Get all your news in one place.
100’s of premium titles.
One app.
Start reading
Fortune
Fortune
David Meyer

AMD’s Lisa Su wants to dethrone Nvidia as AI-hardware industry leader. Success depends on software

Dr. Lisa Su, Chair and CMO, AMD speaks onstage during Vox Media's 2023 Code Conference at The Ritz-Carlton, Laguna Niguel on September 26, 2023 in Dana Point, California. (Credit: Jerod Harris—Getty Images for Vox Media)

Lisa Su has already had a very successful tenure at the home of the fabled (but these days fabless) chip giant AMD. In the past nine years, she’s rescued the company and ensured its products are found in countless data centers, not to mention its previous staple of the desktop PC. And now, despite rival Nvidia holding something in the region of 85% market share, she intends to ensure that AMD makes the most out of the AI explosion.

I spoke to Su for an article that will appear in Fortune’s October/November magazine, and that just went up on Fortune.com. The interesting thing about her AI-processor-related ambitions is that she doesn’t merely see AMD gaining a foothold by virtue of Nvidia being unable to satisfy the market’s rampant demand for AI processors—though that’s an obvious opportunity that AMD won’t let pass by.

Su thinks AMD’s rival processors, as currently exemplified by the MI300 line, can make the company “the industry leader for inference solutions”—that is, AI deployments that take already-trained models and use them to infer stuff from the fresh data that’s thrown at them.

That’s a pretty tall order when Nvidia is also trying to seize the inference market, and analysts are currently predicting that AMD would do very nicely from gaining something like 30% market share. But if AMD really is going to become the industry leader, it’s going to need to change the well-founded perception that it’s far easier to build AI models on Nvidia than on AMD, owing to Nvidia’s superior software offerings.

One AI company that thinks AMD is already there is Lamini, which provides a platform to help enterprises build their own large-language models. Last week Lamini revealed that it’s been running LLMs on AMD’s graphics processors for a year now, and said AMD’s ROCm software had now achieved “parity” with Nvidia’s CUDA. “AMD has been putting hundreds of engineers behind their general-purpose AI initiative,” Lamini cofounder Gregory Diamos, who used to be a CUDA architect at Nvidia, told me as I was researching my article.

Microsoft chief technology officer Kevin Scott also talked up AMD’s GPU chips at the Code conference last week, predicting that they would become “more and more important to the marketplace in the coming years”—comments that bumped AMD’s shares up nearly 5% the next day.

Meanwhile, Nvidia’s dominant market share is also drawing scrutiny from antitrust authorities in France, who reportedly raided its offices last week.

So maybe Su has reason to be so bullish. You can read my piece on her here. More news below.

Want to send thoughts or suggestions to Data Sheet? Drop a line here.

David Meyer

Sign up to read this article
Read news from 100’s of titles, curated specifically for you.
Already a member? Sign in here
Related Stories
Top stories on inkl right now
One subscription that gives you access to news from hundreds of sites
Already a member? Sign in here
Our Picks
Fourteen days free
Download the app
One app. One membership.
100+ trusted global sources.