulf shares jumped as much as 19 percent intraday before settling near 4 percent.

That $19 billion figure is larger than TeraWulf’s entire market value of roughly $12 billion.

One tenant now underwrites more future revenue than the whole company was worth the day before.

TeraWulf was a bitcoin miner. It has no hyperscale operating history. What made it financeable was not the concrete in Kentucky. It was Anthropic committing for 20 years against an investment-grade credit.

The asset was never the campus. The asset is the promise to pay for the campus. That promise is only as strong as the counterparty standing behind it.

Read the mechanics.

The campus will deliver 401MW of critical IT load, with first power in the second half of 2027 and full build by early 2028. The buildout implies roughly $3.2 to $4 billion in capex.

The $19 billion sits on the revenue side, while the capex sits on the risk side. That gap only closes if Anthropic continues paying.

That is why the same week carried a second signal.

TeraWulf sold its 50.1 percent stake in the Abernathy joint venture for approximately $530 million, redeploying into assets it owns and controls outright. The operator is concentrating capital where the offtake is secured and the control is direct.

So here is the discipline.

Do not price a data center operator on its square footage or its megawatts. Price it on the credit quality of the tenant and the length of the lease.

Decompose the revenue by counterparty. Size the single largest one. Then ask what the termination terms recover if that tenant walks before term. For a name like TeraWulf, one lease is the entire equity story, and one counterparty is the entire risk.

If Anthropic’s credit moved tomorrow, what would remain of the $19 billion, and who would be left holding 401 megawatts in Kentucky?

THIS WEEK BY REGION

The week’s biggest moves — what happened and what it signals.

North America

Capital concentrated around secured offtake and control this week. Anthropic signed a roughly $19 billion, 20-year lease at TeraWulf's 401MW Kentucky campus, while TeraWulf sold its Abernathy stake to focus on wholly owned assets. Ares also invested in Sabey Data Center Properties, gaining exposure to a stabilized operating platform.

MasTec acquired The Superior Group for $1.65 billion to secure construction capacity. Galaxy energized 133MW for CoreWeave, while Meta committed C$13 billion to its first Canadian data center campus.

For public markets equity investors, the read is that the market is now paying for contracted duration and shovel-ready power, and the operator holding neither trades at the widest discount to its backlog.

Europe

The region drew growth capital into AI infrastructure buildouts. CPP Investments committed about $1.75 billion to EQT and EdgeConneX’s AI data center expansion, backing development risk over stabilized assets.

Nscale, backed by Nvidia, secured about $900 million for data center expansion, showing vendor-backed capital moving toward earlier stage infrastructure risk. Both moves push institutional money closer to construction and demand creation.

For infrastructure funds, the signal is that the safest pools of capital are now underwriting the riskiest stage of the buildout, and the premium for stabilized assets will widen as they do.

Asia-Pacific

The region's story is national scale ambition rather than completed deals. SK Telecom plans up to 15GW of AI data center capacity under South Korea's AI G3 strategy, but the buildout still depends on securing anchor tenants and financing.

Meanwhile, Tata expanded India Singapore subsea connectivity to support growing regional compute demand.

For infrastructure funds, APAC is still building the foundations North America has already priced, making platforms and connectivity more attractive than stabilized assets.

Middle East and Africa

African sovereign capacity still arrives one certified facility at a time. Gabon opened its first Tier III data center, a ~$14 million site built around a digital-sovereignty mandate.

The constraint in the market is certification and power access, not investor appetite, and the state remains the gatekeeper of both.

For development finance institutions, early-market position in Africa is taken at the permitting and power layer, years before the capital-formation stage that dominates developed markets.

NOTABLE TRANSACTIONS

Key structures and capital moves from this week’s deal tape.

TeraWulf: ~$19 billion, 20-year Anthropic lease at the Justified Data campus

TeraWulf signed a 20 year lease with Anthropic for a 401MW Kentucky campus, expected to generate about $19 billion in contracted revenue. The deal transforms a former bitcoin miner into an AI infrastructure operator.

For the financing landscape, the mechanism is offtake-as-collateral, and it means a lender now underwrites the tenant’s credit rather than the operator’s history.

Ares: strategic investment in Sabey Data Center Properties

Ares invested in Sabey through its real estate secondaries arm, gaining direct exposure to a cash generating data center operator. The deal signals secondaries capital is moving into digital infrastructure.

For operators, the mechanism is secondaries-as-primary, and it widens the pool of institutional buyers competing for control of established platforms.

MasTec: $1.65 billion acquisition of The Superior Group

MasTec agreed to acquire The Superior Group for $1.65 billion, expanding its data center construction capabilities. The deal values execution capacity as the scarce asset, not just the land or power behind it.

For the financing landscape, the mechanism is execution-as-scarcity, and it means contractor capacity is becoming a gating input on how fast committed capital can convert into energized megawatts.

If you are new to Global Data Center Hub, visit our Start Here page for more information about the newsletter.

If you found this newsletter useful, share it with a colleague.

Have a great week.

— Obinna

Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

The License Count Is the Signal, Not the Dollar Figure

The number that matters in Egypt is not $400 million.

It is 10.

The National Telecommunications Regulatory Authority has issued 10 data center licenses in two years, and the Hassan Allam and A15 authorization on June 15, 2026 is the tenth.

Mainstream coverage anchored on the capital figure and the conglomerate name.

The structural signal is the cadence.

A regulator that formalized its Public Data Center Provider framework only in 2023 is now clearing licenses at roughly one every ten weeks.

Allocators reading the headline figure are pricing a single facility.

Allocators reading the cadence are pricing a market that is being licensed into existence ahead of the demand that will fill it.

The Platform Is Built to Own a Position, Not to Operate a Building

The Hassan Allam and A15 structure is the first read. This is not a pure-play operator and it is not a hyperscaler captive.

It pairs Hassan Allam Holding, an engineering, procurement, and construction conglomerate with utility-scale delivery history, against A15, an Egyptian technology and venture capital firm.

The EPC parent brings the capacity to build power-dense facilities at cost. The venture partner brings cloud commercialization and the tenant relationships that convert whitespace into contracted revenue.

The structure underwrites a vertically integrated position: land, build, power, and lease held inside one platform.

This is a model that developed markets recently introduced and is now finding its way into frontier digital infrastructure markets.

The pure-play colocation operator carries construction risk it cannot control. The hyperscaler captive serves one tenant.

The EPC-plus-venture platform captures the build margin and the lease margin and diversifies the tenant book.

Where the source material does not disclose the debt and equity split or any lender syndicate, that structure is not read here. What is confirmed is the platform composition, and the platform composition is the signal.

The License Is the Asset, Not the Facility

The second read is the license itself.

Under the NTRA framework, a Public Data Center Provider license runs 15 years and carries the right to contract directly for submarine cable capacity and fiber paths without case-by-case regulatory approval.

In a market where an estimated 17-20% of global internet traffic transits the country and 15 or more submarine cable systems land on its coasts, the direct-interconnection right is the scarce asset.

Hassan Allam and A15 have acquired a 15-year position on the transit corridor before the hyperscale demand that monetizes it has fully arrived.

The party that holds the interconnection right when that demand crystallizes captures the rent. The license is a call option on transit-corridor scarcity, and it was struck ahead of the move.

Localization Law Converts Regulatory Risk Into Captive Demand

The third read is the demand pull that most coverage treats as a compliance footnote.

Egypt’s Personal Data Protection Law restricts cross-border transfer of citizen data, and the executive regulations issued in late 2025 started a compliance clock running to October 2026.

The read most observers miss is directional. A localization regime is not a constraint on this platform. It is the demand engine.

Every multinational, financial institution, and cloud provider serving Egyptian users must domicile that data in-country.

That mandate converts a regulatory requirement into captive, non-discretionary colocation demand that cannot be served from Gulf or European facilities.

The platform that holds licensed in-country capacity when the compliance deadline forces the migration underwrites demand it did not have to win on price.

The localization law is the offtake the source material does not name in contract form but names in statute.

Investor Action

Private Capital. Infrastructure and private equity allocators evaluating frontier digital infrastructure should assess the integrated platform against the pure-play operator model.

A vertically integrated structure captures build and lease margins while controlling construction risk.

The action is to diligence the interconnection rights and the localization-driven demand curve as the underwriting spine, not the single-facility economics.

The cost of waiting is entry after the transit-corridor licenses are allocated, when the scarce asset is no longer available to underwrite and the position must be bought secondary at a premium.

Public Markets. Public equity and credit investors with hyperscaler and cable-operator exposure should sequence Egypt into the regional map now.

The action is to calibrate the localization deadline and the licensing cadence as leading indicators of in-country capacity demand.

The cost of inaction is pricing the region off developed-market comparables and missing the demand inflection the compliance clock forces.

Operators. Data center developers and hyperscalers acting as buyers should read the licensing cadence as a closing window.

Ten licenses in two years signals a regulator moving fast and a corridor position that is being claimed.

The action is to negotiate interconnection and colocation positions before the localization deadline converts optional demand into forced demand and compresses the negotiating leverage of late entrants.

The cost of waiting is leasing capacity at deadline-driven pricing from the platforms that positioned early.

The Verdict

Egypt is being licensed into a market position before the demand that validates it has fully arrived, and the Hassan Allam and A15 authorization is the clearest instance of a private platform underwriting that timing gap.

The market inflection is the point at which the October 2026 localization deadline and the sequenced hyperscaler pipeline convert corridor optionality into contracted rent.

The platforms holding 15-year interconnection rights at that inflection capture the spread between an option struck early and demand priced late.

The open question is whether the power grid keeps pace.

Egypt underwrites this thesis on a surplus-electricity advantage, though its grid is still scaling for hyperscale load. The transit corridor is a durable asset.

The power to support what lands on it is the variable the next 24 months will resolve.

Join Global Data Center Hub

Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

AI infrastructure did not just raise more money in Q2 2026. It changed the kind of money it raises.

