
AMD is hosting its first dedicated AI event since June 2025 next week in San Francisco, where analysts expect it to announce a higher addressable market for AI CPUs and potentially reveal Anthropic as a new customer diversifying its GPU suppliers. The event will also showcase details on AMD's next-generation MI500 GPUs, which are claimed to deliver a 1,000-times AI performance uplift versus current hardware and are targeted for 2027 release.
Summaries like this, in your inbox every morning.
Sign up free →What happened
AMD hosts its Advancing AI 2026 event next week in San Francisco, where Jefferies analysts expect the company to announce a higher addressable market estimate for AI CPUs (above $200 billion(約32兆円)) and potentially reveal a new customer partnership with Anthropic, the AI company.
Why it matters
New customer wins have been the biggest driver of momentum at AMD's past two AI events. Anthropic's reported hiring of engineers with ROCm experience suggests the company is preparing to diversify its computing infrastructure away from sole reliance on other chip makers, which could validate AMD's competitive standing in the GPU market.
What to watch
Jefferies is monitoring details on AMD's next-generation MI500 GPUs—previewed at CES 2026 with a claimed 1,000-times AI performance uplift versus an eight-GPU MI300X node and targeting 2027 release—and confirmation of the co-packaged optics approach and its supplier for the platform's 256-GPU-per-rack scale-up architecture.
Advanced Micro Devices will host its Advancing AI 2026 event next week in San Francisco, marking its first dedicated AI day since June 2025, when it launched its MI350 series GPUs and previewed its Helios rack system. According to Jefferies analysts covering the event, AMD is expected to raise its addressable market estimate for AI CPUs above $200 billion(約32兆円), surpassing the figure Nvidia provided in May.
The centerpiece of analyst expectations is a potential announcement involving Anthropic, the AI company. Jefferies noted that Anthropic has been hiring engineers with ROCm experience—AMD's open-source GPU compute framework—which the analysts interpret as a sign that Anthropic is preparing to diversify its computing infrastructure. Supply chain checks also suggest Microsoft has become a customer for AMD's MI400 series GPUs, joining previously disclosed customers OpenAI and Meta Platforms. Jefferies cautioned that deal economics matter more than headlines, pointing out that AMD has already committed 20% of the company to OpenAI and Meta, so future partnerships would need to operate on smaller incentive packages. A more traditional Anthropic agreement, the analysts said, would reinforce confidence in AMD's ability to compete without equity incentives.
On the product side, Jefferies expects new disclosure on the MI500 series, which was previewed at CES 2026 as CDNA 6 architecture built on an advanced 2nm process with HBM4E memory and targeted for 2027. AMD claims a 1,000-times AI performance uplift for the MI500 platform versus an eight-GPU MI300X node. The MI500 platform is expected to move to a native Ultra Accelerator Link scale-up domain with 256 GPUs per rack, a configuration that may require optical interconnects. Jefferies is watching closely for confirmation of a co-packaged optics approach and identification of its supplier, noting AMD's investment in Ayar Labs and its ongoing work with Astera Labs on Ultra Accelerator Link and with Broadcom on scale-up networking.
AMD's Advancing AI 2026 event arrives at a critical juncture for the company's competitive positioning in AI infrastructure. Jefferies analysts indicate that new customer announcements have historically been the biggest swing factor at AMD's AI events, suggesting the market views customer wins as the true measure of AMD's ability to compete with Nvidia. The reported Anthropic engagement is significant precisely because it would represent a customer making an independent choice to diversify its supplier base, rather than relying on equity-heavy deal incentives. AMD has already committed 20% of the company to OpenAI and Meta, so future deals would need to operate on more balanced commercial terms to avoid unsustainable dilution.
The MI500 roadmap—with its claimed 1,000-times uplift and 2027 target—represents AMD's bet on next-generation architecture and process technology to maintain competitiveness. However, the technical details matter less than the commercial credibility: Jefferies expects disclosure on the MI500's infrastructure requirements, including confirmation of co-packaged optics and its supplier partner, because the feasibility and cost of scale-up connectivity directly affect the platform's real-world competitiveness and customer adoption.
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytime
No comments yet. Be the first to share your thoughts!
Log in to join the discussion



Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack