NVIDIA: AI tokens now profitable

Why it matters

Nvidia’s latest earnings call offered investors a more specific defense of the AI trade than another large revenue beat. On May 20, the chipmaker reported first-quarter fiscal 2027 revenue of $81.6 billion, up 85% from a year earlier, and data-center revenue of $75.2 billion, up 92%. (investor.nvidia.com) Jensen Huang used the call to argue that the economics of AI inference are improving, not just the scale of spending. Huang said agentic AI had arrived and that token generation — the process of producing output from AI models after training — “is now profitable,” according to the earnings-call transcript and call coverage. (fool.com) That matters because Nvidia’s stock reaction suggested investors were still asking whether demand is durable at current valuations. (thestandard.com.hk) CNBC reported that the quarter was strong but the shares slid after the report, and a separate CNBC item on May 26 said traders were focused on whether the stock could hold a key technical level after a volatile stretch. ### What exactly did Huang say about AI economics? Huang told investors on the May 20 call that AI had crossed a “critical threshold” and that token generation was now profitable, according to the transcript cited by Benzinga and other call reports. (investor.nvidia.com) In practical terms, token generation is the revenue-producing side of generative AI. Training builds a model; inference serves queries and generates text, images or code. (fool.com) Huang’s point was that customers are no longer buying Nvidia systems only to experiment or build capacity in advance — they can now earn money from running models at scale, based on his description of the market on the call. ### How big was the quarter beneath that claim? Nvidia reported $81.6 billion in quarterly revenue and said second-quarter revenue would be about $91 billion, plus or minus 2%. (cnbc.com) The company also announced an additional $80 billion share repurchase authorization and raised its quarterly cash dividend to $0.25 per share from $0.01. (fool.com) Colette Kress gave investors a more detailed look at who is driving the demand. CNBC reported that hyperscalers accounted for more than half of Nvidia’s data-center revenue in the quarter, reaching $38 billion and rising 12% sequentially. ### Why are investors focused on hyperscalers now? Stratechery reported that Nvidia is changing its reporting to separate hyperscaler sales from the rest of the business. (fool.com) The publication said the distinction matters because Nvidia faces more commoditization pressure with the largest cloud customers than in parts of the market where it can sell a broader hardware-and-software stack. That split gives investors another way to test Huang’s argument. If hyperscaler spending remains large but growth broadens outside the biggest cloud companies, Nvidia can point to demand from enterprise, sovereign AI and AI-native cloud customers as support for its platform strategy; Huang cited those segments on the call. (investor.nvidia.com) ### Why didn’t the stock simply rally on a beat? (cnbc.com) Nvidia’s shares fell in extended trading after the results even as revenue and guidance topped Wall Street expectations, Reuters and CNBC reported. Reuters said Huang used the call to reassure investors that Nvidia could sustain growth through a broad customer base and new products as competition changes. The stock response showed that investors were weighing more than the quarter’s headline numbers. (stratechery.com) The open question after the call was whether Huang’s claim about profitable token generation would ease concerns that AI infrastructure spending has run ahead of near-term returns. CNBC’s May 26 follow-up framed the next test in market terms: whether the shares could stabilize after the post-earnings wobble. ### What comes next for this story? Nvidia’s next formal update is its second-quarter report, anchored by guidance for roughly $91 billion in revenue. Investors will also be watching the company’s revised disclosures on hyperscaler sales and whether management continues to emphasize inference economics, customer breadth and demand for the full Nvidia stack. (thestandard.com.hk) (link.cnbc.com) (wifc.com)

Key numbers

• Nvidia said on May 20 that first-quarter revenue rose 85% to $81.6 billion as CEO Jensen Huang told investors AI demand had crossed a profitability threshold.
• (investor.nvidia.com) Huang said “token generation is now profitable,” while CFO Colette Kress said hyperscalers contributed $38 billion, or more than half, of data-center revenue.
• (fool.com) Nvidia’s next checkpoint is second-quarter revenue guidance of about $91 billion, with investors also watching the company’s revised hyperscaler disclosures.
• On May 20, the chipmaker reported first-quarter fiscal 2027 revenue of $81.6 billion, up 85% from a year earlier, and data-center revenue of $75.2 billion, up 92%.

What happens next

• On May 20, the chipmaker reported first-quarter fiscal 2027 revenue of $81.6 billion, up 85% from a year earlier, and data-center revenue of $75.2 billion, up 92%.
• (thestandard.com.hk) CNBC reported that the quarter was strong but the shares slid after the report, and a separate CNBC item on May 26 said traders were focused on whether the stock could hold a key technical level after a volatile stretch.
• Huang told investors on the May 20 call that AI had crossed a “critical threshold” and that token generation was now profitable, according to the transcript cited by Benzinga and other call reports.

Sources

• investor.nvidia.com
• fool.com
• thestandard.com.hk
• cnbc.com
• stratechery.com
• link.cnbc.com
• wifc.com

Quick answers