AMD's Biggest AI Event Yet
AMD held its flagship Advancing AI 2026 event on July 23 in San Francisco, unveiling a sweeping portfolio of next-generation AI infrastructure. CEO Dr. Lisa Su introduced the 6th Gen AMD EPYC Venice CPUs, the AMD Instinct MI400 Series GPUs, and the AMD Helios AI rack-scale solutions.
Helios: Rack-Scale AI at Gigawatt Scale
AMD Helios rack-scale solutions are built with co-optimized silicon spanning 72 high-performance AMD Instinct MI455X GPUs and 18 powerful 6th Gen EPYC CPUs, connected by AMD Pensando networking and accelerated by AMD ROCm open software. AMD claims Helios delivers up to 30% more tokens per dollar than the leading competitive solution.
Leading AI labs are already choosing Helios: OpenAI, Anthropic, Meta, Microsoft, Oracle, and several smaller cloud providers have committed to deployment.
Anthropic, OpenAI, Meta Partnerships
Anthropic and AMD announced a strategic partnership to deploy up to 2 gigawatts of AMD Instinct MI455X GPUs in Helios solutions. The companies are launching a multiyear engineering collaboration to use Claude to optimize workloads for AMD Instinct GPUs.
OpenAI and AMD are partnering to optimize the full AI stack, leveraging OpenAIs Triton framework with AMD ROCm software. OpenAI expects to bring Helios online beginning in Q4 2026.
Meta and AMD are co-designing for gigawatt-scale deployments. Cerebras and AMD are collaborating to deliver combined ultra-low-latency AI compute and high-throughput rack-scale infrastructure.
New MI400 and MI350 GPUs
The AMD Instinct MI455X GPU delivers 34x higher token throughput compared to the previous MI355X. The Instinct MI430X accelerator offers up to 288 TFLOPS of FP64 performance for scientific computing. AMD also launched the MI350P GPU, delivering up to 4.2x more tokens per second per dollar than competition.
Roadmap Through 2030
AMD extended its annual cadence: Zen 7 EPYC CPUs arriving in 2028, Zen 8 Ravenna CPUs in 2030. The Instinct MI500 Series GPUs in 2027, MI600 in 2028.
Enterprise and Physical AI
AT&T detailed its use of AMD Instinct GPUs and ROCm software to power its OTel 2.0 model for telecoms. Cisco and AMD are collaborating to combine AMD inference engines with Cisco networking. AMD also introduced Kria AI solutions for robotics.
