Nvidia posted $96.2 billion in fiscal Q2 revenue, up 106% year-over-year, with data center sales alone hitting $89 billion — a scale gap that now defines who can and can’t compete in AI infrastructure.

Nvidia closed its fiscal second quarter with $96.2 billion in revenue, more than double the year-ago period, and data center sales climbed 117% to $89 billion. Edge computing, the company’s smaller but fastest-diversifying segment, added $7.2 billion on 27% growth. CEO Jensen Huang told investors the company will “sell twice as many chips next year” as this year, framing accelerating demand as a floor, not a ceiling. “AI has reached its inflection point,” Huang said. “It’s doing useful work. Its tokens are productive and profitable. Now, compute is revenue.”

Why the Data Center Number Shifts the Competitive Balance

Nvidia’s $89 billion in quarterly data center revenue isn’t just a strong result, it’s larger than the annual revenue of most standalone semiconductor companies. AMD posted $11.5 billion in total Q2 revenue, with data center sales of $6.7 billion, up 107%, and Broadcom’s custom AI silicon revenue hit $16.7 billion, up 221%. Both numbers represent real acceleration, and both remain a fraction of what Nvidia now generates in data center sales alone in a single quarter. Every hyperscaler buildout announced this year, from Microsoft to Amazon to Meta, still routes the majority of its compute dollars through Nvidia silicon, regardless of how aggressively rivals price custom accelerators. Custom silicon programs at Amazon, Google, and Microsoft were supposed to erode that dependency by now; instead, Nvidia’s data center growth rate accelerated rather than slowed, evidence that in-house chips are supplementing Nvidia purchases, not replacing them.

What AMD and Broadcom’s Growth Actually Proves

The fact that AMD’s data center revenue doubled and Broadcom’s AI revenue nearly tripled isn’t a sign Nvidia is losing ground, it’s a sign the total addressable market is expanding faster than any single competitor can capture share from Nvidia’s base. With Google, Microsoft, Meta, and Amazon collectively guiding to roughly $725 billion in 2026 capital expenditure, up 77% from the prior year, there’s enough demand for multiple chipmakers to post triple-digit growth simultaneously without any of them closing the absolute gap with Nvidia. That’s the uncomfortable math for AMD and Broadcom: strong growth rates on a much smaller base still leave them years away from parity.

The Circular-Financing Question Nvidia Still Has to Answer

Skeptics have pushed back on the durability of this growth, pointing to the web of vendor financing arrangements between Nvidia and the hyperscalers and AI labs buying its chips. Huang has publicly rejected the “circular financing” characterization, but the concern isn’t going away: if AI capital expenditure decelerates even modestly, Nvidia’s growth rate has the furthest to fall of any company in the stack, simply because it has the most revenue exposed to the category. A single quarter of decelerating hyperscaler capex guidance would do more damage to Nvidia’s stock than to any of its customers, which is exactly the asymmetry critics are pricing into their skepticism.

What Intel’s Absence From This Conversation Says

Intel isn’t mentioned in the same breath as AMD and Broadcom when investors discuss AI accelerator competition, and that omission is itself the story. The company that once defined the semiconductor industry has yet to ship an AI accelerator at the volume or performance tier needed to appear in Nvidia’s rearview mirror, and every quarter that passes without a credible Intel entrant hands Nvidia, AMD, and Broadcom a three-way market with one fewer serious competitor than the historical chip landscape would suggest. For enterprise buyers building multi-year infrastructure roadmaps, that narrows the realistic vendor shortlist for frontier AI compute to a list that gets shorter, not longer, the more this quarter’s results sink in.

How This Changes the Playing Field

A 117% year-over-year data center growth rate at $89 billion in quarterly scale is a moat, not a milestone. Legacy chipmakers built their roadmaps around multi-year product cycles; Nvidia is now compounding revenue fast enough to fund its own next-generation architecture before competitors ship their current one. AMD’s data center GPU revenue remains a fraction of Nvidia’s single quarter, and Intel is still working to ship a credible AI accelerator at volume. Until a rival closes that scale gap, “competing with Nvidia” in data center AI infrastructure is now a game of chasing a moving, better-funded target, one that gets harder to catch every ninety days Nvidia posts numbers like these.