AMD just showed signs of progress. But can it really take on Nvidia?

AMD unveiled its Instinct MI350 series accelerators and future MI400/MI450 plans at its Advancing AI event, with CEO Lisa Su projecting a $500B+ data-center AI chip market by 2028. Analysts view AMD as solidifying its position as the second source for GPUs behind Nvidia, though challenges remain in scaling system-level offerings and competing with custom solutions; some foresee revenue upside to $10-12B for AMD's data-center GPU revenue next year, while others emphasize the importance of full-stack solution execution to capture market share, as shares underperformed due to no new customer announcements.

Analysis

Advanced Micro Devices (AMD) showcased progress in its artificial intelligence strategy at its Advancing AI event, unveiling the Instinct MI350 series accelerators and outlining a roadmap for the MI400 and MI450 series. CEO Lisa Su amplified optimism by forecasting the total addressable market for data-center AI chips to exceed $500 billion by 2028. Analysts generally concur that AMD is solidifying its position as a credible second source to Nvidia, with the MI350 series expected to close the raw performance gap with Nvidia's Blackwell offerings, albeit with a one-year lag, and the MI450 aiming for closer competition with Nvidia's next-generation Rubin platform. Cantor Fitzgerald analysts see potential for AMD's data-center GPU revenue to reach $10 billion to $12 billion next year, up from their current $8 billion model, though they note AMD's 2024 AI revenue of over $5 billion fell short of some higher prior expectations and deem 2024 a "stopgap year" before more significant revenue acceleration in 2026-2027. Despite existing partnerships with Meta and Oracle, and a potential future engagement with Amazon Web Services hinted by Bank of America analysts, AMD's shares underperformed post-event, declining approximately 2%, which Cantor analysts attribute to the absence of new customer announcements. Nevertheless, analysts at Raymond James express increased conviction in AMD's market opportunity, suggesting a 10% to 20% long-term share of the data-center GPU market is attainable, while Seaport Research finds AMD's AI chips competitive enough to sustain business, particularly as large customers seek leverage against Nvidia and a hedge against internal chip development failures.

Key Decisions for Investors

• Investors should monitor AMD's execution on its AI chip roadmap, particularly the timely delivery and performance of the MI350, MI400, and MI450 series, as this is crucial for closing the gap with Nvidia and capturing projected market share.
• Pay close attention to new customer announcements, especially from hyperscalers, as these will be key validation points for AMD's traction and revenue growth in the AI data center market, potentially catalyzing share performance.
• Consider AMD's strategic position as a growing alternative in the expanding AI hardware market, but remain cognizant of Nvidia's entrenched leadership and the significant execution risks AMD faces in scaling its full-stack solutions to effectively compete.