Intel’s former CEO, Pat Gelsinger, recently made waves by criticizing NVIDIA’s current pricing approach for AI GPUs. Speaking at a podcast during NVIDIA’s GTC 2025 event, Gelsinger didn’t hold back, suggesting NVIDIA’s success in the AI sector might have more to do with luck than strategy. He argued that the company’s GPUs, popular for training AI models, are prohibitively expensive for AI inferencing applications.
Taking a trip down memory lane, Gelsinger reminisced about the days when CPUs ruled the computing world. He commended NVIDIA’s CEO, Jensen Huang, for his steadfast belief in a different path, starting with graphics and evolving into AI. However, Gelsinger implied that while Huang found success, he got a bit lucky along the way.
Gelsinger asserts that while NVIDIA has excelled in AI training, the next step is developing cost-effective solutions for AI inferencing. He argues that continuing to use GPUs for this purpose is overly expensive, perhaps even unsustainable, if the ambitions around AI deployment are to be fully realized.
NVIDIA’s commitment to AI and its long-term strategy has been praised. In contrast, Intel, once a dominant force in computing, has struggled to keep pace in the AI field. Intel’s ambitious “Falcon Shores” AI lineup has been shelved, putting its hopes in the forthcoming “Jaguar Shores.” Unfortunately, their current Gaudi series has not matched the performance of competitors like AMD’s Instinct or NVIDIA’s Hopper.
With Lip-Bu Tan now at the helm of Intel, there’s a cautious optimism about the company’s direction. However, details remain hazy on how Intel plans to tackle its AI challenges. Gelsinger highlighted another potential shift on the horizon—quantum computing. He believes it could define the next era of computing, possibly gaining traction by the end of this decade.
In summary, while Gelsinger offered some pointed critiques of NVIDIA’s strategy and pricing, he also acknowledged the company’s success in positioning itself at the heart of the AI movement. As Intel grapples with its AI trajectory, the broader computing world may soon shift focus to yet another frontier: quantum computing.
