"Incredible Piece Of Engineering:" Ex-Intel CEO Pat Gelsinger Hails DeepSeek

Pat Gelsinger, former Intel CEO, has reacted to DeepSeek AI's meteoric rise, highlighting three key lessons from computing history that he believes the AI community must embrace in light of the Chinese company's disruptive success.

Sharing his insights on X, Mr Gelinsger said computing obeyed the gas law, meaning when computing becomes dramatically cheaper, it doesn't shrink the market - it expands it. He criticised market reactions that saw AI stocks tumble, arguing that lowering AI costs will only "make AI much more broadly deployed."

Engineering thrives on constraints: With limited access to cutting-edge chips due to export restrictions, DeepSeek's team of Chinese engineers had to innovate with fewer resources and deliver a world-class model at 10-50X lower cost than competitors.

Open wins: Mr Gelsinger voiced concern over AI's increasingly closed ecosystem, supporting DeepSeek's decision to open-source its model. He called for more transparency in AI development, aligning himself with Grok's Elon Musk's advocacy for open AI rather than OpenAI CEO Sam Altman's closed approach.

"Thank you DeepSeek team.," he concluded his post.

Wisdom is learning the lessons we thought we already knew. DeepSeek reminds us of three important learnings from computing history:
1) Computing obeys the gas law. Making it dramatically cheaper will expand the market for it. The markets are getting it wrong, this will make AI…

— Pat Gelsinger (@PGelsinger) January 27, 2025

In response to a commentator on X, Mr Gelsinger, who retired from Intel last year after being given the choice to step down or be removed from the board, said that if he were still leading the company, he would be open to working with DeepSeek.

yes

— Pat Gelsinger (@PGelsinger) January 27, 2025

DeepSeek's innovative approach is a major industry disruptor. Instead of relying on expensive, high-end hardware like its competitors, DeepSeek trained R1 using 2,000 Nvidia H800 GPUs over just two months at a cost of approximately $5.5 million - a fraction of the billions US firms typically spend on AI infrastructure.

The immediate market response has been dramatic, with Nvidia's stock taking a hard hit - losing almost $600 billion - and DeepSeek's app surging in popularity.

But while the market panicked, Pat Gelsinger doubled down. In a detailed LinkedIn post, he revealed he was still buying Nvidia stocks, convinced that cheaper AI would expand the industry's potential, not shrink it. "Today I'm an Nvidia and AI stock buyer and happy to benefit from lower prices," he wrote.

In his post, he also reflected on a conversation with legendary computer scientist Donald Knuth, who once said that "he did his best work when the resources were most severely limited and schedules most demanding". Mr Gelsinger connected this philosophy to DeepSeek's achievements, praising its highly constrained Chinese team for delivering a breakthrough AI model.

Further, drawing comparisons to Linux, USB, and Wi-Fi, he claimed that history has repeatedly shown "open wins every time it is given a proper shot."

In closing, Pat Gelsinger called DeepSeek "an incredible piece of engineering" that will usher in greater AI adoption and reset the industry's approach to open innovation.