Q3 2026 Cross-Asset Report · Out Now Subscribe to The Allocation Desk to read it today →

Make better-informed real estate decisions.

Independent research and underwriting for allocators, designed to fit alongside your existing investment process.

Start Monthly · $175/mo
The Allocation Desk

What you get as a subscriber.

Who it's for
RIAs, family offices, wealth platforms, and other capital allocators.
Quarterly Cross-Asset Report & Briefing
The full sector outlook, plus a live briefing and Q&A with William McSweeney.
Monthly Sector Brief
What moved, the risks worth watching, and what it means for allocation and underwriting.
Chart of the Week
One chart, one development, and whether it supports or challenges our outlook.
The Allocation Memo · Full Workbook
Sector deep dives and the data and assumptions behind The Allocation Memo.
Subscriber Introduction
A 30-minute call with the team to share your firm's research priorities.
$175 / month, or $1,680 / year ($140/mo)
Start Monthly · $175/mo →
Experience
Merrill Lynch Merrill Lynch Merrill Lynch Merrill Lynch

What we do

✓Underwrite live opportunities independently
✓Monitor positions against the original thesis
✓Prepare committee-ready conclusions
✓Custom & white-labeled research

What we never do

✕Act as a placement agent
✕Raise capital for sponsors
✕Solicit or offer securities
Press

As featured in

Research informed by Steuart AI, our internal cross-asset engine, and interpreted by our team.

Contact us

Talk to the team.

Questions about the subscription or a custom engagement.

Preview our research

Read before you subscribe.

See our prior representative issue, then start monthly when you're ready.

Follow us

Stay close to the research.

Firm updates and The Allocation Memo on LinkedIn.

EM Capital provides independent research, underwriting, and monitoring support to allocators. EM Capital does not act as a placement agent, raise capital for sponsors, solicit or offer securities, or provide legal, tax, or investment advice.

The Price Changes the Answer on Data Centers

The Price Changes the Answer on Data Centers

November 19, 2026•8 min read

Blackstone’s Amazon deal tests the price objection behind our Q3 underweight call.

Custom HTML/CSS/JavaScript

We were underweight data centers in our Q3 Cross-Asset Report. At a 7% entry cap, the same model puts them above our return hurdle.

Blackstone Digital Infrastructure Trust’s announced agreement to acquire interests in Amazon-leased data centers at a 7% initial cap rate gives that change a concrete reference point. In our existing base case, replacing the report’s 5.74% entry cap with 7% raises the modeled annualized unlevered return from 5.5% to approximately 10.7%, above our 8.3% pricing benchmark.¹ ²

For an allocator, that is a reason to prioritize offers with long-term hyperscaler leases at yields around 7%. At that price, our Q3 framework supports a more favorable view.

The price changes the answer

Our Q3 underweight reflected a strong demand story offered at a price that left too little return. We gave data centers our strongest favorable adjustment to the sector’s historical exit spread. Even with that support, the modeled return remained 2.8 percentage points below its benchmark.

The entry cap measures annual property income relative to the purchase price. The 7% case keeps our operating outlook and expected sale valuation while buying that income at a better yield. It does not require a stronger AI demand forecast.

Custom HTML/CSS/JavaScript

Five-year model, before financing, investor fees and taxes. The 7% column reconstructs the report’s calculation from rounded inputs, retaining its operating contributions and 3.5% exit Treasury, which produces a 5.96% exit cap. It is an entry-price sensitivity, not a forecast of BXDC’s return. Calculation details appear below.²

A higher starting yield helps twice in this model. It buys more income per dollar invested. It also changes the relationship between the purchase price and the expected sale price. The original case entered at 5.74% and exited at 5.96%, creating a small valuation drag. Entering at 7% while retaining the same exit assumption makes that contribution positive.

That second benefit matters. About 3.3 percentage points of the 10.7% modeled return come from the lower exit cap. Sell at the same 7% cap instead, with the other published contributions unchanged, and the result is approximately 7.5%. The base case works at the new entry price; the sale assumption still needs to earn its place.

Custom HTML/CSS/JavaScript

Our report had already put a number on the price that would change the call. Its base-case Accumulate threshold was an entry cap above 6.64%, holding other assumptions constant. Seven percent clears it. The conclusion follows from the framework we published before this transaction.

What the landlord is buying

The announced transaction would give BXDC a 75.1% interest in joint ventures owning 13 newly constructed powered-shell buildings. The approximately $5 billion aggregate gross value covers the whole portfolio. Amazon affiliates retain the balance.¹ ³

Amazon Data Services will lease the properties under triple-net leases, with payment obligations guaranteed by Amazon.com. The initial lease term averages 12 years and annual rent increases are 2.75%. In this structure, the landlord owns the property interest and receives lease payments; the tenant runs the computing business.³

BXDC defines its 7% initial cap rate as first-year base rent divided by aggregate gross value, including transaction costs. That definition belongs beside the headline yield when comparing another offering. Investor returns will also reflect financing, ownership expenses and the eventual sale.⁴

The rent schedule does not automatically rise when Nvidia reports a strong quarter or fall when the chip market disappoints. For the landlord, the immediate income questions are who owes the rent, what guarantees that obligation and what spending the lease leaves with the owner.

