AMD takes on Nvidia with its Helios AI rack scale system
AMD says Helios will ship later this year with Microsoft, OpenAI, Meta, Oracle, and Anthropic lined up to deploy it.
Lisa Su pitched Helios as AMD’s “highest-performance AI rack,” built for frontier-model training and inference at massive data-center scale. The piece says AMD is trying to challenge Nvidia’s dominant Vera Rubin and Grace Blackwell rack systems, with reported benchmark advantages over Vera Rubin. Anthropic and AMD also announced a partnership to deploy up to two gigawatts of GPUs through Helios, while Microsoft plans to expand Azure infrastructure with it. AMD also introduced its Venice-X data-center CPU, expected in 2027. TechCrunch AI's note
Lisa Su pitched Helios as AMD’s “highest-performance AI rack,” built for frontier-model training and inference at massive data-center scale. The piece says AMD is trying to challenge Nvidia’s dominant Vera Rubin and Grace Blackwell rack systems, with reported benchmark advantages over Vera Rubin. Anthropic and AMD also announced a partnership to deploy up to two gigawatts of GPUs through Helios, while Microsoft plans to expand Azure infrastructure with it. AMD also introduced its Venice-X data-center CPU, expected in 2027. TechCrunch AI's note
score 6