AMD ups the AI infrastructure ante

• The chipmaker has been churning out major industry announcements this week
• It now claims to have the ‘highest performance rack-scale solution’ for AI factory deployments

Chipmaker AMD has used its Advancing AI 2026 event in San Francisco to launch its new AI infrastructure and physical AI portfolio, which includes AMD Helios rackscale solutions that are now in production and on course to be “deployed by leading AI companies at gigawatt scale”. 

AMD claims Helios is “the highest performance rack-scale AI solution”, boasting that it “combines leadership compute performance, memory capacity and networking bandwidth to deliver up to 30% more tokens per dollar than the leading competitive solution” (by which it means an Nvidia Vera Rubin NVL72 rack). 

“The next phase of AI will span frontier models, agents and physical AI, creating new opportunities to bring intelligence everywhere,” stated AMD’s CEO Dr Lisa Su. “Realising that potential will take the entire industry working together. AMD is partnering across the ecosystem to deliver leadership compute and open platforms that give customers the performance, flexibility and choice to scale AI from the datacentre to the edge.”

AMD has been rolling out big-name customers and partners this week. TelecomTV has already reported on the wide-ranging strategic partnership with AI giant Anthropic, which includes an equity investment of up to $5bn, and an expanded partnership with Microsoft Azure. In addition, the chip vendor is working with Meta to jointly design AMD-based compute stacks for Meta workloads and is partnering with OpenAI to optimise the performance of ChatGPT workloads on AMD Helios racks. It is also collaborating with large AI processor developer Cerebras “to deliver a combined solution of Cerebras ultra-low-latency AI compute and AMD Helios high-throughput rack-scale infrastructure to help improve inference efficiency, scalability and economics for ultra-low-latency inference serving.”

Meanwhile, next-generation, virtualised networking specialists DriveNets and Arrcus have signalled their support for AMD’s pitch to AI infrastructure developers. Virtual routing vendor DriveNets, in which AMD invested earlier this year, has teamed up with the semiconductor giant to develop a reference architecture “that defines a validated design blueprint for building high-performance AI infrastructure using AMD Instinct MI350 series GPUs and DriveNets AI Fabric.” And Arrcus has integrated its ArcOS network operating system with UfiSpace’s Broadcom-based open switch platforms to offer an open, disaggregated “alternative to proprietary networking fabrics” for AI infrastructure deployments based on AMD’s GPUs and CPUs.  

- Ray Le Maistre, Editorial Director, TelecomTV