Nvidia's Multiple Has Collapsed Back to Pre-Boom Levels Despite Record Sales

Nvidia's sinking stock valuation is flashing a warning signal about sustaining booming profit growth, Bloomberg reported on September 22, 2026. Bloomberg
The warning came weeks after Nvidia reported revenue of $96.2 billion for the second quarter ended July 26, 2026, up 106% from a year ago. Nvidia Data Center revenue was $89.0 billion, up 117% from a year ago. The company, which identifies itself as the pioneer of GPU-accelerated computing, specializes in products and platforms for gaming, professional visualization, data center, and automotive markets.
Forward multiples have compressed even as the numerator in the growth equation kept expanding. Nvidia lost roughly $1 trillion in market value in less than two months before July 8, 2026, and the stock was the cheapest it had been since before the AI boom. Bloomberg That July reset took the forward P/E back toward levels last seen in June 2019, when Nvidia traded at 22.47 times forward earnings. Yahoo Finance
The de-rating did not start in July. Nvidia's P/E sank to a seven-year low as of March 30, 2026. Reuters MarketWatch's quote page for NVDA references a historic $150 billion buyback announcement by Nvidia. MarketWatch A separate MarketWatch report, undated and therefore lower priority than the 2026 prints, had put the stock at 35.99 times 12-month forward earnings and described it as relatively inexpensive after a large rally.
Sell-side framing has leaned against the tape. In August 2026, BofA said Nvidia shares could be trading at a discount of as much as 50% on AI risks. Bloomberg On August 24, 2026, MarketWatch published a piece titled It's time to bet big on Nvidia's stock, says this analyst who thinks the market has it all wrong. MarketWatch Earlier MarketWatch pieces from December 2025 made a similar mean-reversion case, arguing the stock was unusually cheap and almost historically cheap, with history suggesting scope for gains ahead.
The broader context here is a market pricing lower earnings durability, not lower current earnings. Multiples lead earnings. When a forward multiple contracts through triple-digit revenue growth, the market is discounting terminal growth, incremental margins, or capex intensity. That is de-rating in its textbook form.
In my view, practitioners should read the sequence, not any single print, for what it implies about the cost of equity and capital return. A seven-year low in March, a pre-AI-boom multiple in July, a 50% risk discount frame in August, and a September warning on profit sustainability trace a consistent repricing of AI capex cyclicality. The $150 billion authorization and the $89.0 billion Data Center quarter sit on opposite sides of that repricing. One supports EPS through share count reduction. The other has to be repeated to justify the prior multiple. For underwriting, the relevant question is persistence of Data Center growth and its conversion to free cash flow, not whether the multiple looks low against trailing fundamentals.