Here’s What’s Inside

Top 15 global announcements — CoreWeave's $8.5 billion GPU-backed facility, Oracle's $14 billion debt repricing, Meta and Microsoft's lease expansion, AirTrunk's Mumbai commitment, Amazon's $17.5 billion loan, and the platforms reshaping AI infrastructure capital.

5 structural trends — Investment-grade ratings on novel collateral, a growing secondary market, institutional origination at scale, demand contracting in gigawatts, and power and permitting as the binding constraint.

5 strategic opportunities — Rated infrastructure credit as a new institutional allocation, emerging markets re-rating to primary destinations, and power-adjacent origination as its own investable position.

5 structural risks — Delivery risk shifting into debt, secondary-market volatility, GPU depreciation, concentrated origination, and social license as a siting constraint.

Q2 2026 marked the quarter capital markets turned compute into a rated, tradable credit product and quietly inherited the power risk sitting underneath it.

Top 15 Global Announcements (Q2 2026)

Below are the most consequential capital events that defined the global data center and AI infrastructure landscape in Q2 2026, ranked by capital scale, structural signal, and long-term market impact.

1. CoreWeave Closes $8.5 Billion Investment-Grade GPU-Backed Financing

CoreWeave closed what it described as the first investment-grade-rated financing facility backed by GPUs. More important than the size was the structure: rating agencies accepted rapidly depreciating GPU collateral, moving AI infrastructure debt toward a formal credit category and establishing a template for future issuers. [Read here]

2. Investors Reprice $14 Billion of Oracle-Backed Data Center Debt

Investors demanded higher yields on more than $14 billion of Oracle-backed data center debt, while PIMCO sold portions of its holdings. Active repricing and trading marked the emergence of a secondary market. This debt is no longer held to maturity; it is priced continuously, bringing capital-markets volatility into infrastructure finance. [Read here]

3. Meta and Microsoft Lead $850 Billion Data Center Lease Boom

Bloomberg reported an $850 billion data center leasing boom led by Meta and Microsoft. At this scale, leasing shifts balance-sheet risk from hyperscalers to developers and their lenders. Hyperscalers provide demand certainty, capital markets provide financing, and developers absorb construction and power risk. [Read here]

4. AirTrunk Commits $21 Billion to Data Center Project Near Mumbai

AirTrunk committed $21 billion to a data center project near Mumbai, one of the quarter's largest investments. Backed by Blackstone and a Canadian pension, the project signals India's emergence as a primary destination for institutional infrastructure capital rather than a diversification market. [Read here]

5. Amazon Secures $17.5 Billion Loan for AI Data Center Buildout

Amazon secured a $17.5 billion loan for AI data center buildout. More significant than the size was the financing choice: even the sector's strongest balance sheets now prefer dedicated infrastructure debt over corporate capital expenditure, confirming debt as the default funding instrument. [Read here]

6. Related Digital Finances $16 Billion Oracle Data Center in Michigan

Related Digital announced financing for a $16 billion Oracle data center project in Michigan. The deal exemplifies the quarter's model: a hyperscale tenant, a developer intermediary, and structured financing. It also links new origination with the active repricing of Oracle-backed debt, showing primary and secondary markets maturing in parallel. [Read here]

7. Hut 8 Closes $3.25 Billion of Investment-Grade Senior Secured Notes

Hut 8 closed $3.25 billion of investment-grade senior secured notes for its River Bend project. A former crypto miner accessing the investment-grade bond market highlights how quickly the credit category has expanded. It later paired the financing with a 352 MW lease valued at $9.8 billion, combining long-term contracted demand with rated debt. [Read here]

8. SoftBank Group to Build 5 GW of AI Data Center Capacity in France

SoftBank announced plans to build 5 GW of AI data center capacity in France. The commitment reinforces France's emergence as a top-tier AI infrastructure market and reflects the quarter's defining shift to measuring demand in gigawatts rather than floor space. The scale also underscores how power, not buildings, has become the industry's primary unit of growth. [Read here]

9. Blackstone Acquires 49% of Rowan Digital Infrastructure

Blackstone agreed to acquire a 49% stake in Rowan Digital Infrastructure as part of a strategic recapitalization. The investment extends its shift from acquiring stabilized assets to taking positions in development platforms. Control of origination, rather than ownership of individual facilities, is becoming the strategic objective for the largest infrastructure investors. [Read here]

10. KKR Launches Helix Digital Infrastructure

KKR launched Helix Digital Infrastructure, a dedicated platform to finance and develop AI infrastructure. Creating a permanent operating company rather than investing through a generalist fund signals a multi-cycle commitment to industrialized origination. It also concentrates future deal flow within a smaller group of large, specialized platforms. [Read here]

11. Applied Digital Surpasses 1 GW Contracted With Investment-Grade Hyperscaler Leases

Applied Digital surpassed 1 GW of contracted capacity with an investment-grade hyperscaler and paired it with a proposed $1.59 billion senior secured notes offering and a $550 million revolving facility. The company exemplified the quarter's dominant model: secure an investment-grade tenant, then raise debt against contracted cash flows. Tenant credit now underpins infrastructure credit. [Read here]

12. AWS and Anthropic Expand Collaboration for Up to 5 GW of Compute

Amazon and Anthropic expanded their collaboration to include up to 5 GW of new compute alongside an additional $5 billion investment. Gigawatt-scale commitments provide the long-term demand certainty required to support large-scale infrastructure financing. Without offtake at this scale, many of the quarter's debt structures would be difficult to underwrite. [Read here]

13. Meta Partners With Reliance on AI-Enabled Data Center in India

Meta partnered with Reliance on an AI-enabled data center in India. The partnership illustrates how hyperscalers are using established domestic operators to secure power, land, and regulatory access in strategic markets. Alongside AirTrunk's investment, it reinforces India's emergence as a primary destination for global AI infrastructure capital. [Read here]

14. DayOne Closes $4.5 Billion Series C Equity Financing

DayOne closed a $4.5 billion Series C financing, one of the quarter's largest equity raises and an important counterweight to the debt-heavy narrative. The raise reinforces a capital stack that is increasingly bifurcated: equity finances early development risk, while investment-grade debt funds contracted, stabilized assets. [Read here]

15. America’s Data Center Build-Out Reported Behind Schedule

The Wall Street Journal reported that U.S. data center development is falling behind schedule. This is the principal execution risk underlying many of the quarter's financings. As capital is committed faster than power and construction can deliver capacity, delivery risk increasingly migrates from projects into the debt itself. [Read here]

Thematic Analysis: Five Structural Trends (Q2 2026)

1. Investment-Grade Ratings Arrived on Novel Collateral

Q2 2026's most significant shift was qualitative: rating agencies granted investment-grade status to GPU-backed, single-tenant AI assets for the first time. That opens the sector to institutional fixed-income capital, expanding the investor base well beyond private project finance. Once compute is ratable, it becomes bankable at scale, lowering the cost of capital for qualified issuers.

2. A Secondary Market Formed and Began Repricing the Paper

The second shift was the emergence of active trading. Investors repricing more than $14 billion of Oracle-linked debt, and PIMCO moving to sell down, is the signature of a secondary market rather than a buy-and-hold book. This is maturation with a sharp edge: paper that trades can sell off, and a sector-wide repricing event would raise the cost of capital across every issuer at once. Liquidity and volatility arrived together.

3. The Largest Allocators Industrialized Origination

Major sponsors shifted from opportunistic investing to building permanent AI infrastructure platforms. Taking strategic stakes in developers, rather than buying finished assets, is a bet on controlling long-term pipeline. That concentrates future deal flow among a handful of large balance sheets and raises barriers to entry.

4. Demand Is Now Contracted in Gigawatts, Not Square Feet

Anchor tenants are committing to power capacity as the unit of demand, which is exactly what makes long-dated infrastructure debt underwritable. Gigawatt-scale offtake supplies the revenue certainty lenders require to rate the paper. The same shift exposes the sector’s fault line: these commitments presuppose power that, as the delivery data shows, the grid cannot yet reliably provide.

5. Power and Permitting Became the Binding Constraint

Across three continents, the limiting factor is no longer capital or demand but electrons and the social license to consume them. When a hyperscaler-backed project in Kenya stalls on energy availability and Russia freezes billions over power and rate pressure, the constraint is structural rather than local. This is the risk the credit market is systematically underpricing as it races to originate.

Opportunities

1. Rated Infrastructure Credit as a New Institutional Allocation

Investment-grade, asset-backed data center debt, established this quarter by CoreWeave and Hut 8, opens the asset class to institutional fixed-income investors. Allocators limited to rated securities can now participate directly in AI infrastructure. The advantage will accrue to issuers that secure ratings early and the arrangers that standardize the structures the market will follow.

2. Emerging Markets Re-Rating to Primary Destinations

Investment-grade, asset-backed data center debt, established this quarter by CoreWeave and Hut 8, opens the asset class to institutional fixed-income investors. Allocators limited to rated securities can now participate directly in AI infrastructure. The advantage will accrue to issuers that secure ratings early and the arrangers that standardize the structures the market will follow.

3. Power-Adjacent Origination as Its Own Investable Position

Deals pairing generation with compute, including Google-Intersect’s Meitner Energy Center and Chevron-Microsoft’s West Texas arrangement, show power becoming a standalone investable position, not just a site-selection input. Controlling generation near demand turns the sector’s constraint into an asset. As delivery timelines slip, secured dispatchable power near compute will command a wider premium.

4. The Bifurcating Capital Stack

The quarter’s financings split by risk stage: equity funded development, while rated debt financed contracted assets. DayOne, Firmus, and Fluidstack raised large equity rounds alongside investment-grade note issuances, signaling distinct capital pools for different points on the risk curve. Capital is increasingly specializing rather than blending risk.

