NVIDIA splits hyperscaler sales

Published May 27, 2026 by The Daily Scout

- Nvidia’s May 20 earnings and follow-up coverage showed the company breaking out hyperscaler sales, giving investors a clearer view of cloud concentration. - Jensen Huang said AI crossed a “critical threshold” because “tokens are now profitable,” as Nvidia, AMD, Qualcomm and Micron rose in early trade. - Nvidia’s next public milestones are its quarterly filings and future earnings materials, where investors can track any continued hyperscaler disclosure.

Why it matters

Nvidia’s latest reporting change matters because it gives investors a cleaner look at where the company’s AI demand is coming from. After its May 20 fiscal first-quarter 2027 results, coverage of the earnings call and materials pointed to a new distinction between hyperscaler sales and the rest of Nvidia’s business. That arrives as Chief Executive Jensen Huang argues the economics of AI are improving, saying token generation is now profitable. It also arrives as Nvidia reported record quarterly revenue of $81.6 billion, including $75.2 billion from data center, according to the company. ### Why does the hyperscaler split matter now? Nvidia’s May 20 earnings call included what CNBC described as a shake-up in the company’s reporting breakdown, with Huang addressing the reason for the change. Stratechery said the new framing separates hyperscaler sales — where Nvidia faces more commoditization pressure — from the rest of the market, where it still controls more of the stack. (investor.nvidia.com) That distinction matters because hyperscalers are not just customers; they are also designing more of their own infrastructure. Cloud companies including the biggest AI buyers have been building internal silicon programs and broader software stacks, even as they continue buying Nvidia systems at scale. The split gives investors a way to watch whether growth is broadening beyond a small group of cloud platforms. That last point is an inference from the reporting change and the competitive context described by Stratechery and CNBC. (cnbc.com) ### What exactly did Jensen Huang say about AI economics? Benzinga reported on May 26 that Huang said AI had crossed a “critical threshold” and that “tokens are now profitable.” CNBC separately quoted Huang saying demand had “gone parabolic” and tied that to the arrival of agentic AI. Those comments are notable because they shift the argument from training large models to earning money from using them. (cnbc.com) If token generation is profitable, that supports continued spending on inference infrastructure, not just headline model development. That interpretation follows from Huang’s own framing about profitability and demand. ### How did markets react? (benzinga.com) Charles Schwab’s market update said Nvidia rose more than 1% in early trade after the report, while AMD gained nearly 3% and Qualcomm rose more than 2%; Micron also got a boost, Schwab said. The move suggested investors treated Nvidia’s commentary as supportive for the broader AI chip complex, not only for Nvidia itself. Nvidia’s own quarter gave investors fresh numbers to anchor that reaction. (benzinga.com) The company said first-quarter revenue rose 85% from a year earlier to $81.6 billion, while data center revenue rose 92% to $75.2 billion. Nvidia also announced an additional $80 billion share repurchase authorization and a higher quarterly cash dividend, according to its financial release. ### What does this change for customers and suppliers? (schwab.com) Large cloud buyers have been pushing the market toward more architecture choices, including merchant GPUs, custom accelerators and mixed systems. A reporting split that isolates hyperscalers may intensify scrutiny over how much of Nvidia’s growth depends on a concentrated set of cloud customers and how much comes from enterprises, sovereign projects and other buyers. That is an inference based on the new reporting approach described by Stratechery. (investor.nvidia.com) For companies selling into the AI stack, that can translate into more work around trade-off analysis, subsystem integration and migration support. Customers weighing Nvidia against in-house chips or alternative architectures still have to connect hardware, networking and software into deployable systems. Nvidia’s future quarterly materials and SEC filings will show whether the company keeps expanding that disclosure and how investors use it. (investor.nvidia.com) (stratechery.com)

Key numbers

Nvidia’s May 20 earnings and follow-up coverage showed the company breaking out hyperscaler sales, giving investors a clearer view of cloud concentration.
After its May 20 fiscal first-quarter 2027 results, coverage of the earnings call and materials pointed to a new distinction between hyperscaler sales and the rest of Nvidia’s business.
It also arrives as Nvidia reported record quarterly revenue of $81.6 billion, including $75.2 billion from data center, according to the company.
Nvidia’s May 20 earnings call included what CNBC described as a shake-up in the company’s reporting breakdown, with Huang addressing the reason for the change.

