Chipmaker AMD used its sold-out Advancing AI conference in San Francisco to introduce Helios, a rack-scale AI system intended to challenge incumbent Nvidia in the data‑center accelerator market. Dr. Lisa Su, AMD Chair and CEO, presented the new system and highlighted a roster of launch customers — including Microsoft — as AMD prepares to begin shipments later this year.
What Helios is and where it’s aimed
Helios is a rack-scale configuration that aggregates many processors into a single high‑performance unit, designed for data centers that train and run large AI models and other compute‑intensive workloads. Lisa Su described Helios as the industry’s "highest‑performance AI rack," built to train and operate the most demanding frontier models at massive scale. AMD said leading AI companies will deploy the system at gigawatt scale.
Customers and partnerships
AMD named several well‑known customers that plan to deploy Helios: OpenAI, Meta, Oracle, Anthropic and Microsoft. The system was first revealed in 2025 and was shown onstage at CES in January 2026. Microsoft CEO Satya Nadella said on Monday that Microsoft will expand its Azure infrastructure with Helios. Separately, Anthropic and AMD announced a strategic partnership on Wednesday to deploy up to two gigawatts of GPUs via the new rack system.
Competition and performance
Nvidia has historically dominated the rack-scale AI market with products such as Vera Rubin and Grace Blackwell. AMD is explicitly trying to enter this space, and The Register reported that Helios outperforms Vera Rubin on a number of metrics, suggesting AMD may have a competitive opportunity.
Venice‑X CPU
Alongside Helios, AMD introduced the Venice‑X server CPU, targeted at data centers and high‑compute workloads. The Venice‑X is expected to launch in 2027.
Market outlook and strategic views from AMD
During her remarks, Lisa Su discussed the trajectory of the chip industry and forecast that by 2030 chips that power AI will become a very large part of the overall computing market. She said the sector is experiencing a step change in compute demand driven largely by the rise of agentic AI, which requires repeated reasoning, tool calls and data access — behaviors that multiply GPU needs.
Su provided a concrete market estimate: she expects the AI accelerator market to reach about $1.4 trillion by 2030, and she said GPUs are likely to make up the vast majority of that market because current algorithms and changing workloads favor programmability in silicon.
Why this matters
Helios and AMD’s accompanying hardware announcements mark a notable move into a market that has been dominated by Nvidia. The presence of major customers and the Anthropic agreement to deploy up to two gigawatts of GPUs indicate significant commercial interest. If Helios delivers the promised performance and scale in real‑world deployments, it could influence how large AI labs structure compute capacity and affect competition, pricing and supply dynamics in the AI accelerator market.
Timeline — key dates
- 2025: Helios first revealed
- January 2026: Helios shown at CES
- Thursday (Advancing AI conference): Helios presented in San Francisco at the sold‑out event
- Monday: Satya Nadella announced Microsoft will expand Azure with Helios
- Wednesday: Anthropic and AMD announced a partnership to deploy up to two gigawatts of GPUs
- 2027: Venice‑X CPU expected to launch