5. Platform Consolidation Over Single-Asset Acquisition

Capital shifted from individual assets to platforms. Blackstone acquired 49 percent of Rowan Digital Infrastructure, I Squared bought Brazil’s Elea, and DigitalBridge agreed to acquire ArcLight, each securing pipeline, power, and operating capability in one transaction. Platform acquisitions are becoming the preferred way to deploy capital at scale.

Challenges

1. Delivery Risk Migrating From Operations Into the Debt

With the United States build-out reported behind schedule, the gap between contracted and deliverable capacity is no longer merely an operational problem. It is a credit problem. Debt underwritten against leases that assume on-time energization is exposed when interconnection and construction slip, and that exposure is now embedded in rated paper held by institutional investors.

2. Secondary-Market Volatility Raising the Cost of Capital

Now that Oracle-linked paper is actively repriced, sentiment can move the sector’s cost of capital quickly and without warning. A repricing event on one large issuer can cascade across comparable credits, raising yields for issuers who have done nothing wrong. The liquidity that makes the asset class attractive is the same mechanism that transmits shocks through it.

3. GPU-Collateral Depreciation Outrunning Debt Tenor

The collateral underpinning the quarter’s signature financings is compute hardware that depreciates on a far shorter cycle than the debt written against it. If GPUs lose economic value faster than the notes amortize, the collateral coverage assumed at issuance erodes, and refinancing risk rises precisely when the paper most needs support. Rated does not mean immune to obsolescence.

4. Origination Concentrating in a Handful of Platforms

As KKR, Blackstone, and Brookfield industrialize origination through dedicated platforms, deal flow is concentrating within a shrinking group of large balance sheets. Concentration improves execution but reduces competition and increases systemic linkage: distress at a major originator would ripple across the sector's pipeline. Market breadth is narrowing as the sector scales.

5. Social License Hardening Into a Systemic Siting Constraint

Opposition is becoming a structural constraint rather than a local hurdle. Maine enacted a statewide ban, the UK government intervened in a $500 million London project, and developers withdrew projects in New Jersey, Ontario, and Australia. Siting constraints are increasingly limiting capacity regardless of available capital or power.

Conclusion

Q2 2026 marked the moment digital infrastructure completed its transition from a real estate story into a credit story one with ratings, secondary trading, and dedicated institutional platforms.

Rating agencies, lenders, and major sponsors converged on one goal: rate the collateral and scale origination.

This quarter’s investment-grade GPU facilities, repriced debt, and gigawatt-scale leases show that advantage now lies in rated capital and secured power, not location.

That shift expands the capital pool but also increases repricing risk as power and permitting delays continue.

The next few quarters will test whether capital can keep financing gigawatts the grid cannot yet deliver. The financing is ready. The power is not.

Join Global Data Center Hub

Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

The Buyer And The Seller Are Reading The Same Market Differently

The signal in this transaction is not the $3.5 billion.

It is that the largest financial sponsor of digital infrastructure sold finished, fully leased capacity in the world’s most constrained market, and the operator paid roughly $27 million per megawatt to take it.

Mainstream coverage read the 5.77% single-day drop in Digital Realty stock as the story.

The drop was a technical function of a secondary block sale clearing in one session, not a verdict on the assets.

What the tape obscured is a structural repricing: in a market where new supply cannot be built on schedule, the premium has migrated from the development spread to the finished asset itself.

The Exit Timing Prices Greenfield Risk, Not Asset Quality

Blackstone committed capital to the underlying joint ventures in December 2023 and exited the Northern Virginia tranche in under three years.

A three-year hold on a development-stage infrastructure position is fast.

The conventional read is a clean development-to-core return, and that read is correct as far as it goes.

The structural read is that Blackstone chose to realize a ~6.5% stabilized yield rather than hold it, at the exact moment its own greenfield pipeline in the same market was failing.

Two days after this deal closed, Blackstone-backed QTS abandoned the Prince William Digital Gateway, a project once envisioned as the world’s largest data center campus, after zoning approvals were voided and its co-developer exited.

The same firm that could not push a greenfield megaproject through Prince William County took a clean cash-and-stock exit on finished capacity 30 miles away.

That is not a conviction reversal. It is a firm allocating away from the part of the lifecycle that has become unshippable and toward the part that still clears.

Powered-Shell Scarcity Now Sets The Price, Not The Development Spread

The $27 million per megawatt valuation is the number to underwrite, not the headline.

Historically, the return in this asset class was manufactured in the development spread: buy land, secure power, build, lease, and capture the gap between development cost and stabilized value.

That model assumes greenfield delivers on schedule.

In Northern Virginia it no longer does. Dominion Energy reported a 70,000 MW backlog of large-load interconnection requests at the end of 2025, with wait times of three to five years.

Virginia’s biennial budget, signed June 30, 2026, added a $0.011 per kWh consumption tax, water-use mandates east of I-95, and noise limits enforceable from 2030. When greenfield cannot be scheduled, the finished, powered, leased asset becomes the scarce instrument, and its price decouples from replacement cost.

Over the next 12 to 24 months, expect further operator buy-ins of stabilized JV positions in primary markets, priced off scarcity rather than off the development spread that scarcity has closed.

Investor Action

Private Capital. For infrastructure and real estate funds holding development-stage positions in constrained primary markets, this deal benchmarks your exit.

Blackstone realized a stabilized position into a strategic operator rather than holding for yield or transferring to its own retail vehicle.

Diligence which of your assets are genuinely powered and permitted versus which are queue-dependent, because the two are now priced as different asset classes.

The cost of waiting is holding a greenfield position into a market that has stopped rewarding the development spread and started penalizing execution risk.

Public Markets. For public equity and credit investors, separate the technical from the structural.

The 5.77% Digital Realty decline was a secondary block clearing, not a fundamental repricing; two institutional desks held buy ratings with price targets above $230 through the drop.

The structural read is that Digital Realty is converting joint-venture exposure into wholly owned, escalator-bearing cash flow at a leverage-neutral cost.

Underwrite the 2027 and 2028 Core FFO accretion the company guided, and treat the dilution overhang as a timing entry, not a thesis break.

The cost of reading the tape as the story is mispricing a structural consolidation as a technical stumble.

Operators. For developers and operators, the strategic lesson is that full ownership of finished, powered capacity in a primary market is now worth paying a scarcity premium to secure.

Digital Realty removed joint-venture governance friction and locked 288 MW of Aa3/AA- leased capacity with 3.6% escalators at a moment when it cannot reliably build the equivalent.

Sequence your own JV buy-ins around which assets are stabilized and secured, not around which carry the highest headline yield.

The cost of inaction is watching finished capacity in your core market clear to a competitor while your own greenfield pipeline sits in an interconnection queue.

The Verdict

This transaction marks the point at which the digital infrastructure premium visibly detached from the development spread and attached to the finished asset.

The constraint is no longer capital.

Capital is abundant. The constraint is power, permitting, and community consent, and those constraints have made the powered-shell asset the scarce instrument in the primary market.

Blackstone recycled out of the yield and into its next greenfield and credit deployments; Digital Realty consolidated the scarce end of the lifecycle. Both are rational, and both point the same direction.

The open question for the next 24 months is whether scarcity pricing holds as the AI capacity wave either clears the interconnection queues or breaks against them.

If the queues clear, the premium compresses. If they do not, $27 million per megawatt is a floor, not a peak.

Join Global Data Center Hub

Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

Digital Realty Trust agreed to purchase Blackstone’s blended 64% equity interest in three fully leased hyperscale data centers in Northern Virginia for $3.5 billion in net equity consideration.

interests sat inside two joint ventures, the Digital Carver Dulles 9 and Digital Carver Brickyard vehicles, backed by Blackstone’s Real Estate, Infrastructure, and Tactical Opportunities funds.

The proceeds gave Digital Realty full ownership of 288 MW of critical IT capacity across three facilities.

The transaction was announced on June 29, 2026, and closed on June 30, 2026.

Portfolio And Asset Detail

The portfolio comprises three 96 MW facilities.

Digital Realty acquired Blackstone’s 80% interest in two 96 MW data centers in Manassas, Virginia, and its 50% interest in one 96 MW data center on the Digital Dulles campus in Sterling, Virginia.

On close, both joint ventures became wholly owned subsidiaries of Digital Realty’s operating partnership.

The gross valuation of $7.8 billion on a 100% basis includes assumed debt and the remaining capital expenditure required to reach full stabilization.

Lease Terms And Tenant Credit

The three data centers are 100% leased to three distinct investment-grade hyperscale tenants on 15-year contracts.

The leases carry a blended average tenant credit rating of Aa3/AA- and a 3.6% annual rental escalator weighted by contractual rent.

The transaction implies an initial stabilized capitalization rate above 6.5%, calculated on estimated first-year stabilized net operating income.

Two facilities are expected to reach full stabilization in the first half of 2027 and the third in the first half of 2028.

Consideration And Capital Sourcing

The $3.5 billion consideration was split into a $1.2 billion cash component and a $2.3 billion stock component.

Digital Realty funded the cash portion through its at-the-market equity program, selling 6,158,839 shares between April 29 and June 29, 2026, to raise approximately $1.2 billion.

The stock component was paid by issuing 12,310,249 shares of newly designated non-voting common stock, struck at the June 29 closing price of $190.58.

The non-voting shares convert automatically into voting common stock on transfer to unaffiliated buyers.

Secondary Offering And Close

Blackstone’s affiliates opted to monetize the stock immediately.

On June 29, Blackstone entered an underwriting agreement with Morgan Stanley as sole underwriter for a registered secondary offering of the full 12,310,249-share block.

The offering priced after the close on June 30 at $185.00 per share, a 2.9% discount to the pre-announcement price, raising approximately $2.277 billion in gross proceeds for Blackstone.

Digital Realty received no proceeds from the secondary.

The company projects the acquisition to be leverage-neutral and accretive to Core Funds From Operations per share in 2027 and 2028 as the final construction phases complete.

