Back to feed
News Story
SSignal90
InfoQ
2 sources

AMD Unveils World's First 2nm GPU MI455X, with OpenAI, Meta as Early Customers

At the Advancing AI 2026 conference, AMD announced the Instinct MI455X accelerator, the first data center GPU using 2nm compute chiplets, and the Helios rack system integrating 72 GPUs. The MI455X features 320 billion transistors, 432GB of HBM4 memory, and delivers up to 2.9 ExaFLOPS FP4 per rack. OpenAI, Meta, Microsoft, and Oracle are among the first customers, with shipments expected from late Q3 2026. The launch targets NVIDIA's dominance by combining advanced process technology with massive memory capacity for AI workloads.

SynthePulse Insight · AI deep reading

AMD Helios Takes on Nvidia: AI Infrastructure Landscape Shifts

Version 1 · 1 source

At the 2026 Advancing AI conference, AMD unveiled the Helios AI rack system, claiming performance surpassing Nvidia's Vera Rubin rack, and has secured key customers including OpenAI and Microsoft. CEO Lisa Su forecasts the AI accelerator market will reach approximately $1.4 trillion by 2030, approaching the size of the entire semiconductor industry.

  • AMD launches Helios AI rack system, positioning it as the highest-performance AI rack, directly challenging Nvidia.
  • Helios has secured major customers including OpenAI, Meta, Oracle, Anthropic, and Microsoft; Microsoft plans to use it to expand Azure infrastructure.
  • AMD and Anthropic announce strategic partnership to deploy up to 2 gigawatts of GPUs via Helios.
  • Reports claim Helios outperforms Nvidia's Vera Rubin rack in multiple benchmarks.
  • AMD also unveils Venice-X CPU, targeting data centers, expected to launch in 2027.
  • Lisa Su predicts the AI accelerator market will reach approximately $1.4 trillion by 2030, with GPUs continuing to dominate due to programmability.
Open section navigationHelios Launch: Performance and Customer Roster

Helios Launch: Performance and Customer Roster

On July 27, 2026, AMD officially launched the Helios AI rack system at the Advancing AI conference in San Francisco. CEO Lisa Su described it as "the highest-performance AI rack in the industry," designed for training and running cutting-edge AI models, with deployment set to begin later this year.

Helios has secured major customers including OpenAI, Meta, Oracle, Anthropic, and Microsoft. Microsoft CEO Satya Nadella confirmed the use of Helios to expand Azure infrastructure. Additionally, AMD and Anthropic announced a strategic partnership to deploy up to 2 gigawatts of GPUs via Helios.

Reports claim Helios outperforms Nvidia's Vera Rubin rack in multiple benchmarks. However, the report did not provide specific benchmark names or scores, so this performance advantage is a source claim pending independent verification.

Market Forecast and Strategic Positioning

At the conference, Su noted that the rise of agentic AI—which requires extensive multi-step reasoning and tool use—is driving significant growth in computing demand. She predicted the AI accelerator market will reach approximately $1.4 trillion by 2030, approaching the size of the entire semiconductor industry today.

She further stated that due to the rapid evolution of AI workloads and algorithms, GPUs are expected to continue dominating the market because of their programmability. This assessment provides strategic justification for AMD's GPU roadmap.

Complementary Product: Venice-X CPU

AMD also unveiled the Venice-X CPU, designed for data center workloads, expected to launch in 2027. This product complements the Helios rack system, but the report did not provide specific specifications or performance data.

Credibility boundary

This article is primarily based on a single source, The AI Insider, which is a secondary industry media outlet. The performance comparison (Helios outperforming Vera Rubin) is a source claim and has not been independently verified. The customer list (OpenAI, Meta, etc.) and partnership details (2 GW GPU deployment) are also source claims. The market forecast ($1.4 trillion) is a statement from AMD's CEO, representing a company executive's prediction, not an independent third-party forecast.

Insight takeaway

AMD has officially challenged Nvidia with the Helios rack system, leveraging a strong customer roster and performance claims to capture a larger share of the AI infrastructure market. However, the performance advantage requires independent verification, and Nvidia's ecosystem moat remains a key variable.

Primary report

InfoQ

Primary source

Same-event coverage

Also covered by 1 sources