Lisa Su's AMD nearly quadrupled in value to $335B since 2022 AI pivot, now challenging Nvidia

Summary

AMD's market value has nearly quadrupled to $335 billion since Lisa Su announced an AI pivot at a board meeting in late 2022. Su told the board she would redirect the entire company toward AI as a once-in-a-lifetime opportunity. That bet has made Su a billionaire and positioned AMD as the only viable challenger to Nvidia's dominance in AI chip design.

AMD has moved from essentially zero market share in data center CPUs five years ago to an estimated 41% today. The company's valuation climbed from $90 billion to $335 billion despite recent volatility.

Two deals in October accelerated investor confidence. Oracle and OpenAI both agreed to purchase tens of thousands of AMD's MI 450 chips, the company's newest generation. Su described the OpenAI partnership as a huge vote of confidence that turbocharged the roadmap. Semi Analysis research shows the MI 450 is performing well and has caught up significantly to Nvidia this year, even outperforming Nvidia on certain models in some cases.

Su is now making bolder market predictions. A year ago, she forecast the AI chip market would reach $500 billion annually by 2028. At an investor presentation last week in New York, she raised that to $1 trillion per year by 2030. For context, Nvidia CEO Jensen Huang has said his company alone has line of sight to $500 billion in revenue by 2026, a figure that underscores how much larger Nvidia's current position is.

A critical advantage for AMD lies in the shift from training to inference. As AI models like ChatGPT become embedded in daily software, demand for inference computing is expected to surge by a billion times. Inference requires less raw computing power than training, playing to AMD's relative strengths. The company has strong inference chips but has historically struggled to design processors powerful enough to compete with Nvidia in training workloads.

Su's background shaped this strategy. A Taiwan-born executive who spent time at IBM, she took over AMD in 2014 when the company had a market value under $3 billion. She cut deals with Chinese partners to stabilize finances and capitalized on Intel's stumbles to build AMD's foothold in CPUs and data center processors. When she proposed the AI pivot in 2022, board members trusted her conviction and execution track record.

Nvidia remains the dominant force, a company expected to exceed $200 billion in revenue next year with thousands of AI developers locked into its proprietary software ecosystem. But Nvidia cannot manufacture enough chips to meet demand. The OpenAI deal demonstrated that large AI companies are willing to diversify suppliers rather than depend entirely on Nvidia.

Su has begun to embrace celebrity status as AMD's profile has risen. Earlier this year at an industry summit in Paris, she was treated like a celebrity with attendees asking for selfies and autographs. In July, at a Trump AI summit, she pulled an MI 450 chip from her pocket to show the audience, a device with 185 billion transistors that took nine months to manufacture. Trump later mentioned her in his remarks, though he gave more prominent applause to Huang.

On the question of an overheating AI market and bubble risk, Su is unambiguous. Concerns about oversupply are exaggerated and shortsighted. It is more dangerous to underinvest than to overinvest in AI infrastructure. That conviction, and the board's willingness to back it, is what separates this moment from Intel's decline. Su had the vision and the conviction to move fast.