Join Global Data Center Hub

Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

Design determines seventy to eighty percent of a data center’s lifecycle cost while consuming three to seven percent of project spend.

The market treats design as a documentation exercise that precedes the real money in construction.

That inversion is why projects miss their commercial operation dates by twelve to twenty-four months and why underwriters price schedule risk they never had to carry.

The AI buildout has made this expensive.

Rack densities moved from ten kilowatts to fifty and now toward three hundred within a single equipment generation.

Liquid cooling shifted from option to requirement above roughly thirty to forty kilowatts per rack.

Every facility designed for yesterday’s density is a stranded-capital problem the moment it energizes.

The question is not whether a site can be built. The question is what gets committed, and when, before a single foundation is poured.

Two Design Regimes, One Point of No Return

Understanding this requires separating the two design regimes that govern every project.

Conceptual design came first as a discipline because the industry learned that premature precision destroys optionality.

It sets the Basis of Design: the redundancy philosophy, the cooling technology, the power density target, the phasing strategy.

It produces an order-of-magnitude cost model at plus or minus twenty-five to thirty percent. It runs five to fifteen percent complete and takes two to four months.

Detailed design then converts that architecture into construction-ready instruction, progressing through schematic design, design development, and construction documents at thirty, sixty, and ninety percent completion.

It reaches plus or minus five to ten percent cost accuracy and takes six to twelve months.

The Reliability Class Is Set Before Anything Is Built

The single most consequential decision in the entire sequence sits in conceptual design. It is the Tier and redundancy topology.

Moving from Tier II to Tier III typically adds twenty to forty percent to mechanical and electrical capital cost. Tier III to Tier IV can add another thirty to fifty percent.

Concurrent maintainability, the Tier III standard that lets an operator service any component without interrupting IT load, cannot be retrofitted.

It requires fully redundant distribution paths and cross-connected buses designed in from the first schematic.

A facility’s reliability class is set in conceptual design and cannot be meaningfully improved afterward without major capital reinvestment.

Electrical infrastructure carries twenty-five to thirty percent of the total budget, the largest single line.

Mechanical systems carry twenty to twenty-five percent. These proportions hold across the uploaded evidence and they explain where design leverage actually lives.

The Uptime Institute reinforces the point structurally.

It stopped issuing design-only certifications for commercial operators in the US after 2015, having observed facilities that claimed Tier III design certification but built to a lower standard.

Certification is now earned by the constructed facility, not the drawing set.

Procurement Is Design, Not a Later Phase

The narrative has focused on construction as the capital event. That misses where the commitments become irreversible.

The capital reality is that procurement is a design activity, not a downstream one.

Long-lead equipment sets the true critical path.

Transformers, switchgear, generators, and chillers carry manufacturing lead times that span both source estimates, from roughly twelve weeks at the low end to sixty-five weeks at the high end in constrained markets.

Utility interconnection is longer still: twelve to thirty-six months, and five to twelve years where transmission upgrades enter congested queues in ERCOT, PJM, or MISO.

A project that waits for one hundred percent construction documents before ordering a transformer has already lost six to twelve months it will never recover.

Issue-for-procurement packages must go out at sixty to seventy percent design completion.

The design locks first. The order follows.

Where You Sit Changes What Binds You

This lands differently depending on where you sit.

For independent operators and developers, the binding constraint is specification discipline.

A thirty percent change to rack density assumptions late in detailed design propagates across structural loading, electrical bus capacity, and cooling plant sizing at once.

White space slabs designed for standard load will not carry AI racks that exceed four hundred pounds per square foot in concentrated areas. You cannot value-engineer your way out of a density decision you got wrong at concept.

Benchmark the ultimate density before you set the structural grid, or you will underwrite a retrofit you cannot finance.

For private equity and infrastructure investors, the decisive underwriting constraint is the interconnection timeline, not the construction schedule. Utility power is the critical path of most programs.

The earliest achievable commercial operation date is set by the grid, and a facility that enters the interconnection queue with incomplete load specifications suffers material delay.

Evaluate the Point of Interconnection feasibility and the utility coordination status before you rely on the build timeline.

The construction schedule is the visible number. The interconnection queue is the one that determines the return.

For public equity investors, the structural shift sits with the hyperscalers and their reference designs.

Meta, AWS, Microsoft, and Google now co-design the building, the power path, and the cooling loop as a single machine. Meta’s Louisiana campus was designed entirely for AI at ten billion dollars.

Compass Datacenters, using Schneider Electric’s prefabricated modular approach, compressed deployment from the industry-standard two to three years to under ten months with up to thirty percent TCO savings.

Standardization and prefabrication are becoming the competitive moat.

The operators still stick-building bespoke facilities on twenty-four-month timelines are ceding the AI demand wave to those who industrialized the design.

What This Costs the Ones Who Wait

Standard air-cooled construction reached 10.7 million dollars per megawatt in 2025 on JLL’s index, forecast to rise six percent to 11.3 million in 2026.

AI-ready facilities with liquid cooling regularly exceed twenty to thirty million per megawatt.

Those numbers are set at concept, not at handover.

Design is not the phase before the money moves. It is the phase where the money is decided.

The operators who locked their density, redundancy, and procurement sequence early will energize on schedule and capture the demand.

Everyone still treating design as documentation will pay for that mistake in stranded capital and missed operation dates, and they will pay it at a moment when the grid queue no longer has room for the correction.

Join Global Data Center Hub

Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

Türkiye has committed $3 billion in public funding to mobilize $10 billion in private data center and AI investment by 2030.

The capital design is the signal.

This is a deliberately engineered leverage structure aimed at building sovereign compute capacity, and the sophistication of the design is what separates it from the announcement-driven programs it will be compared to.

The Capital Structure Is Engineered, Not Announced

The $3 billion is not a procurement budget.

The Ministry of Industry and Technology structured it as anchor capital intended to crowd in roughly $10 billion in private investment, a 3.3x leverage target.

The mechanism underneath it is the HIT-30 program, a $30 billion high-tech incentive umbrella that allocates a dedicated $1.5 billion tranche to hyperscale and AI-compatible infrastructure.

Eligibility is gated with precision: minimum 30 MW IT load, at least 50 percent AI hardware compatibility, PUE at or below 1.4, and mandatory integration with cloud or AI platforms.

The state is not buying data centers.

It is underwriting the conditions under which private operators build them, and it is reserving the incentive for facilities at hyperscale density only.

This is disciplined capital-formation design.

The 2 percent mandatory allocation of public investment programs to AI projects manufactures a domestic demand baseline, which gives private operators a predictable state revenue anchor.

Türkiye has built a structure that de-risks private capital at the entry point and guarantees it a floor of demand at the exit.

That is the part institutional readers should study, because it is more sophisticated than the headline figure suggests.

Localization Has Already Converted Policy Into Demand

The clearest evidence the design works is that it has already pulled a hyperscaler commitment.

Data localization rules enacted in late 2024 require platforms with more than one million daily Turkish users to process data domestically.

That mandate converted a policy preference into a physical infrastructure requirement, and it compelled Google to commit $2 billion to a Turkcell cloud region rather than serve the market from Frankfurt.

Türkiye engineered captive demand through regulation and then attached incentives to it.

The localization mandate is doing exactly what capital-formation policy is supposed to do: it created the demand the capital now funds.

The structural advantages sit underneath this. Istanbul sits within 50 ms of Europe, MENA, and Central Asia.

Power runs at $0.08 to $0.10 per kWh, roughly 40 percent below European averages, and construction costs run below Gulf markets.

Türkiye is not competing on raw scale against Gulf programs.

It is competing on geography, latency, and cost, and that is a defensible position for the workloads it is built to serve.

The Grid Is the Sequencing Variable

The 1 GW installed-capacity target for 2030 sits against a current operational baseline of 160 to 471 MW, depending on methodology.

The gap is real, and the variable that governs how fast it closes is power.

Two-thirds of new renewable energy projects currently face grid connection approval delays, and the $10 billion national transmission upgrade is scheduled for completion in the early 2030s.

This is the sequencing question every allocator should track: not whether Türkiye can build the grid, but in what order the grid and the data centers arrive.

Grids get built, and Türkiye’s nominal generation capacity of 115 GW is not the constraint. Transmission timing is.

Renewable generation concentrates in central and eastern Anatolia and the Aegean coast, while demand concentrates in the industrial northwest, and the interconnection to bridge that gap is what paces conversion.

This does not make the roadmap fail.

It makes the roadmap a timing question, and timing questions reward the allocators who track interconnection milestones rather than announcement cadence.

The state is already responding, with operators testing liquid immersion cooling and feasibility work underway on naturally cooled sites to reduce the power draw per facility.

The Adequacy Question Is Optionality, Not a Ceiling

Türkiye does not yet hold an EU data adequacy ruling, and the divergence in hyperscaler behavior reflects it.

Google has committed $2 billion. AWS and Microsoft have not committed full cloud regions.

That split is the market pricing an open question rather than a settled verdict.

Read correctly, the adequacy gap is optionality: domestic and regional demand supports the buildout today, and an adequacy ruling would unlock the EU-regulated workload pool on top of it.

The base case does not depend on the ruling. The ruling is upside the current entrants are positioned to capture first

Investor Action

Private capital should treat this capital-formation framework as a reason to begin diligence now. The HIT-30 program improves project economics through tax incentives and energy subsidies, materially de-risking the capital stack for facilities exceeding the 30 MW threshold.

The variable to underwrite is interconnection timing.

The cost of waiting for the grid to visibly clear is entry at a repriced basis once the sequencing resolves. The window to structure at current land and power cost is open now.

Public Markets investors holding Turkish telecom and infrastructure equity should benchmark exposure against the state-linked operators carrying this buildout, which sit inside the sovereign wealth portfolio.

The 25 percent-of-revenue capital expenditure commitment from the anchor operator and the multi-billion concession pledges from the incumbents are the public-market expression of this roadmap.

Calibrate position sizing to grid execution milestones.