What happens next

After its May 20 fiscal first-quarter 2027 results, coverage of the earnings call and materials pointed to a new distinction between hyperscaler sales and the rest of Nvidia’s business.
Nvidia’s May 20 earnings call included what CNBC described as a shake-up in the company’s reporting breakdown, with Huang addressing the reason for the change.
Benzinga reported on May 26 that Huang said AI had crossed a “critical threshold” and that “tokens are now profitable.” CNBC separately quoted Huang saying demand had “gone parabolic” and tied that to the arrival of agentic AI.

Sources

Quick answers

What happened in NVIDIA splits hyperscaler sales?

Nvidia’s May 20 earnings and follow-up coverage showed the company breaking out hyperscaler sales, giving investors a clearer view of cloud concentration. Jensen Huang said AI crossed a “critical threshold” because “tokens are now profitable,” as Nvidia, AMD, Qualcomm and Micron rose in early trade. Nvidia’s next public milestones are its quarterly filings and future earnings materials, where investors can track any continued hyperscaler disclosure.

Why does NVIDIA splits hyperscaler sales matter?

Nvidia’s latest reporting change matters because it gives investors a cleaner look at where the company’s AI demand is coming from. After its May 20 fiscal first-quarter 2027 results, coverage of the earnings call and materials pointed to a new distinction between hyperscaler sales and the rest of Nvidia’s business. That arrives as Chief Executive Jensen Huang argues the economics of AI are improving, saying token generation is now profitable. It also arrives as Nvidia reported record quarterly revenue of $81.6 billion, including $75.2 billion from data center, according to the company. Why does the hyperscaler split matter now? Nvidia’s May 20 earnings call included what CNBC described as a shake-up in the company’s reporting breakdown, with Huang addressing the reason for the change. Stratechery said the new framing separates hyperscaler sales — where Nvidia faces more commoditization pressure — from the rest of the market, where it still controls more of the stack. (investor.nvidia.com) That distinction matters because hyperscalers are not just customers; they are also designing more of their own infrastructure. Cloud companies including the biggest AI buyers have been building internal silicon programs and broader software stacks, even as they continue buying Nvidia systems at scale. The split gives investors a way to watch whether growth is broadening beyond a small group of cloud platforms. That last point is an inference from the reporting change and the competitive context described by Stratechery and CNBC. (cnbc.com) What exactly did Jensen Huang say about AI economics? Benzinga reported on May 26 that Huang said AI had crossed a “critical threshold” and that “tokens are now profitable.” CNBC separately quoted Huang saying demand had “gone parabolic” and tied that to the arrival of agentic AI. Those comments are notable because they shift the argument from training large models to earning money from using them. (cnbc.com) If token generation is profitable, that supports continued spending on inference infrastructure, not just headline model development. That interpretation follows from Huang’s own framing about profitability and demand. How did markets react? (benzinga.com) Charles Schwab’s market update said Nvidia rose more than 1% in early trade after the report, while AMD gained nearly 3% and Qualcomm rose more than 2%; Micron also got a boost, Schwab said. The move suggested investors treated Nvidia’s commentary as supportive for the broader AI chip complex, not only for Nvidia itself. Nvidia’s own quarter gave investors fresh numbers to anchor that reaction. (benzinga.com) The company said first-quarter revenue rose 85% from a year earlier to $81.6 billion, while data center revenue rose 92% to $75.2 billion. Nvidia also announced an additional $80 billion share repurchase authorization and a higher quarterly cash dividend, according to its financial release. What does this change for customers and suppliers? (schwab.com) Large cloud buyers have been pushing the market toward more architecture choices, including merchant GPUs, custom accelerators and mixed systems. A reporting split that isolates hyperscalers may intensify scrutiny over how much of Nvidia’s growth depends on a concentrated set of cloud customers and how much comes from enterprises, sovereign projects and other buyers. That is an inference based on the new reporting approach described by Stratechery. (investor.nvidia.com) For companies selling into the AI stack, that can translate into more work around trade-off analysis, subsystem integration and migration support. Customers weighing Nvidia against in-house chips or alternative architectures still have to connect hardware, networking and software into deployable systems. Nvidia’s future quarterly materials and SEC filings will show whether the company keeps expanding that disclosure and how investors use it. (investor.nvidia.com) (stratechery.com)