NVDA Q2 Guidance at $91B Tests Whether Hyperscaler Capex Can Hold

Nvidia delivered one of the strongest earnings in corporate history on May 21, reporting revenue of $78.75 billion in Q1, a stunning 85% year-on-year increase from the prior year's $44.06 billion. Net income nearly doubled, growing 135% to $1.76 per share from $0.96. The company's data-center segment, the crown jewel of AI infrastructure demand, generated $75.2 billion in revenue, more than double the prior-year level. Yet despite these superlative numbers, the stock fell 2.5% in after-hours trading, a lackluster response that revealed an uncomfortable truth: the market's concern is no longer whether Nvidia can deliver growth, but whether the world's hyperscalers can afford to keep spending.

Nvidia's forward guidanceCompany-issued forecasts of future financial performance. of $91 billion in Q2 revenue, implying 94% growth, underscores the magnitude of AI capex still flowing into the company's servers and GPUs. The company also authorized an $80 billion stock buybackA company repurchasing its own shares from the open market. and reported gross margins at all-time highs, signaling pricing power and operational excellence. Yet the market's muted reaction reflects a growing debate: are hyperscalers like Google, Microsoft, and Amazon willing or able to sustain multi-hundred-billion-dollar annual capex investments when bond yields have risen sharply, 10-year Treasuries approach multi-year highs, and the Fed is now pricing a 37% odds of a rate hike in 2026?

The earnings miss wasn't in the numbers; it was in positioning. Sell-side analysts were uniformly bullish ahead of the report, and retail options flow skewed heavily bullish. With such lopsided expectations, even a beat with no upside surprise disappointed. The real question now is whether rising borrowing costs will force hyperscalers to reevaluate capex budgets later this year, a risk that will weigh on semiconductor demand in 2026.

Some observers also pointed to a structural concern: Nvidia's H100 GPUs, launched in 2022, are now three generations old, yet rental prices have risen roughly 20% in 2026. This suggests chip utilization is constrained and supply remains tight, but it also hints that older hardware commands premium pricing precisely because new chips are in short supply. If capacity comes online or demand slows, pricing power could erode quickly.