Operators and hyperscalers evaluating market entry should sequence on the adequacy question as an optionality trade.

The operator that structures land and power agreements now captures the latency arbitrage at the lowest basis, with an adequacy ruling as upside rather than a precondition.

The asymmetry favors early structuring for operators whose workloads are domestic or regional, and the adequacy ruling extends the addressable market on top of a case that already clears.

The Verdict

The capital commitment is credible and the leverage design is genuinely well-engineered.

Türkiye has built a capital-formation structure and a regulatory demand driver that most sovereign compute programs announce but do not deliver.

Whether $13 billion converts into 1 GW by 2030 is a sequencing question, governed by grid timing and adequacy positioning, and both are variables an allocator can track rather than risks that sink the case.

The market inflection arrives when the first hyperscale facility energizes at full load against a resolved interconnection.

Allocators should position for that moment now, while the basis is low.

The open question is the pace of conversion, and the next twelve months of grid-queue data will tell you whether to accelerate or wait.

Join Global Data Center Hub

Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

The Number That Felt Wrong

In 2024, an infrastructure equity team at a major institutional fund sat down to value a compute factory.

The asset was purpose-built liquid-cooled, GPU-dense, engineered from the component level for AI training workloads.

The offtake agreement was in place. The counterparty was credible. The power position was secured.

The team ran the model they had used to evaluate data center assets for fifteen years.

Cap rate applied to net operating income. Comparable transaction multiples from the colocation market.

Weighted average lease term as the primary duration variable. The number the model produced was discussed for three hours. Nobody in the room believed it.

The model was not wrong in its arithmetic. It was wrong in its assumptions.

It had been built for a different asset class and applying it to the compute factory produced a valuation that was simultaneously too high on the wrong variables and too low on the right ones.

The team rebuilt the model from first principles. This article is what they found.

What the Standard Model Was Built For

The colocation data center model that became the industry standard over two decades was built on real estate logic. The asset is the building. The value driver is the lease.

The risk is vacancy. The return is the spread between the cap rate on acquisition and the cap rate on exit, with the net operating income in between.

That model fit the colocation asset class precisely.

A colocation facility with diversified tenancy, staggered lease expirations, and a building that retains functional value across tenant changes is a real estate asset.

The model prices it correctly. The cap rate, the lease multiple, and the NOI walk all describe the same asset from different angles.

Open Compute Project: The Hardware Moat Facebook Built and Gave Away introduced the first crack in that framework. Two facilities at identical capacity produce different returns when one was engineered at the component level for its workload.

The design efficiency variable was the first sign that the standard model was measuring the wrong thing.

Two GPUs: The Infrastructure Shift That Priced Most Investors Out introduced the second crack. Workload compatibility determines whether the asset qualifies for the demand driving lease rates.

A facility at 95% occupancy on the wrong workload is not the same asset as one at 95% on the right one, regardless of what the NOI line reads.

The Retrofit Problem: Why Legacy Data Centers Cannot Serve AI Workloads named the consequence most of the global installed base cannot economically qualify for the workload driving the market.

The standard model prices that disqualification as a valuation discount when it arrives in the occupancy data. By then, the allocation decision has already been made.

The compute factory compounds all three cracks simultaneously.

It requires a framework built from different starting assumptions.

The Five Variables the Standard Model Misses

The compute factory valuation framework starts where the standard model ends.

Offtake duration, not lease term. Lease term is a reasonable measure of cash flow duration in diversified colocation facilities.

In a compute factory, however, the offtake agreement is the cash flow.

Its duration, renewal probability, and termination terms determine value, making a long-term hyperscaler contract materially more valuable than a shorter lease with weaker counterparties.

Counterparty credit quality, not occupancy rate. Occupancy is a useful revenue metric for multi-tenant data centers.

In a compute factory, where one or two customers generate nearly all revenue, counterparty credit quality is a far better measure of cash flow certainty and asset value.

GPU hardware cycle, not building depreciation. Traditional models assume decades of building value. Compute factories depend on GPU refresh cycles every two to three years and their ability to support each new hardware generation.

Power cost basis, not power capacity. Traditional underwriting asks whether a facility has enough power. The more important question is what it pays for that power.

Electricity cost directly determines the cost of compute and creates a lasting competitive advantage or disadvantage.

Design qualification rate, not square footage. Square footage measures physical capacity, but compute factories are constrained by thermal and power architecture.

A facility’s value depends on whether it can qualify for current and future AI workloads, not simply how much space it contains.

The Rebuild

The compute factory valuation framework sequences its inputs in the inverse order of the standard model.

Start with the offtake agreement.

Duration, counterparty credit quality, pricing structure under stress, and termination mechanics. These four variables produce the revenue profile.

The asset analysis follows from the revenue profile, not the other way around.

Layer in the power cost basis.

The compute factory’s operating cost structure is dominated by power expenditure.

A facility producing AI training compute at $0.04 per kilowatt-hour holds a structural margin advantage over a facility producing the same output at $0.08.

That margin advantage compounds at every scale point and is the primary driver of competitive durability.

Apply the GPU hardware cycle as the primary depreciation variable.

The productive asset is the GPU cluster. Its economic life is two to three years per generation.

The building’s value depends on its ability to support next-generation GPU density without structural changes. Thermal capacity, power distribution, and interconnect scalability are what matter.

Price the design qualification rate against the current and projected workload specification.

A facility that qualifies for today’s hyperscaler procurement specification holds a different forward value from one that requires retrofit to qualify.

The retrofit cost, against the lease rates available to a qualifying facility, is the valuation adjustment.

Close with the exit analysis.

The standard model exits on a cap rate applied to stabilized NOI. The rebuilt model exits on the durability of the offtake structure.

Whether the buyer remains committed, the facility can support the next GPU generation without major capital expenditure, and power costs remain competitive.

Three Positions on the Framework Rebuild

For private equity and infrastructure funds, the rebuilt model produces systematically different valuations. Rather than pricing stabilized cap rates, it values offtake duration, counterparty credit quality, GPU cycle exposure, and power cost basis.

The largest pricing gaps emerge where the standard model overvalues strong revenue streams with weak design specifications or undervalues technically superior assets because it focuses too narrowly on initial lease duration.

For public market investors, the rebuilt model exposes the disclosure gaps behind systematic mispricing. Traditional metrics such as occupancy, lease term, and NOI per square foot reveal current performance but not future value.

The real drivers power cost basis, counterparty credit quality, GPU cycle exposure, and design qualification are rarely disclosed.

Investors who can estimate these variables gain a durable information advantage over those relying solely on reported metrics.

For compute factory operators, the rebuilt model distinguishes operational improvements from true valuation events. Extending long-term offtake with a creditworthy counterparty directly increases value.

Efficiency gains such as improving PUE also matter, but not simply because they reduce operating costs.

They strengthen design qualification, expand the addressable tenant base, and reduce GPU cycle exposure, producing a materially different return on the same capital expenditure.

The Framework the Capital Series Built

This series has assembled one analytical framework across five articles.

Why Meta’s $26B Leaseback Rewrote AI Infrastructure Financing established the ownership structure.

In AI Infrastructure, the Offtake Agreement Is the Asset identified the primary credit instrument.

Why Sovereign Wealth Funds Are Becoming the Compute Factory’s Anchor Capital named the anchor dynamic that changes the cost of capital for every investor in the structure.

What the Hyperscaler Balance Sheet Actually Tells Investors About AI Infrastructure decomposed four capital strategies that a single capex multiple cannot capture.

This article assembles the valuation framework those four articles were building toward.

The compute factory is an industrial asset. Its value depends on the offtake agreement, power cost, GPU hardware cycle, and design qualification in that order.

The standard data center model misprices the compute factory because it applies the wrong framework.

Capital that rebuilds the model will identify assets the market is systematically mispricing.

The next series applies this framework to emerging compute factory markets, where sovereign demand, power economics, and technology transfer determine outcomes.

Join Global Data Center Hub

Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

Permitting is the longest phase in data center development, and it is the one most newcomer’s underestimate.

Land, power, and cost get the early attention.

The approvals that turn a viable site into a buildable one take longer than any of them, and they run on their own logic.

This piece walks through that process in the order a developer actually encounters it.

Start with scale. Data center development typically takes 3–6 years, with permitting and approvals alone often requiring 6–18 months.

Between May 2024 and March 2025, at least $64 billion in U.S. projects were delayed or blocked by regulatory hurdles and local opposition.

Understanding the permitting process is essential for setting realistic timelines.

The Process Is Not One Approval. It Is Eight Running in Parallel.

The first thing to understand is structure.

Permitting is not a single approval a developer waits on.

It is roughly eight distinct workstreams zoning, environmental review, building permits, fire and life safety, grid interconnection, water, air, and utility coordination.

They do not happen one after another. They run at the same time, each handled by a different agency, each on its own timeline.

A delay in one can force changes in another.

Learning the process means learning how these workstreams interact, not just what each one requires.

Zoning and Land Use Come First

Zoning is where most projects begin, because it determines whether a data center is even an allowed use on a given parcel.

Historically, data centers occupied an ambiguous position in municipal codes, and many jurisdictions permitted them “by-right” meaning the use was allowed without discretionary review or public hearings.

That is changing. In 2000, Loudoun County, Virginia the world's largest data center market classified data centers as office parks, allowing by-right development in key industrial zones.

In March 2025, the Board of Supervisors replaced that with a Special Exception requirement, adding public hearings and a Board vote to every new application.

Fairfax County followed with equipment enclosure rules, a one-mile Metrorail buffer, and 200-foot residential setbacks.

The educational point is that zoning posture is jurisdiction-specific and changes over time.

A market that was straightforward to build in five years ago may require a full discretionary process today.

Reading the current zoning treatment of a specific county is the starting point of the entire timeline.

Environmental Review Runs in Two Layers

Environmental review is the workstream with the longest potential tail, and it operates at two levels.

