AMD is going after Nvidia

Nvidia may still dominate the artificial intelligence chip market, but Advanced Micro Devices Inc. (AMD) is barreling ahead, and CEO Lisa Su believes the $500 billion tipping point is now closer than ever.

At the “Advancing AI” event in San Jose, California, on Thursday, Su made a bold announcement: AMD’s MI350 series — led by the MI355X — has began shipping earlier this month and delivers “35 times faster” inference performance than its predecessor, she said, which would dramatically accelerate AMD’s competitiveness in the global market.

What stood out in particular, though, was Su’s recalibrated market outlook. Previously, she predicted that the global AI processor would reach $500 billion in market revenue by 2028. Now, she’s projecting that the threshold will be surpassed in under three years.

“People used to think that $500 billion was a very large number,” she said after her presentation. “Now, it seems well within grasp.”

That projection is backed by AMD’s expanding portfolio. At the event, Su showcased the company’s coming MI400 series — the product, she said, which will position AMD as clear leaders over Nvidia’s available tech when AMD’s series debuts next year. Su said AMD is adding memory and components to access information more quickly, offering an important advantage. She also gave a glimpse of the 2026 “Helios” rack-scale server, which packs up to 72 of these chips and is built on open networking standards — an intentional pivot away from Nvidia’s closed NVLink architecture.

“The future of AI is not going to be built by any one company or in a closed ecosystem. It's going to be shaped by open collaboration across the industry,” Su said.

AMD has seen strategic partnerships follow. OpenAI CEO Sam Altman joined Su on stage at the event, remarking that initial specs for the MI400 were so ambitious he thought, “No way.” He confirmed OpenAI is working with AMD on the company’s MI450 chips. And major names such as Meta, Oracle, and xAI are already testing or deploying MI300X successors and supporting AMD’s modular approach. AI cloud provider Crusoe told Reuters that it’s planning to buy $400 million of AMD’s new chips.

Still, AMD’s challenges remain steep. 

Despite gains, the company still commands just a sliver of the market, while Nvidia continues to dominate with a 90–98% share. And AMD’s stock dipped 2.2% after the San Jose event — as investors are seemingly waiting for proof that momentum can translate into significant market share shifts. The company’s stock is down 3.26% year-to-date, while Nvidia’s is up 2.74%.

Nvidia’s software ecosystem remains the gold standard for developers, and its NVL72 rack-scale systems are deeply embedded across major data centers. The company’s integration of hardware and software makes it difficult for competitors to chip away at its lead — even with better pricing or open standards.

Additional headwinds linger for both companies, however. U.S. export restrictions — particularly on China-bound chips — could affect AMD’s revenue to the tune of around $1.5 billion this year. AMD is advocating for eased curbs, citing a Saudi deal as a step toward ensuring its technology remains globally accessible.

But as AI inference becomes the dominant workload, price‑performance and open architectures will become more and more significant. On Thursday, Su’s message was clear: AMD isn’t just ready to chip away at Nvidia, it wants to challenge it head-on — with silicon and systems that span the full stack.

If the global market for AI products and services truly surges past a trillion dollars in the next few years, as Bain & Company predicted earlier this year, AMD is positioning itself to claim a larger portion of the market.