The AI funding split resolved, and both names won for opposite reasons

For two weeks we framed AI capex as a funding question, not a demand question. On Tuesday both test cases reported after the close, and the results are a clean lesson in how the market prices a self-funder versus a financed builder.

The self-funder missed revenue and rose anyway. Super Micro reported revenue of $11.1B, below the ~$11.55B consensus, with adjusted EPS of $1.70 against ~$1.59 expected (Seeking Alpha, MarketBeat). The stock rose about 8% in extended trading, and the reason was entirely forward: FY27 net sales guided to $65-72B against ~$52.5B expected, Q1 FY27 sales $14.5-15.5B versus ~$11.68B, Q1 FY27 EPS $0.89-0.98 versus ~$0.76. A company that funds its growth from product margin gets to trade on the size of the order book, because it does not need anyone's permission to fill it. The printed quarter was almost irrelevant.

The financed builder's loss narrowed, and that was the whole story. CoreWeave reported revenue of $2.58B, up 112% year on year and just ahead of the ~$2.56B consensus, but the number that moved the stock was the adjusted loss of -$1.03 per share against -$1.20 expected (CNBC). It signed new business with Anthropic and Meta and management called an operating-leverage inflection (Invezz). The stock rose about 13%. For a business that buys GPUs and data-center capacity with debt and equity, revenue growth was never in doubt. What was in doubt was whether each dollar of that revenue loses less over time. On this print, it did.

Why it matters. Same theme, two economics, both rewarded, but on different variables. The self-funder is underwritten on demand durability and margin. The financed name is underwritten on the slope of its losses and its cost of capital. If you had graded both on the revenue line, you would have been wrong twice: SMCI missed it and CoreWeave's beat was a footnote.

So what for the allocator (ABC). This is Alpha-sleeve work, and the two names demand different diligence. On a self-funder, the question is whether the guide converts to cash without a trip to the capital markets; a $65-72B revenue guide is only as good as the balance sheet that carries the working capital behind it. On a financed builder, the question is whether the loss keeps narrowing per dollar of revenue across quarters, because that trajectory, not the growth rate, is what refinances the debt. Neither belongs in the Beta sleeve, where you own the index and skip the single-name funding risk entirely.

Takeaway: in a capex boom, revenue growth is the easy part and tells you almost nothing; who pays for the growth, and whether their economics are improving, is the entire investable question.

Analytics & education, not advice. DYOR.