At the federal level, the National Environmental Policy Act (NEPA) applies when a project involves federal land, permits, funding, or federally regulated resources.

Reviews range from Categorical Exclusions requiring little analysis to full Environmental Impact Statements (EIS), which assess alternatives, cumulative impacts, and public input and can take two years or more.

A 2025 executive order expanded fast-track eligibility for qualifying large projects, though implementation still depends on agency rulemaking and legal challenges.

State environmental reviews typically run separately from federal reviews, creating multiple approval layers.

In Pennsylvania, a single project may require permits for earth disturbance, stream crossings, sewage planning, and water withdrawals each with its own agency, timeline, and documentation.

Coordinating these parallel approvals is a core responsibility of the developer.

Air Permitting Depends on How the Permit Is Classified

Air permitting has become more complex because large data center campuses can house dozens or even hundreds of diesel backup generators.

Facilities with a Potential to Emit above 100 tons per year are classified as major sources and face stricter permitting. Most data centers avoid this by operating as synthetic minor sources with binding operational limits.

These limits cap how many hours generators can run for non-emergency purposes, typically between 50 and 100 hours per year depending on the permit.

Emergency use during grid outages is not federally limited, provided all operating hours are properly logged.

Standards are tightening in major data center markets. In Virginia, air permit applications for diesel generators submitted on or after July 1, 2026 must meet Tier 4-equivalent controls, increasing costs and raising the bar for future projects.

On-site primary generation is also evolving. In the Vantage VA2 case in Loudoun, a data center received a minor air permit for natural gas turbines because emissions stayed below major-source thresholds.

Virginia has since updated its rules, requiring a Special Exception for associated power generation facilities and closing regulatory gaps exposed by gigawatt-scale projects.

Water Permitting Is Becoming More Significant

Water is a tightening part of the process as cooling demand draws regulatory attention.

Any discharge from cooling operations into US waters requires an NPDES permit.

Ohio EPA has proposed a draft general NPDES permit for data center discharges, which streamlines the approval while codifying discharge limits.

Construction that disturbs one or more acres requires Construction General Permit coverage and a Stormwater Pollution Prevention Plan before ground-breaking.

In water-stressed regions, withdrawal permits have moved from routine to project-defining.

Grid Interconnection Is Usually the Longest Single Item

The workstream that most often sets the true timeline is grid interconnection.

Virginia’s DEQ permitting queue now estimates wait times of up to seven years for data center grid connections in Northern Virginia.

Interconnection studies alone can run longer than all other development phases combined.

Dominion Energy signaled as early as 2022 that it could not supply new Loudoun data centers until at least 2025 to 2026.

For anyone learning the process, this is the item to map first, because it frequently determines when a project can actually energize.

How Different Players Experience the Same Process

The same process looks different depending on who is moving through it.

For independent operators, the work centers on the completeness of the application. Incomplete or placeholder submissions restart review timelines, and a single resubmission cycle can add three to nine months.

Smaller entrants without established agency relationships tend to absorb those delays more heavily, while experienced developers move the same review faster on the strength of their track record.

For private equity and infrastructure investors, the central concern is carry cost. A 12 to 18 month permitting delay on a one-billion-dollar project, at a 7 percent cost of capital, represents 70 to 105 million dollars before any revenue.

For this group, understanding the process means underwriting permitting timelines as a real financial variable jurisdictional history, local posture, developer track record rather than treating them as a fixed assumption.

For public equity investors, the relevant dynamic is how policy shifts move across markets at once. New York’s legislature passed a one-year moratorium on data centers of 20 MW or more on June 4, 2026, pending the Governor’s signature. At least 14 states have proposed or enacted similar measures.

For investors holding operators concentrated in tightening markets, the effect is felt across a whole footprint, not one project.

The Process Is Easier in Some Jurisdictions Than Others

The final thing to understand is that the process is not uniform. As several mature US markets add steps, other jurisdictions are streamlining.

Singapore reopened a multi-year moratorium in April 2026, approving four facilities under defined efficiency conditions PUE at or below 1.3, fully renewable energy certificates, and liquid cooling for at least 60 percent of IT load.

Gulf states can issue approvals in 48 hours for credentialed operators. Texas processes interconnection faster than PJM and imposes no Tier 4 mandate.

The practical lesson for anyone learning this process is that jurisdiction selection shapes the timeline as much as any single permit.

The markets that move quickly today may tighten as they mature, just as Loudoun did.

Understanding where a given jurisdiction sits on that path is the foundation of a realistic development plan.

Join Global Data Center Hub

Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

The Binding Constraint Is Fabrication, Not Capital

China has committed roughly $295 billion dollars over five years to a sovereign AI computing grid, and the capital is the part everyone is watching.

The capital is not the constraint.

The constraint sits one layer down, in a single domestic foundry running above 93 percent utilization at a node two generations behind the frontier, and in a domestic high-bandwidth memory packaging base that cannot yet yield at volume.

The program guarantees the money and guarantees the market. It cannot guarantee the silicon.

That gap, between funded demand and constrained supply, is the entire investment question, and the headline figure is built to obscure it.

Guaranteed Demand Is Not The Signal Investors Think It Is

The first read is that demand-side certainty has been engineered to an unusual degree.

State policy now requires any state-funded data center project to source at least 80 percent of its underlying technology domestically.

The May 2026 Anke certification converted that mandate into a procurement catalog, listing nine processors from seven domestic entities, with provincial bureaus and state-owned enterprises legally restricted to certified silicon.

This manufactures a captive market. It also concentrates risk.

Demand certainty is only an asset if supply can meet it, and a guaranteed buyer for a chip that cannot be fabricated at volume is a queue, not a market.

For the next 12 to 24 months, the binding question is not whether demand exists.

The state has legislated that it does. The question is throughput.

The Anke List Is A Map Of Winners And A Map Of Exclusions

The second read is in who the certification names and who it omits. The Anke catalog certifies Huawei Ascend, Alibaba T-Head, Biren, Hygon, Iluvatar, MetaX, and Moore Threads.

It omits Cambricon and Baidu-backed Kunlunxin. Because state-owned facilities are legally restricted to certified silicon, omission is not a delay. It is exclusion from the procurement channel that drives the majority of high-end domestic demand.

Cambricon reported first-quarter 2026 revenue of 423 million dollars and strong growth, and is still outside the catalog.

The signal is that the certification list, not the market, now allocates demand.

For an investor in Chinese chip designers, the diligence question has changed: financial performance is secondary to catalog inclusion, because inclusion is the licence to sell into the only buyer that matters at scale.

Position on the list, not on the income statement.

Carrier-Operators Are A Structural Source Of Stranded Capacity

The third read is who builds and operates the grid. The state has tasked telecom carriers, China Mobile and China Telecom, with constructing and operating the bulk of the national facilities.

S&P Global Ratings notes that traditional carriers often lack experience building the high-power-density, low-PUE facilities that generative AI training requires.

The operators with a proven record of delivering optimized facilities for the private AI labs, the carrier-neutral providers, sit largely outside the sovereign funding flows.

This is a structural mismatch between who holds the capital and who holds the capability.

The consequence is concrete: state capital may fund facilities that cannot host the workloads they were built for, leaving local governments with debt against idle infrastructure.

The risk is not theoretical demand softness.

It is capable demand routed to incapable supply by the funding structure itself.

Investor Action

Private Capital. Infrastructure and private equity allocators evaluating exposure to China’s domestic chip and data center layer should underwrite the supply chain, not the program size. The diligence sequence runs through SMIC wafer allocation, HBM packaging yield, and Anke catalog position, in that order. A certified designer with no secured wafer allocation is a brand with a licence and no product.

The cost of underwriting the 295-billion-dollar figure as a proxy for deliverable capacity is that an allocator funds a queue and books it as an asset.

Price catalog inclusion and secured fabrication capacity as the value drivers; treat the headline program as context, not thesis.

Public Markets. Public equity investors hold the cleanest instrument to express this view, because the bottleneck is observable. SMIC capacity utilization, HBM packaging progress, and Anke catalog revisions are trackable signals, and they lead the buildout. Benchmark certified designers on secured wafer allocation rather than order books, because an order the foundry cannot fill is not revenue.

The Cambricon case is the template: strong reported growth, catalog exclusion, and a step-change in risk that the income statement did not show.

Position on the spread between certified-and-supplied and certified-but-unsupplied names and watch the certification list as the event that moves it.

Operators. Data center developers and the carrier-operators themselves face the execution risk directly. A facility built without the power density and PUE that AI training requires is a facility built into the idle tier, regardless of how much state capital funded it.

The carrier-neutral operators with the delivery record have the capability the sovereign grid needs and the funding access it does not extend to them.

The structural read is that capability and capital are misaligned by design, and the operators that close that gap, through partnership, mandate, or acquisition of capability, capture the workloads the carrier builds cannot host. Underwrite delivery capability as the scarce input, not capital.

The Verdict

China's sovereign computing program is the most fully funded and most fully mandated AI infrastructure buildout in the world, and that is precisely why the supply constraint is the story.

When demand is legislated and capital is guaranteed, the only variable left is whether the physical supply chain can deliver, and this one runs through a single foundry above 93 percent utilization, an immature HBM packaging base, and operators who may not be able to build what the workloads require.

The program targets full operational capability by 2031, a date chosen as strategic insurance against the decoupling it anticipates.

The open question for the next 24 months is whether SMIC and domestic packaging can scale fast enough to convert funded demand into delivered compute.

Or whether physical infrastructure outpaces the certified silicon available to fill it, leaving the grid as a funded promise waiting on chips.

Underwrite the bottleneck until the fabrication numbers say otherwise.

Join Global Data Center Hub

Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

The Earnings Call Question

In the first quarter of 2025, four public companies reported capital expenditure commitments collectively exceeding $300 billion for the fiscal year.

Amazon. Microsoft. Google. Meta. Each CEO fielded the same question from equity analysts on their respective earnings calls: when does the return on this capital become visible?

