Q1 Earnings: Already a Record Quarter, and Vera Steals the Spotlight
Nvidia reported Q1 revenue of US$81.62 billion, crushing analyst estimates of US$78.86 billion. Q2 guidance came in at US$91 billion — well above the US$86.84 billion Wall Street expected. By any measure, another dominant quarter. Yet the headline number masked something more strategically important: Jensen Huang used the analyst call to plant a flag on an entirely new battlefield.
Nvidia's new Vera central processors target a US$200 billion market — one that sits entirely outside the US$1 trillion the company has already forecast from its Blackwell and Rubin AI GPU lineup between 2025 and 2027. Huang told analysts he expects Vera chip revenue to hit US$20 billion by the end of this fiscal year, calling it "the second largest" sales contributor. That's not a footnote. That's a second front.
Why Nvidia Needs a Second Front
The reason is surprisingly simple: Nvidia's biggest customers — Google, Amazon, and Microsoft — are building their own chips.
Collectively, these hyperscalers are expected to pour over US$700 billion into AI infrastructure this year, up sharply from roughly US$400 billion in 2025. But the deeper threat isn't the spending — it's the silicon. Google has its TPU line. Amazon has Trainium. Microsoft has Maia. These aren't experiments; they're strategic bets that custom chips can run inference workloads cheaper and faster than Nvidia's GPUs.
The industry narrative has quietly shifted: who can train the biggest model is no longer the only question. Who can serve it cheapest and fastest is equally important. And on that second question, Nvidia's GPU dominance is most exposed.
Training still belongs to Nvidia — the company's CUDA ecosystem and HBM memory integration create switching costs that custom silicon can't easily overcome. But inference, generating answers at scale, in real time, across millions of users? That's where Google's TPUs, Amazon's Trainium, and AMD's data-center CPUs are making real inroads.
Meet Vera: Nvidia's Answer to the Inference Challenge
Nvidia's response is the Vera chip, a central processor purpose-built for inference workloads. The chip was developed partly using technology licensed from Groq, a startup specializing in fast inference that Nvidia reportedly acquired for around US$17 billion. The full Vera Rubin platform — pairing the Vera CPU with Rubin GPUs — is set to launch later this year.
The strategic logic is elegant. If customers won't use Nvidia GPUs for inference, Nvidia will supply the CPUs they do use. By owning the inference compute stack, Nvidia preserves pricing power even as custom silicon nibbles at the GPU monopoly.
Huang was characteristically blunt about the biggest risk: supply.
"My sense is that we'll be supply-constrained through the entire life of Vera Rubin," he told analysts. It's a telling admission for a product Nvidia is positioning as a major growth pillar. To get ahead of potential bottlenecks, Nvidia disclosed that its supply commitments rose to US$119 billion in Q1, up from US$95.2 billion the previous quarter — a sharp jump that reflects both confidence in demand and concern about a global memory chip crunch.
Investors Are Asking the Right Question
Despite the beat, Nvidia shares fell 1.6% in after-hours trading. The market's reaction tells a story: investors have priced in the quarter. What they want to know is durability.
eMarketer analyst Jacob Bourne captured it well: *"Nvidia delivered another beat, but at this point that's essentially priced in. The lingering question is whether it can convince investors the AI buildout has durability into 2027 and 2028, especially as the narrative shifts toward inference workloads and competing silicon from Google, Amazon, AMD, and Intel."*
Huang pushed back with a revealing data point. He pointed to a growing sub-segment of AI-native cloud customers whose spending is now roughly equal to the hyperscalers — but growing faster quarter-over-quarter. *"We should be growing faster than hyperscale capex,"* he said. The Vera chip is central to that argument.
Nvidia also announced an US$80 billion share repurchase program and raised its quarterly dividend from 1 cent to 25 cents per share. These are moves that signal financial confidence, even as Huang warned of tightening supply.
What This Means for the AI Infrastructure Race
The Vera story is ultimately a story about the changing shape of AI compute. For years, the winning strategy was: build the best GPU, sell as many as possible. Nvidia executed that perfectly, and the Blackwell cycle proved it once again.
But the inference pivot changes the game in two ways:
Whether Vera can sustain US$20 billion in annual revenue — and whether the supply chain cooperates — will define the next chapter of Nvidia's growth story. For now, the company has earned the benefit of the doubt. But the second front is contested territory.
Key Takeaways
- Nvidia Vera chip targets a US$200 billion inference market
- Expected revenue: US$20 billion by end of fiscal year
- Built partly on Groq's inference technology (deal ~US$17 billion)
- Vera Rubin platform (CPU + GPU) launching later this year
- Supply commitments jumped to US$119 billion in Q1
- Investors are watching whether the AI buildout has durability into 2027-2028
*Source: AI News — Nvidia's Vera chip is the US$200 billion bet Jensen Huang doesn't want you to overlook (Dashveenjit Kaur, May 21, 2026)*