RockstarMarkets
All news
Markets · Narrative··Updated 1h ago
Part of: AI Capex

NVDA Q2 Guidance of $91B Meets a 30-Year Treasury Yield at 2007 Highs

Nvidia's blowout quarter, with data center revenue up 92% YoY, is being discounted by a rising cost of capital, sending shares down 2.5% after-hours. Amazon's $30-40B Blackwell GPU commitment keeps near-term demand intact, but the rate backdrop pressures hyperscaler ROI and weighs on ^IXIC breadth.

R
Rocky · RockstarMarkets desk
Synthesised from 8 wires · 55 mentions in the last 24h
Sentiment
+20
Momentum
80
Mentions · 24h
55
Articles · 24h
127
Affected sectors
Related markets

Key facts

  • NVDA Q1 revenue $81.6B (+85% YoY), data center revenue $75.2B (+92% YoY)
  • Q2 guidance $91B vs. $84-86B analyst expectation; $80B new share buyback announced
  • Amazon adding 1M+ Blackwell and Rubin GPUs this year, roughly $30-40B capex
  • US 30-year Treasury yield at highest level since 2007; 37% Fed hike probability priced for 2026
  • NVDA shares declined 2.5% after-hours despite beating estimates

What's happening

Nvidia delivered the earnings beat the market had been waiting for, with Q1 revenue reaching $81.6 billion against an expected $79.2 billion, and data center revenue alone hitting $75.2 billion, up 92% year-over-year. Q2 guidance of $91 billion further underscored that Blackwell demand is soaring, and the company added an $80 billion share buyback alongside a dividend lift. On paper, the print was textbook perfect for the AI narrative that has dominated mega-cap equity gains for the past 18 months.

Yet despite this blowout, shares retreated in after-hours trading as the broader market sentiment curdled around macroeconomic headwinds. Treasury yields climbed sharply on expectations that elevated rates may persist longer than hoped, a dynamic amplified by ongoing Middle East tensions and inflation concerns. Bond markets are pricing in a 37% probability of a Fed hike in 2026, a stark reversal from the earlier consensus of rate cuts. This creates a structural problem for the AI trade: if hyperscalers face higher borrowing costs for the massive capex cycles required to build out Blackwell and Rubin infrastructure, the return-on-capital calculus deteriorates, even if near-term revenue growth remains robust.

Amazon disclosed it will add over 1 million Blackwell and Rubin GPUs this year, roughly $30 to $40 billion in chip purchases, confirming that institutional appetite for AI infrastructure remains insatiable. Yet the pricing environment for GPU rentals is tightening. Internal data suggests H100 rental prices are up roughly 20% in 2026 despite the chip being three generations old, a sign of constrained supply and pent-up demand but also a red flag for margin compression if that capacity is later underutilized. The market's muted reaction to Nvidia's beat signals that investors are shifting focus from growth to risk-adjusted returns; when treasury yields near decade highs, even 85% revenue growth must contend with a higher discount rate.

The real debate now centers on whether hyperscalers can profitably deploy capital at these lending rates, or whether the capex cycle will plateau sooner than consensus expects. Bearish voices argue that the AI infrastructure buildout is front-loaded and that excess capacity will emerge within 12 to 18 months, capping pricing power. Management's guidance assumes zero contribution from China data center sales, a prudent stance but one that masks upside optionality if geopolitical tensions ease. Either way, Nvidia remains the bellwether for the entire AI ecosystem, and this earnings cycle proved that growth alone is no longer sufficient when real rates are elevated.

What to watch next

  • 01Amazon, Microsoft earnings calls for capex guidance details: next 2 weeks
  • 02Fed speakers on rate path in light of inflation concerns: this week
  • 03US CPI data print: June 10
Mention velocity · last 24 hours
Coverage from these sources
Previously on this story

Related coverage

More about $NVDA

Topic hub
AI Capex: Who's Spending, Who's Earning, and What's at Risk

Tracking AI infrastructure capex — hyperscaler spend, data center buildouts, memory demand and the margin compression risk.