The question was right. The assumption embedded in it was wrong.

The assumption was that four companies committing $300 billion to AI infrastructure were making the same bet: the same buildout, scaling logic, and path to monetization through usage fees.

But the balance sheets told four different stories.

The capital strategies pointed in four different directions. Applying a single capex multiple across all four compresses those differences into noise and misprices each of them.

How Hyperscaler Capex Has Always Been Read

Public equity analysts tracked hyperscaler capital expenditure as a single category for twenty years.

The metric was legible: capex rises when the company invests in growth, falls when the cycle turns, and return on invested capital emerges with a lag as infrastructure fills and utilization rises.

This framework worked through the cloud era. Amazon, Google, and Microsoft were each pursuing versions of the same strategy build owned infrastructure, sell metered access, achieve utilization economics at scale. The capital structures were similar.

The return profiles were similar. Comparing capex across the three was a useful exercise because the underlying logic was the same.

Why AWS EC2 Rewrote the Economics of Compute Ownership established why that convergence made sense: EC2’s utility model created an asset class with a defined cost structure, revenue model, and competitive dynamic. The analysts who modeled it correctly captured the returns.

The AI infrastructure buildout has broken that convergence.

Four companies are committing record capital to the same general category AI infrastructure through four materially different capital structures with four materially different return profiles.

The cloud-era framework maps none of them correctly.

Four Companies, Four Strategies

Amazon is building and owning AWS infrastructure at hyperscale the strategy Why AWS EC2 Rewrote the Economics of Compute Ownership traced from its 2006 origins.

Capital expenditure sits on-balance sheet. The return accrues to AWS margins as utilization rises against the fixed cost base.

The risk is the capital commitment itself approximately $80 billion in annual capex against a utilization curve that must materialize over a two-to-four year lag.

Microsoft is pursuing a hybrid strategy. Owned infrastructure build for Azure, supplemented by committed offtake agreements with CoreWeave and a deep financial relationship with OpenAI.

The Microsoft balance sheet carries the owned build as capital expenditure and the CoreWeave commitments as future lease obligations two different line items, two different return profiles, one strategic position.

The CoreWeave relationship applies industrial offtake logic to an equity investment thesis in a way no prior hyperscaler relationship has.

Google is the most vertically integrated of the four. Custom silicon the Tensor Processing Unit custom network fabric, custom cooling architecture, wholly owned facilities.

The capital intensity is comparable to Amazon’s. The design specificity is materially higher.

The return profile depends on Google maintaining the technical advantage embedded in its custom stack a bet that the factory’s design moat compounds rather than commoditizes with each GPU generation reset.

Meta is the outlier. Why Meta’s $26B Leaseback Rewrote AI Infrastructure Financing traced the leaseback structure private credit builds and owns, Meta leases back.

Operational control is separated from asset ownership. Meta’s owned capex line is lower than the other three. Its operating expense line carries the lease obligations.

The balance sheet reads differently from all three competitors and most equity frameworks are not yet correctly adjusting for the distinction.

Four Questions the Framework Must Answer

Four strategies require four distinct questions. Applying the same question to all four produces the wrong read on each.

For Amazon, the right question is utilization velocity. Owned infrastructure earns its return as AWS workloads scale against a fixed cost base. The capex commitment is front-loaded, while the return is back-loaded.

The timing gap between commitment and return is the primary valuation variable and compressing that gap through aggressive demand generation is the operational lever that justifies the capital commitment.

For Microsoft, the right question is integration premium. The hybrid strategy produces its return only if owned build, CoreWeave offtake, and OpenAI relationship compound together. If the three components operate independently, the blended return is lower than any single-strategy alternative.

The integration bet is real but requires a different analytical tool from utilization-lag modeling.

For Google, the right question is design moat durability. The vertically integrated strategy earns its premium return when the custom silicon and custom architecture maintain a cost-per-inference advantage against commodity GPU clusters. Each new GPU generation resets the competitive question.

The return is real when the moat holds. It compresses when it does not and no publicly visible metric tells an outside investor in real time which is happening.

For Meta, the right question is lease obligation visibility. The leaseback model produces a smoother near-term free cash flow profile than the other three the capital burden is spread across lease payments rather than front-loaded into capex.

The cost of that smoothness is operating leverage: the lease obligations are fixed costs that do not flex with utilization. If AI training demand softens, Meta’s operating expense base adjusts more slowly than Amazon’s or Google’s capex.

The Complication

The divergence in capital strategies produces a divergence in return timing that standard earnings frameworks cannot fully capture.

Amazon and Google are committing capital that produces returns on a utilization lag.

The free cash flow is suppressed in the near term and recovers as infrastructure fills.

Valuing either on near-term free cash flow multiples during the capex deployment phase systematically undervalues the asset being built and consistently punishes the management teams making the most strategically correct capital allocation decisions.

Microsoft’s hybrid strategy produces a blended return timeline harder to model than either pure-play alternative. The owned infrastructure follows the Amazon pattern.

The CoreWeave offtake and OpenAI relationship produce returns through revenue attribution that may not map cleanly to the capital committed to them.

The integration premium is real when the components compound and invisible when they do not.

Meta’s leaseback strategy produces the smoothest near-term free cash flow. The capital burden is spread across operating expense rather than concentrated in capex.

The trade-off is operating leverage: lease obligations are stickier than capex under demand softness.

If AI training demand grows as Meta expects, the leaseback is optimal. If not, fixed operating expenses create earnings pressure that owned-infrastructure competitors can mitigate through capex deferral.

Three Positions on the Hyperscaler Balance Sheet

For public equity investors, the four capital strategies require four distinct valuation frameworks applied concurrently. Amazon warrants a utilization-adjusted infrastructure value approach.

Google warrants a design moat premium that decays with each GPU generation reset. Microsoft warrants an integration premium that is real only if the hybrid components compound.

Meta warrants an operating expense visibility model that prices lease obligation durability alongside the free cash flow advantage. The analyst applying one framework to all four will be consistently right on none of them.

For private credit investors and infrastructure lenders operating in each company’s supply chain, the four strategies produce four different counterparty credit profiles. Amazon and Google counterparties carry utilization risk.

Microsoft counterparties carry integration dependency risk. Meta counterparties carry the cleanest credit profile: the leaseback is already structured as an offtake agreement, and the framework In AI Infrastructure, the Offtake Agreement Is the Asset identified applies directly.

For a lender choosing between supply chain positions, the Meta supply chain carries the most legible credit structure of the four.

For sovereign and institutional capital allocators evaluating AI infrastructure exposure, the four strategies identify four different vectors of risk concentration in the broader market. A portfolio with significant exposure to all four hyperscalers through index positions carries four distinct capital structure risks that index weighting obscures.

Decomposing the hyperscaler capex story into four separate capital thesis evaluations produces a more accurate picture of where the transition premium actually accumulates and where the capital structure risk compounds if demand materializes slower than any single company’s commitment assumes.

What the Balance Sheet Cannot Tell You

The hyperscaler balance sheet tells you the capital strategy. It tells you the return timeline. It tells you the operating expense structure and the free cash flow profile under different utilization scenarios.

It does not tell you how to underwrite the assets those capital strategies produce or the independent assets competing alongside them.

A compute factory financed through Meta’s leaseback model, an Amazon-adjacent colocation facility, and a standalone CoreWeave data center each carry different return profiles that the hyperscaler balance sheet framework does not directly address.

That requires a valuation methodology built specifically for the compute factory era one that starts with the offtake agreement, accounts for the anchor capital stack, and prices the design efficiency variable Open Compute Project: The Hardware Moat Facebook Built and Gave Away identified.

The standard data center investment model, applied to compute factory assets financed through AI-era capital structures, produces the wrong answer with the consistency of a framework applied to the wrong asset class.

That is the subject of the final article in this series.

Join Global Data Center Hub

Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

The Contract Is The Signal, Not The Capacity

Meta contracted 1.6 gigawatts of compute from a developer it does not own rather than building the capacity itself.

The mainstream coverage fixed on the gigawatt number and the missing price.

The number that should move you is the one nobody disclosed, because the absence of a price tells you what the contract is doing.

This is a frontier hyperscaler routing its scarcest input through a third-party balance sheet.

The build-versus-buy line in AI infrastructure has moved, and the operators who do not reprice against it will underwrite the wrong risk for the next two years.

Offtake Is Becoming The Collateral That Underwrites Developer Debt

Crusoe does not fund gigawatt campuses on its own credit.

It funds them on contracted demand.

The Abilene flagship sits inside a $15 billion joint venture with Blue Owl Capital and Primary Digital Infrastructure, structured as a forward takeout against an Oracle build and an OpenAI lease.

The Meta contract extends the same logic.

A named hyperscaler commitment converts a development site into a financeable asset, and the lender underwrites the tenant, not the developer.

Expect this pattern to harden over the next 12 to 24 months.

The developers who control energized sites and can attach an investment-grade offtake will raise capital at terms unavailable to operators carrying merchant capacity risk.

Power plus a signed tenant is now the financing unit, and capital will price that bundle at a premium.

Counterparty Concentration Has Moved From The Tenant To The Developer

Crusoe holds 4.9 gigawatts under contract against a pipeline exceeding 40 gigawatts.

The contracted book now spans Oracle, OpenAI, Microsoft, Google, and Meta.

One developer sits at the center of several of the largest US AI infrastructure projects.

That concentration is a strength for Crusoe and a risk for everyone underwriting it.

The dependency runs both directions.

Crusoe’s lenders carry exposure to a handful of anchor tenants, and any hyperscaler that slips a delivery date or renegotiates an offtake transmits stress straight into the developer’s capital stack.

Crusoe already exited a Wyoming project this month after a counterparty path collapsed.

Diligence on any Crusoe-adjacent exposure now requires mapping the full tenant book, not the single contract in front of you.

