Nvidia’s $5 Trillion Crown Wobbles as Hyperscaler Spending Hits $725 Billion — and Customers Build Their Own Chips

Nvidia's stock fell below $5T as cloud giants build their own AI chips, China gray market surges, and earnings expectations loom ahead of May 20 report.

Nvidia’s stock has been on a rollercoaster, touching a historic $5 trillion market cap in April before sliding back below that threshold on the last trading day of the month. The shares closed at $199.57, down nearly five percent in a single session, as a potent mix of customer self-reliance, Chinese gray-market turbulence, and sky-high earnings expectations converged.

The headline numbers tell a tale of two forces pulling in opposite directions. On one hand, Nvidia’s shares have rallied roughly 14 percent in April, hitting an all-time high as the broader Philadelphia Semiconductor Index surged nearly 47 percent year-to-date. On the other, the company’s biggest customers are rewriting the rules of engagement.

Hyperscalers Go Rogue

Amazon, Google, Meta, and Microsoft have collectively raised their 2026 capital expenditure budgets for AI infrastructure to an estimated $725 billion. That torrent of spending would normally be a straight tailwind for Nvidia. But the cloud giants are increasingly diverting those dollars toward their own silicon — Google’s TPU, Microsoft’s Maia chips, and Amazon’s Trainium processors — reducing their dependence on Nvidia’s GPUs.

This strategic pivot is reshaping the competitive landscape. Nvidia’s graphics processors are no longer the sole beneficiaries of the AI spending boom. The hyperscalers are effectively becoming frenemies: they pour record sums into infrastructure while simultaneously building alternatives to Nvidia’s dominant hardware.

China’s Gray Market Hits Fever Pitch

A parallel headache is brewing in Asia. A crackdown on chip smuggling in China has sent prices for Nvidia’s B300 servers toward $1 million per unit in sanctioned regions. The extreme scarcity is pushing Chinese customers toward domestic alternatives, a trend that could accelerate as export restrictions tighten on both sides.

Huawei, the primary beneficiary, expects its AI chip revenue to surge at least 60 percent to $12 billion by 2026, with the Ascend 950PR entering mass production. The technology gap remains wide — current US chips are roughly five times more powerful than Huawei’s best offerings, and that gap is projected to widen to 17 times by 2027. But in the inference segment, where raw compute is less critical, Huawei is gaining traction. Nvidia’s management has previously warned that Chinese revenue could vanish entirely over the long term due to trade restrictions.

The Earnings Gauntlet

All eyes are now on May 20, when Nvidia reports fiscal first-quarter results. Management has guided for 77 percent revenue growth, while the analyst consensus sits at 79 percent. Options markets are pricing in a swing of more than ten percent in either direction. Anything below 80 percent growth is unlikely to excite a market that has grown accustomed to blowout numbers.

The company’s new Blackwell and Vera Rubin architectures will be the centerpiece of the narrative. Nvidia must convince investors that its technological edge remains insurmountable even as hyperscalers develop custom chips and Huawei chips away at the low end.

Wall Street remains broadly bullish, with a consensus buy rating and an average price target of $274.38 — implying upside of more than 35 percent from current levels. But the risks are mounting. The $5 trillion valuation was a milestone; defending it will require Nvidia to prove that its customers’ massive spending still flows disproportionately to its own coffers, even as those same customers build the tools to bypass them.