Nvidia’s $81.6B Quarter Meets a Math Problem: Hyperscaler Capex Outpaces Revenue by 4-to-1

Nvidia's Q1 revenue hit $81.6B, but stock is down 11% amid concerns over hyperscaler ROI, falling GPU rental rates, and China export controls.

The numbers coming out of Nvidia are staggering by any historical standard. Revenue surged 85% to $81.6 billion in the first fiscal quarter of 2027, with data-center sales hitting $75.2 billion — well above the $73.5 billion analysts had penciled in. The company guided $91 billion for the current quarter and authorized an $80 billion share buyback while raising the dividend to $0.25 a share. Yet the stock trades around €171, roughly 16% below its 52-week high of €202.50, and has shed nearly 11% over the past 30 days.

The disconnect stems from a deepening tension between Nvidia’s blistering execution and the sustainability of the AI investment cycle. The four hyperscalers — Amazon, Microsoft, Alphabet and Meta — are set to plow roughly $725 billion into infrastructure in 2026, a 77% jump from the prior year. Their combined revenue, however, is only growing at 15% to 16%. That 4-to-1 gap between capex growth and top-line expansion is the kind of math that keeps Wall Street strategists up at night.

D.A. Davidson analyst Gil Luria put it bluntly: Nvidia and Micron are trading as if the AI cycle has already peaked. He’s not alone in worrying that when hyperscaler returns on investment fail to keep pace, the spending spree will inevitably slow — and Nvidia’s pricing power will take the first hit.

A real-time signal emerged in late May. The hourly rental rate for Nvidia’s B200 GPU fell from $6.11 to $4.22 between May 30 and June 21 — a 31% drop in just three weeks. Supply of AI compute capacity is outstripping the ramp-up of new workloads, eroding the leverage Nvidia once had. For a company whose data-center segment accounts for the vast majority of revenue, that’s a warning flag that theoretical demand forecasts can’t dismiss.

Meanwhile, the trade winds have shifted against Nvidia in Washington. Early June saw the Commerce Department close a loophole that had let Chinese AI firms buy Nvidia’s Rubin and Blackwell chips through offshore subsidiaries. JPMorgan and Bernstein now peg the cumulative revenue hit from export controls and domestic Chinese competition at $5.5 billion to $16 billion this fiscal year. China’s contribution to Nvidia’s data-center business is effectively zero.

The irony is that Nvidia’s own AI boom is lifting competitors more than itself. Micron’s stock has nearly tripled this year, AMD has roughly doubled, and the iShares Semiconductor ETF has more than doubled — yet Nvidia has lagged. Investors are rotating into memory and CPU plays as the next phase of infrastructure build-out takes shape, leaving the chipmaker that ignited the cycle watching from the sidelines.

The bull case still commands a majority of analyst opinion. The median price target sits at $309 — more than 60% above current levels — and 58 of 61 analysts rate the stock a Buy or Outperform, with only one Sell. Firms including Bank of America, JPMorgan, Morgan Stanley and Wedbush have lifted their targets to the $250–$300 range. The bull camp points to Nvidia’s next-generation Vera Rubin chip, slated for production in the second half of 2026, with AWS, Google Cloud, Microsoft Azure and CoreWeave already lined up as early adopters. A multiyear technology partnership with SK hynix to develop memory solutions for AI supercomputers, AI PCs and robotics platforms adds further momentum.

But the valuation leaves little room for error. With an RSI of 40.9 and the stock hovering roughly 6% below its 50-day moving average, technical weakness is already baked in. The next catalyst — and the one most likely to settle the bull-bear debate — will be the upcoming round of hyperscaler quarterly results. If their capital expenditure growth begins to moderate, Nvidia’s record revenue run may start to feel like a peak rather than a plateau.