Nvidia's Rebound Puts OpenAI Jitters in Rearview as Hyperscaler Earnings Take Center Stage

Nvidia shares hit a fresh 52-week high as OpenAI refutes revenue concerns and Qualcomm's mobile AI partnership lifts the sector, despite production cuts ahead.

The chipmaker's stock has been on a wild ride this week, but by Wednesday it had already erased the damage from a troubling report about its marquee customer. Nvidia shares hit a fresh 52-week high of €182.94, extending their monthly gain to nearly 27%, after a brief selloff triggered by concerns over OpenAI's financial health.

The turbulence began when the Wall Street Journal reported that the ChatGPT developer had fallen short of internal targets for both user growth and revenue. OpenAI's finance chief, Sarah Friar, reportedly warned internally that the company might struggle to meet future commitments on data center contracts — a development that threatened a massive 10-gigawatt deal with Nvidia worth billions.

The news sent Nvidia shares down more than 3% on Tuesday, dragging down Oracle and CoreWeave along with it. The broader iShares semiconductor ETF shed 3.7% in sympathy. But the selling proved short-lived.

OpenAI pushed back hard against the narrative. Infrastructure chief Sachin Katti took to X to describe the partnership as "fundamental," insisting that the entire computing fleet runs on Nvidia processors. CEO Sam Altman also chimed in, stating the company intends to buy as much computing power as it can get its hands on. The company dismissed the Journal's report as "clickbait" in a statement to Bloomberg.

Analysts largely sided with the denials, calling the initial selloff overdone. The demand for AI chips isn't going away, they argue — it's simply being redistributed across a broader set of applications and customers.

A Qualcomm Curveball Reshapes the Narrative

The real catalyst for Wednesday's rebound came from an unexpected corner of the semiconductor world. Qualcomm announced a partnership with OpenAI to embed AI models directly into mobile processors, extending the technology's reach from massive data centers into handheld devices. The news lifted the entire sector and gave Nvidia's stock fresh momentum.

That broader demand picture will be put to the test later Wednesday, when Alphabet, Microsoft, Amazon and Meta all report quarterly earnings. Market watchers expect the four hyperscalers to confirm capital expenditure plans for 2026 that could top $700 billion. Nvidia has already locked in supply commitments worth roughly $95 billion, giving it a formidable backlog regardless of what happens with any single customer.

The pricing power behind those commitments is also strengthening. Lease costs for Nvidia's H200 chips have climbed sharply since January, underscoring the persistent shortage of high-end computing capacity.

Production Constraints Cloud the Outlook

While the demand side looks robust, Nvidia faces a structural bottleneck on the supply front. KeyBanc Capital Markets estimates the company has been forced to slash production targets for its upcoming Rubin graphics chip. Instead of the 2 million units originally planned for 2026, the bank now expects just 1.5 million.

The culprit is a delay in certifying the new HBM4 memory chips from suppliers SK Hynix and Micron. The knock-on effect for Nvidia's Vera-Rubin server racks is even steeper: forecasts have been cut in half, from as many as 14,000 units to roughly 6,000.

The next major data point for investors arrives on May 20, after the US market close. Nvidia's management has guided for first-quarter revenue of around $78 billion for fiscal 2027, a figure that already assumes a complete loss of the Chinese market for data center chips. How the company frames the ramp-down of its current Blackwell generation will set the tone for the months ahead.

For now, the market is betting that the hyperscalers' checkbooks remain wide open. Any sign of budget tightening in Wednesday's earnings calls could quickly reverse the recovery, but the early signals suggest the AI spending spree is far from over.