Power-First Siting Has Replaced Hub Proximity As The Underwriting Variable

The sites reported in this deal sit in Childress, Texas, and Warrenton, Missouri, not in Northern Virginia or Santa Clara.

Crusoe’s model locates compute where power already exists rather than fighting interconnection queues in saturated hubs.

The 1.2 gigawatt Abilene campus pairs with a dedicated power strategy spanning gas, renewables, and 12 gigawatt-hours of iron-air battery storage. The siting choice is the underwriting choice.

For the next cycle, the question is not whether a market has tenants or land.

The question is whether it has deliverable, energized megawatts before the build begins.

Markets that solve power first will capture the AI capacity that saturated hubs can no longer physically accommodate.

Investor Action

Infrastructure and private equity funds should re-underwrite digital infrastructure equity around offtake quality, not asset location.

The asset that holds value is the energized site with a contracted investment-grade tenant. Diligence the tenant book and the power strategy before the building.

Funds that continue to price on land and proximity will overpay for assets carrying merchant power risk the market has already learned to discount.

Public equity and credit investors holding hyperscaler names should read this as a capital-efficiency signal.

Meta is securing capacity without carrying the full asset on its balance sheet, which preserves capital against its commitment reaching up to $600 billion through 2028.

Credit investors evaluating developer-adjacent paper should price the tenant concentration directly.

Treat a Crusoe-style contracted book as a single-name exposure cluster, not a diversified pipeline, and benchmark spreads against that read.

Developers and operators should sequence power procurement and offtake before capital raises. The financing follows the signed tenant, not the other way around.

Operators carrying speculative capacity without a named anchor will find the bank syndicate and the private placement market both tightening against them. Negotiate the offtake first.

The operators who control energized sites hold the leverage in every conversation that follows.

The Verdict

This contract marks the point where compute capacity became a procured input rather than an owned asset for the largest buyers in the market.

The inflection is structural.

Hyperscalers are converting their balance sheet advantage into contracted demand, and developers are converting that demand into financeable capacity.

Over the next 24 months, the firms that win will be the ones controlling energized sites with investment-grade offtake attached, because that bundle is what capital now prices at a premium.

The open question is the concentration it creates.

When five hyperscalers anchor one developer’s 40 gigawatt pipeline, who holds the risk when the first anchor moves, and has anyone underwriting that pipeline actually priced it?

Join Global Data Center Hub

Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

Meta Platforms has contracted approximately 1.6 gigawatts of artificial intelligence computing capacity from data center developer Crusoe.

Bloomberg first reported the agreements, citing people familiar with the private discussions.

The capacity supports Meta’s AI infrastructure expansion across two US sites.

The agreements were reported on June 18, 2026, with no disclosed delivery timeline.

Sites And Capacity Under Contract

The capacity spans two Crusoe-developed data centers.

Reports place the facilities in Childress, Texas, and Warrenton, Missouri.

The two sites deliver roughly 1.6 gigawatts combined.

One gigawatt supplies electricity to approximately 750,000 US homes.

Meta is under contract to buy computing capacity from both facilities.

Neither Meta nor Crusoe confirmed the agreements on the record, and the financial value of the contracts was not disclosed.

Crusoe’s Contracted Position

Crusoe is a vertically integrated AI infrastructure company founded in 2018.

The company reported 4.9 gigawatts of contracted AI infrastructure capacity earlier in June, across data center projects and its Crusoe Cloud platform.

Its total development pipeline exceeds 40 gigawatts.

The contracted portfolio spans five US campuses.

Existing relationships include Oracle, OpenAI, Microsoft, and Google.

The Abilene Anchor And Wider Pipeline

Crusoe’s flagship project is a 1.2 gigawatt campus in Abilene, Texas, purpose-built for Oracle and leased to OpenAI on land owned by Lancium.

Crusoe, Blue Owl Capital, and Primary Digital Infrastructure entered the second phase of a $15 billion joint venture to fund that campus.

Crusoe recently broke ground on a second 900 megawatt Abilene campus contracted to Microsoft.

The company is also contracted to build additional large-scale campuses in Texas and Missouri.

Recent Pipeline Movement

Crusoe raised $1.37 billion in a Series E round co-led by Valor Equity Partners and Mubadala Capital, lifting its valuation above $10 billion.

The company recently exited a planned Wyoming data center project after earlier talks with Google did not proceed.

Meta was previously reported to be in discussions for capacity at the Abilene site later contracted to Microsoft.

The Crusoe agreements deepen Meta’s contracted compute position as the company executes against a stated AI infrastructure commitment that reaches up to $600 billion through 2028.

Join Global Data Center Hub

Welcome to Global Data Center Hub. Join investors, operators, and innovators reading to stay ahead of the latest trends in the data center sector in developed and emerging markets globally.

Join Global Data Center Hub

Power deliverability now determines whether land is a viable data center site or just speculative real estate.

Many projects still prioritize land before power, but without a credible energization path, the site isn’t an asset it’s only an option on an infrastructure constraint.

The shift happened quickly.

Transmission constraints are now the top grid bottleneck, with utilities quoting 4–10 year connection timelines and up to 12 years just for studies.

While capacity is expected to grow ~14% annually through 2030 (adding ~100 GW and doubling footprint), demand is outpacing the grid.

The core question has shifted from where to build to where power is available, reshaping the entire acquisition process.

The Sequence That Worked for a Century Stopped Working

Start with the history, because the inversion is recent.

For most of commercial real estate, developers identified land first and engaged utilities second.

That worked when grid capacity was abundant and a connection was a formality.

It stopped working when hyperscale and AI load collided with a transmission system that takes seven to eleven years to permit new regional lines.

The PJM interconnection queue tells the story in two figures that measure different things.

One is scale: PJM studied 294 GW of projects between 2020 and 2026, of which only 23 GW reached service.

The other is attrition: 74% of studied projects ultimately withdrew. Scale reflects demand, but attrition shows how little survives the process. Both figures point to a system where land was never the constraint.

What a 9x Move in Eighteen Months Actually Priced

The value of solving for power shows up directly in land prices.

In Stafford County, Virginia, Peterson Companies assembled 504 acres for $32.25 million across 2023 and 2024, roughly $64,000 per acre.

The firm secured data center entitlements in September 2024, then sold the entitled site to Stack Infrastructure in January 2025 for $302.3 million, approximately $600,000 per acre.

That is close to a 9x move in eighteen months. The land did not change.

What changed was certainty entitlement and a power path converted raw acreage into a deliverable site.

More broadly, entitled, power-ready land now trades at two to four times the value of standard industrial parcels, and power-ready land in Ashburn exceeds $3.5 million per acre.

Site Control Is the Entry Ticket, Not the Purchase

This is where the acquisition process has to be understood as a structure, not a purchase.

Site control comes first outright purchase, ground lease, option, joint venture, or land banking and it is the prerequisite for everything downstream.

Utilities will not study a load application without proof of site control.

Lenders will not advance pre-development capital without it.

Under PJM’s first-ready, first-served rules, a developer must demonstrate continuous, exclusive site control or face non-appealable removal from the study queue.

Control is no longer a real estate step. It is the entry ticket to the power process.

The Scarcity Is Not Where the Market Thinks It Is

The narrative has focused on land scarcity in Tier 1 markets.

That misses the actual constraint. The capital reality is that scarcity is not about acreage at all.

Northern Virginia carries roughly 2,745 MW across more than 150 facilities with core Loudoun County vacancy below 1%.

Developers are pushing into Stafford, Spotsylvania, Culpeper, and secondary markets in Iowa, Ohio, and West Virginia.

They are not chasing land. They are chasing power and permission.

The parcels exist. The deliverable parcels do not.

Three Lenses on the Same Parcel

Independent operators face the qualification problem first. Sites must meet soil capacity for heavy loads, usable buildable area after constraints, water availability, and fiber diversity before even reaching tenant discussions.

Many fail late because power, land, environmental, and community risks were assessed separately.

The discipline that prevents this is concurrent evaluation assess all constraints together before committing capital.

Private equity and infrastructure investors face a different issue: underwriting power that does not yet exist. Power must be segmented into four buckets existing capacity, confirmed expansion, queued capacity, and developer-controlled supply.

Only the first two are bankable. Investments based on queue positions are exposed to withdrawal risk.

The key question is not potential power, but defensible power in underwriting.

Public equity reflects a structural shift in financing. Joint ventures now dominate hyperscale development. Meta’s Hyperion campus (~$27B) brought in Blue Owl Capital with an 80% stake while Meta became a tenant, supported by PIMCO debt.

Equinix applies a similar 80/20 xScale model, and Digital Realty has shared billions in development with partners across major markets.

The signal is clear: hyperscale infrastructure is too capital-intensive and power-constrained to finance alone.

Regulation Is Tightening the Screw Power Already Turned

The regulatory environment is now tightening the screw that power already turned.

Permitting has become the stage where projects die one analysis attributes 77% of project delays and cancellations to the permitting and regulatory phase.

Since 2025, at least twenty moratoria have been proposed across local, state, and federal levels.

Georgia’s HB 1059 proposes a development moratorium running from July 2026 through December 2028.

Public opposition blocked or delayed an estimated $64 billion in US developments between 2024 and 2026.

The DATA Act of 2026 moves the other direction, proposing to exempt off-grid data centers from federal oversight so they can build their own power and bypass the queue entirely.

The legislative picture is unsettled, and unsettled regulation is itself a cost.

You Are Underwriting Timing Now

You are underwriting timing now, not land. The developer who secured Stafford acreage before entitlement captured the 9x.

The investor who waits for a confirmed interconnection agreement before committing will pay the entitled price, not the raw one.

Grid connection delays alone stretch launch timelines by an average of one and a half to two years, and in a market where being twelve to eighteen months late can cede a hyperscaler tenant to a faster site, that delay is the whole contest.

The land will still be there.

The power, the permission, and the position will already belong to whoever moved first.

Join Global Data Center Hub