Nvidia reports AI chip surge, cites 300% AI growth and warns graphics shortage

Nvidia’s Q2 earnings call revealed that its data‑center GPU segment, now branded as “AI chips,” generated $6.5 billion in revenue, up 300 percent year‑over‑year, according to the company’s own presentation cited by The Globe and Mail. The surge is driven primarily by “sovereign AI” workloads—large‑scale models deployed by governments and hyperscale cloud providers—whose demand has outpaced the growth of traditional graphics workloads. Nvidia’s CFO noted that the AI‑chip business now accounts for roughly 70 percent of total data‑center sales, a dramatic shift from the 30 percent share recorded a year earlier.

The rapid expansion has immediate supply‑chain implications. Nvidia warned that its graphics‑card inventory, which had previously been oversupplied, will become scarce as fab capacity is reallocated to AI‑focused products. Mix Vale reported that the company’s leading‑edge H100 and the newly announced Vera Rubin GPUs will consume the bulk of its 7‑nanometer and 5‑nanometer wafer allocations through the remainder of 2026, leaving little slack for the GeForce line that powers consumer PCs and gaming consoles. Analysts cited by Mix Vale estimate that the shortage could tighten further after the company’s upcoming CES 2026 announcements, where Nvidia is expected to unveil a new generation of AI accelerators.

At CES 2026, Nvidia presented the Vera Rubin architecture, which it claims delivers five times the AI performance of the current H100 while reducing inference costs by a factor of ten, according to The‑Decoder’s coverage of the event. The company said the new chip leverages a “next‑generation tensor core” and a redesigned memory subsystem that cuts latency for large language model inference. If the performance claims hold, Vera Rubin could accelerate the migration of workloads from on‑premise clusters to cloud‑based AI services, further cementing Nvidia’s dominance in the sovereign AI market.

Despite the bullish outlook, Nvidia’s leadership acknowledged the risk of over‑reliance on a single product line. In the earnings call, the CEO emphasized that the firm is expanding its “AI‑centric” roadmap to include lower‑power variants aimed at edge devices and autonomous vehicles, a move designed to diversify revenue streams beyond the high‑end data‑center segment. The company also announced a partnership with several satellite manufacturers to deploy AI‑enabled processing units in orbit, a venture that The Globe and Mail highlighted as part of Nvidia’s broader strategy to embed its GPUs in “space‑based AI workloads.”

The confluence of explosive AI demand, constrained fab capacity, and aggressive product rollouts positions Nvidia at the center of a tightening graphics market. As the company reallocates silicon toward AI accelerators, gamers and workstation users may face longer lead times and higher prices for GeForce cards, a trend already evident in early‑2024 inventory reports. If Nvidia can sustain the 300 percent growth trajectory while managing supply constraints, it will reinforce its role as the de‑facto hardware provider for the next wave of AI innovation; failure to do so could open a window for rivals such as AMD and Intel to capture a portion of the beleaguered graphics segment.