AMD announced its Helios rack-scale system at the Advancing AI conference in San Francisco, positioning it as a direct competitor to Nvidia’s Vera Rubin and Grace Blackwell systems.

Helios combines many processors into a single high-powered unit designed for data centers training and running AI models at massive scale. AMD CEO Dr.

Lisa Su called Helios the industry’s “highest performance AI rack,” and the company claimed it beats Nvidia’s Vera Rubin on several performance metrics.

The system already has committed customers including Microsoft, OpenAI, Meta, Oracle, and Anthropic, with Microsoft planning to expand its Azure infrastructure using Helios.

Anthropic and AMD also announced a strategic partnership to deploy up to two gigawatts of GPUs via the system.

AMD additionally introduced the Venice-X CPU, a data-center chip for high-computing workloads expected to launch in 2027.

Su projected that by 2030 the AI accelerator market would reach approximately $1.

4 trillion, nearing the size of the entire semiconductor market today, driven by demand from agentic AI which requires many GPUs to handle multi-step reasoning and tool calling.

She expects GPUs to dominate this market due to the early stage of algorithm development and changing workloads favoring programmability.