Sharon Zhou, AMD’s chief AI officer, has moved to Anthropic to focus on frontier model research and GPU kernel optimization as Anthropic prepares for its public listing.
Sharon Zhou’s transition from AMD to Anthropic and talent exchange with OpenAI
Farewell message to Lisa Su and applying reinforcement learning to processor core optimization
Career path from Stanford University and Lamini to leading AMD
Strategic partnership between AMD and Anthropic to deploy 2 gigawatts of chips
Analytical overview of the strategic and economic dimensions of the event
Frequently asked questions
Sharon Zhou’s transition from AMD to Anthropic and talent exchange with OpenAI
Sharon Zhou, who spearheaded the AI research ecosystem at chipmaker AMD and reported directly to CEO Lisa Su, has left her position to join Anthropic. Zhou announced via X that this step marks her return to frontier AI model research, according to a report by State of AI.
Her official announcement was issued on Wednesday, November 4, 2026, according to the State of AI report and NetEase. On the same day, Anthropic lost a prominent product leader to its competitor OpenAI; Kat Wu, who managed the Claude Code product, announced she is joining OpenAI to lead product management for the core products group. Both moves come as Anthropic prepares for its initial public offering on the stock exchange.
Farewell message to Lisa Su and applying reinforcement learning to processor core optimization
A translation of Sharon Zhou’s post published by NetEase showed her expressing excitement about returning to the model research that drew her to the field a decade ago, describing Lisa Su as an exceptional leader, manager, and friend who gave her extensive operational autonomy. Zhou wrote that she founded and led AMD’s AI research ecosystem, which drove partnerships with frontier AI labs and served as executive sponsor for those projects.
She explained that her team was among the first research groups to seriously apply reinforcement learning techniques to optimize GPU cores, adding that this work convinced her achieving frontier progress might require about 100 times fewer individuals than she previously thought, and she now adjusts that estimate to roughly 1,000 times fewer. According to the State of AI report, her work focuses on enabling models to write and tune processor kernels—the low-level code that controls chip execution efficiency—directly tying into the chip business Anthropic presented to IPO investors this week.
Career path from Stanford University and Lamini to leading AMD
Sharon Zhou co-founded and managed the generative AI startup Lamini. In June 2025, she joined AMD as vice president of artificial intelligence alongside several Lamini employees, with an AMD spokesperson telling CRN at the time that the move was a talent acquisition rather than a company buyout. Before founding Lamini, Zhou served as a part-time faculty member in the computer science department at Stanford University, where she earned her Ph.D. under the supervision of prominent scientist Andrew Ng.
Her leadership role expanded in 2026, with AMD congratulating her on assuming the role of chief AI officer. Zhou explained that the expanded position meant she reported directly to CEO Lisa Su.
Strategic partnership between AMD and Anthropic to deploy 2 gigawatts of chips
Her new employer and former employer share a close commercial and strategic partnership; AMD and Anthropic announced a strategic partnership to deploy up to 2 gigawatts of AMD Instinct MI450 series GPUs, according to AMD disclosures.
Sharon Zhou herself worked on securing that massive deal. In a LinkedIn post, she explained that Anthropic’s plan to use up to 2 gigawatts of AMD processors equates to “hundreds of thousands of GPUs and tens of billions of dollars,” writing at the time: “I am excited to see AMD processors become a mainstream choice across AI labs.”
Analytical overview of the strategic and economic dimensions of the event
Sharon Zhou’s move from her high-ranking leadership position at AMD to Anthropic represents a prominent model of rare scientific talent movement between chip manufacturers and major foundation model labs. Zhou’s academic and entrepreneurial background—spanning Stanford University, founding Lamini, and leading AI research at AMD under the direct supervision of CEO Lisa Su—centers on one of the most delicate and sensitive technical nodes: employing reinforcement learning to enable AI models to write and optimize low-level GPU kernels.
This technical skill gains double strategic value for Anthropic as it prepares for its IPO and negotiates multi-billion-dollar compute partnerships, including its agreement with AMD itself to deploy up to 2 gigawatts of Instinct MI450 processors. The more Anthropic succeeds in optimizing the efficiency of the code managing diverse chip execution, the more it can cut massive compute bills and accelerate the training and operation of Claude models, all amidst a competitive landscape that concurrently sees product leaders like Kat Wu moving to OpenAI.
Frequently asked questions
Question: What position did Sharon Zhou hold at AMD, and why did she move to Anthropic?
Answer: She served as vice president and chief AI officer at AMD, reporting directly to CEO Lisa Su, and moved to Anthropic to return to frontier model research and optimize processor kernels.
Question: How does Sharon Zhou’s technical specialization relate to Anthropic’s needs?
Answer: Zhou specializes in using reinforcement learning to enable models to write and optimize low-level GPU kernels, supporting the operational efficiency of the chips Anthropic presented to IPO investors.
Question: What is Sharon Zhou’s academic and entrepreneurial background before joining AMD?
Answer: She earned her Ph.D. from Stanford University under Andrew Ng’s supervision and served on its faculty, then co-founded and managed generative AI startup Lamini before joining AMD in June 2025.
Question: What is the scale of the strategic partnership between AMD and Anthropic?
Answer: The partnership—which Zhou worked on—involves deploying up to 2 gigawatts of AMD Instinct MI450 processors, which she estimated at hundreds of thousands of processors and tens of billions of dollars.