NVDA Drops 5.27% as OpenAI Revenue Shortfall Cracks the AI Capex Thesis

The Catalyst

The move started with a Wall Street Journal report that OpenAI missed internal goals for both weekly active users and monthly revenue. The company was shooting for 1 billion users by the end of 2025, a target it has yet to hit. More damaging than the user miss: CFO Sarah Friar reportedly flagged internal concern that OpenAI could not pay for future computing contracts if revenue doesn't grow fast enough.

NVDA gapped down 3.4% in premarket to $209.28 on April 28, then continued grinding lower through April 29 and into April 30 — the 14-hour leg the Hyperliquid perp captured. OpenAI publicly pushed back, with Sam Altman and Friar stating they remain "totally aligned on buying as much compute as we can." The market did not buy the reassurance.

The $1 Trillion Question

This is not really about whether OpenAI is healthy — it is about whether the unit economics behind the AI buildout actually close. In 2025 alone, OpenAI inked over $1 trillion in compute deals:

• ~$500B with NVIDIA
• ~$300B with Oracle
• ~$270B with AMD

Those are the obligations Friar is reportedly worried about. If OpenAI's revenue ramp slips meaningfully versus plan, the customer that anchors a non-trivial slice of NVIDIA's forward demand suddenly looks shakier — and every other hyperscaler has to ask whether their own AI revenue assumptions are pricing in the same kind of optimism. NVIDIA's quarter-to-quarter prints are still strong (revenue up 73% year-over-year), so this is not a fundamentals-have-cracked story. It is a multiples-recompress story.

Why the Whole Sector Sold Off

The price action makes more sense once you look at the breadth. NVDA was the headline name, but the Philadelphia Semiconductor Index dropped 3.6%, with AMD off 3.1%, Broadcom off 3.29%, and Arm Holdings down roughly 6.7%. Micron, Intel, and Applied Materials joined the move.

Notably, this happened despite strong capex guidance from Meta on April 29 and ahead of an earnings stretch where Alphabet, Microsoft, and Amazon were each expected to defend more than $700 billion in collective AI infrastructure commitments for the year. The market is telling you it is willing to look past one bullish hyperscaler print if the demand-side anchor — OpenAI — looks even slightly wobbly. That is what re-rated NVDA, not anything specific to NVIDIA's own quarter.

What to Watch

Three things to track from here. First, whether any of OpenAI's mega-deal counterparties — particularly Oracle and AMD, who carry the highest implied take-or-pay risk — get pressed on revenue recognition assumptions on their own calls. Second, whether NVIDIA management addresses concentration risk in its next print; the $500B figure is now in the discourse and won't leave it. Third, whether the SOX can hold the level it bounced from late on April 30, or whether this becomes the kind of multi-week derate that turns a single WSJ scoop into the inflection point bears have been waiting for. The bull case isn't broken. The bull case's biggest customer is.