AMD Unveils Helios Rack and MI455X Accelerator in Bid to Challenge Nvidia
Lisa Su brought OpenAI and Anthropic executives on stage to endorse AMD hardware, announced a Cerebras partnership for ultra-low-latency inference, and predicted the total addressable market will reach $2 trillion by 2030
July 23, 2026 – 7:58 pm
Image by: Coolcaesar
Summary
AMD launched MI455X chips, Helios racks, and Venice Epyc CPUs at its Advancing AI event in San Francisco on Thursday, aiming to outperform Nvidia’s competing hardware in AI training and inference. Key announcements included:
- MI455X AI accelerator
- Helios server rack, housing 72 of these chips
- Venice generation of Epyc server processors built on TSMC’s two-nanometre process
AMD forecasts the total market it is targeting will reach $2 trillion by 2030, with AI accelerators alone accounting for over a trillion dollars.
Market and Shares Reaction
Investors showed mixed reactions, with AMD shares dropping about four percent during the presentation despite a year-to-date stock increase of over 145%. Nvidia’s dominance in the market is evident, with its Vera Rubin platform already in full production and deployed by clients like OpenAI, CoreWeave, and Microsoft.
Endorsements and Partnerships
Su invited executives from leading AI companies to the stage to support AMD hardware:
- OpenAI’s Sachin Katti expressed plans to deploy Helios "at massive scale"
- Cerebras CEO Andrew Feldman announced a partnership to combine Helios racks with Cerebras servers for ultra-fast AI inference, competing directly against Nvidia’s Groq acquisition.
AMD’s Competitive Advantage
AMD aims to compete with Nvidia across multiple fronts:
- Accelerators: MI455X vs. Nvidia’s platform
- Server CPUs: Venice Epyc vs. Nvidia’s Vera
- Complete Data Centre Solutions: Helios racks and Cerebras partnership
Independent benchmarks are yet to conclusively settle these direct competitions, but AMD’s robust portfolio positions it for a significant market share in the future.