Technology can still affect the property’s long-term value. Power and cooling requirements influence adaptation costs and the tenant’s decision to renew. Nvidia’s own infrastructure work includes ways to adapt existing facilities to denser computing systems. The useful distinction is between contractual rent and the future usefulness of the building.⁵

What I would underwrite at 7 percent

Our Q3 entry came from a listed colocation proxy based on Equinix and Digital Realty. Blackstone is acquiring powered shells under a specific lease structure. Their growth rights, expenses and capital obligations differ. The gap between 5.74% and 7% therefore does not establish a market-wide price decline.

I would give a stabilized, credit-backed 7% opportunity real diligence time. The next step is to replace the sector proxy with the offering’s actual economics.

How much of the quoted yield reaches the owner? Reconcile the income definition and all-in purchase cost. Include landlord capital obligations, reserves and transaction costs once, on a consistent basis. BXDC’s denominator already includes transaction costs; our sector calculation retains its own cost allowance and cannot simply be relabeled as the deal’s return.

What income can the owner reasonably expect during the hold? Start with the lease, guarantee, rent increases and termination rights. Contractual growth provides a different basis for underwriting from a forecast that assumes the landlord captures the full growth of AI spending.

What will the next buyer be purchasing? A five-year sale during a 12-year lease leaves a different remaining term from the one purchased today. Test the value of that remaining rent and what the property could require at renewal. A lower exit cap should be supported by the asset and the cash flow the next buyer receives.

Seven percent changes the starting position enough to warrant that work. The findings then determine whether the specific investment deserves capital.

The other side of a higher entry yield

The same repricing that improves a new buyer’s prospective return can put pressure on an existing owner’s valuation.

Consider a hypothetical property earning $1 million of annual net operating income. At a 4% cap rate, it is worth $25 million. At 7%, the same income supports approximately $14.3 million, a 43% decline. Income would need to grow 75% to preserve the original value at the higher cap rate.

That is the arithmetic of a comparable asset repricing, not evidence that existing data-center portfolios have lost 43%. Income growth, lease terms and capital requirements differ. It does show why a successful operating business can coexist with a difficult investment for someone who paid too much for its income.⁶

I am interested in whether the next offerings provide the same combination of starting yield, tenant credit and contractual income. The Blackstone announcement supplies a real transaction to investigate; repeated comparable transactions would tell us more about what allocators can actually buy.

At a 7% entry cap, our existing base case gives us a favorable answer. If that is the type of opportunity sponsors are bringing to market, capital allocators should sharpen their pencils. Our Q3 research gives us a reason to pursue the underwriting.


The Allocation Desk: The complete Q3 Cross-Asset Report shows the entry assumptions, sector comparisons and evidence that would change our views across ten real estate sectors. It is available through The Allocation Desk.

Custom HTML/CSS/JavaScript

Sources and calculation notes

1. BXDC transaction announcement, November 16, 2026. The agreement was announced that day and remained subject to closing conditions.

2. EM Capital, Q3 2026 Cross-Asset Report, September 30 research snapshot, 82-page edition, pp. 52–54, 68–74. The original 5.5%, 8.3% and −2.8-point figures are published rounded values. The entry source is a listed colocation proxy. The 7% scenario holds the report’s published 0.44-point recurring-capital deduction, 1.60-point growth contribution and 0.70-point annual transaction-cost allowance fixed. The exit contribution is the fifth root of entry cap divided by exit cap, less one, expressed in percentage points. This additive annualized comparison is not a dated cash-flow IRR.

At a 7% entry and 5.96% exit, the exit contribution is 3.2691 points. Adding 7.00, subtracting 0.44, adding 1.60, adding 3.2691 and subtracting 0.70 gives 10.7291%, rounded to 10.7%. With a 7% exit the contribution is zero and the result is 7.46%, rounded to 7.5%. Re-expressing recurring capital as the appendix’s 7.7% of NOI at the new entry yield instead gives about 10.6%; the favorable base-case conclusion is unchanged.

3. BXDC Form 8-K, Item 1.01, filed November 16, 2026, reporting the November 11 agreement. Source for ownership, guarantee, initial lease term and rent increases.

4. BXDC Transaction Summary, November 2026, slides 2 and 6. Initial cap rate uses first-year base rent over aggregate gross value including transaction costs. The 32-year fully extended term includes two optional ten-year renewals; it is not the initial committed term.

5. Nvidia power architecture discussion, August 11, 2026. Cited for infrastructure requirements and potential adaptation paths, not for the hardware installed at these campuses.

6. The unchanged-income illustration uses value equal to annual NOI divided by the cap rate. $1 million divided by 4% is $25 million; divided by 7% is $14.29 million. The value change is 4 divided by 7, less one. Required income growth to preserve value is 7 divided by 4, less one.

Research cutoff November 16, 2026. This edition is educational and informational, not investment advice, a security recommendation or an offer. Returns are hypothetical model estimates, not realized performance or guarantees. The sector comparison includes stated recurring capital and modeled transaction costs, before financing, investor-level fees, carried interest and taxes. It does not establish suitability or replace underwriting of an actual offering.

blog author image

DanieL Erb

Founder & CEO | Owner & Operator of U.S. Residential Real Estate | Alternative Investment Intelligence | Author of The Allocation Memo

Back to Blog