“As a result, based on my premium data center model, many data center projects are unprofitable even with a paying customer, and that’s assuming they even get built.”
Our reading:
Many AI data center projects are unprofitable even with a paying customer.
Assessment: Unresolved
What happened
Was it true for the period described?
Outcome: Unresolved
We checked, and the evidence so far doesn’t settle it.
Judged on a plain reading. They named no measure, so we used the reading a typical informed reader would take. It is spelled out under “How we judged it”.
Evidence up to . Checked .
Why
CoreWeave reported a 6% adjusted operating margin and a $452 million net loss for Q4 2025; these corporate figures do not establish individual projects' full-cost returns [e3]. Principal's year-end 2025 update reports positive projected returns for a powered-shell development whose tenant is responsible for fit-out [e4]. Neither disclosure establishes how many AI data center projects contemplated in March 2026 would lose money despite having paying customers [e3, e4].
Caveats
- The disclosures provide neither a comparable project sample nor a reproducible reconstruction of the cited model [e3, e4].
- The reported conditions concern year-end 2025 rather than a March 2026 project sample [e3, e4].
- Projected powered-shell returns do not establish realized GPU-rental profitability [e4].
Evidence (2 items)
Context only: CoreWeave; transcript prepared by FactSet,
6%.
Why it matters: CoreWeave reports positive adjusted operating profit, while the same discussion explains that infrastructure costs can precede associated revenue. Neither the adjusted corporate margin nor this timing explanation measures full-life profits across individual projects.
- Publisher
- CoreWeave; transcript prepared by FactSet
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Measures
- CoreWeave, Adjusted operating income margin percent, non gaap, 2025-10-01 to 2025-12-31
- About events on
- Published
Context only: Principal Real Estate,
tenant delivery
Why it matters: The project update describes a contracted powered-shell development whose tenant handles fit-out, alongside construction financing. It does not establish occupied GPU-rental profitability and shows why different types of data center projects cannot be pooled without checking their costs and responsibilities.
- Publisher
- Principal Real Estate
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- About events on
- Published
How we judged it
- Kind of claim
- Observation
- Deadline
- None given. We don’t add one.
Conditions
- even with a paying customer (their words)
Where we had to interpret
- This describes the economics of projects contemplated at the time of the post, including projects not yet completed. It is not a dated forecast of future corporate earnings.
- I read many as a substantial, non-isolated set of projects, without inventing a majority threshold.
- Unprofitable means customer receipts fail to cover the relevant project costs, including hardware consumption or depreciation, construction, operations and financing.
- The target is the March 2026 project-economics snapshot. The model's investment lifetime is unspecified and is not replaced with a one-month profit calculation.
- Having a paying customer restricts the projects being assessed; it is not an antecedent of a conditional forecast.
Right if: Independent project-level records or a reproducible model grounded in actual contracts identify a substantial set of projects whose customer payments fail to cover full costs.
Wrong if: A sufficiently comprehensive, comparable project sample shows that losses under paying contracts are isolated rather than common, after including full project costs.
We wrote this test after the outcome was known.
What would settle it
A sufficiently comprehensive, comparable sample of projects contemplated in March 2026, or an independently reproducible model grounded in actual contracts, matching customer receipts and utilization to construction, hardware, operating and financing costs over a defined investment lifetime.
We found 54 claims in this post. See them all.
Full record
Source and dates
Where's Your Ed At, The Subprime AI Crisis Is Here, paragraph 48.
- Said
- We read the source
- live page, full text, public
- Evidence up to
- Checked
We read this wording on 2026-10-28. The post may have been edited after 2026-03-31.
Where it was said (1)
Said once in the posts we read.
: Where's Your Ed At, The Subprime AI Crisis Is Here
As a result, based on my premium data center model, many data center projects are unprofitable even with a paying customer, and that’s assuming they even get built.
- Where
- paragraph 48
- Published
- 2026-03-31
- We read it
- , live page
- Access
- full text, public
History
Versions
- Current: Unresolved. Evidence up to 2026-10-28.
What happened to this record
PublishedBy us
Published after both AI reviews and the final check passed.
Held backBy us
The final check failed: the write-up said more than the evidence showed.