[
When veteran engineer and executive Pat Gelsinger returned to Intel as CEO in 2021, the once-great chipmaker was in a slump. Having failed to adapt to the mobile age and then missing several steps in manufacturing cutting-edge microprocessors, it was now also lagging behind in supplying chips to meet the tech industry's growing appetite for artificial intelligence.
With optimism that sometimes seemed reckless, Gelsinger promised that Intel would make an epic comeback. He vowed to shake up its sleepy corporate culture, refocus on core engineering and deliver a revitalized manufacturing plan that will put rivals TSMC and Samsung on notice.
This week, Gelsinger announced that Intel's comeback plan is well and truly on track. He announced a rebrand of the company's “foundry” business, which manufactures chips designed by other companies, saying Intel's latest manufacturing process will produce silicon chips equally efficient and capable as TSMC's later this year. Microsoft is the first major customer of this new chipmaking technology – a significant coup for Intel as it tries to convince the industry it can offer competitive products suited to the age of AI.
Pat Gelsinger spoke with WIRED senior writer Will Knight about Intel's AI reboot over Zoom from his home in Santa Clara, California. The conversation has been lightly edited for length and clarity.
Will Knight: You announced this week that Intel would relaunch its business, which makes chips on behalf of other companies, as an “AI-era systems foundry.” What does it mean?
Pat Gelsinger: I started Intel's strategy more than two years ago, and for the company, this has been unprecedented growth in generative AI. This has been Nvidia's land, but we are a company that really has an opportunity to participate 100 percent in the AI market. We know how to connect networks and memory and provide supply chain and all these other elements that we find customers are extremely excited to take advantage of.
Speaking of the AI boom, what did you make of the reports that OpenAI CEO Sam Altman wants to raise $7 trillion to develop and manufacture the chips needed to guarantee progress in AI?
My first reaction was, that's a surprisingly large number. And then I had to do the math. Today, the largest AI models were built on approximately 10,000 GPUs. The assumption is that for the largest AI model to be produced in the future we will probably need 10 million.
We are already saying that we can spend a few billion dollars on training the most advanced models today. Plus, the $7 trillion math also includes power and data centers.
This week you said Intel was on track to deliver its new “18A” manufacturing process, which would compete with TSMC's best offerings. What else are you doing to get an edge?
The entire industry is pursuing this next generation of transistors, which we call RibbonFETs. I think everyone is[asking]who will make the subsequent finest transistor on the planet.
However what everyone seems to be giving us credit score for is bottom energy, this new manner of delivering energy into the machine, which supplies you higher present resistance efficiency, nevertheless it's additionally bettering the density of the chip. Which means the identical wafer, as an alternative of manufacturing 100 chips, can produce 120 chips. It is a nice worth proposition.
You introduced Microsoft as a buyer of your foundry enterprise. However Intel had beforehand lagged behind within the competitors on this market. How do you clarify to prospects that issues are completely different this time?