AMD Unveils Integrated AI Platform with MI350 Series GPUs The company announced broader access to the AMD Developer Cloud, offering developers and the open-source community access to next-gen compute platforms
You're reading Entrepreneur India, an international franchise of Entrepreneur Media.
At its Advancing AI event in San Jose, AMD revealed a wide-ranging AI strategy with a new suite of hardware, software, and end-to-end infrastructure solutions, reinforcing its position as a key enabler of next-generation AI workloads.
The company unveiled its latest AI accelerators, 'the AMD Instinct MI350 Series' which promise a 4x generational performance leap and are already being rolled out in large-scale cloud deployments, including Oracle Cloud Infrastructure.
AMD Chair and CEO Dr. Lisa Su delivered the keynote alongside executives from Meta, Microsoft, Oracle, and OpenAI, showcasing how AMD's integrated platform spanning GPUs, CPUs, networking, and open software is helping power some of the world's most advanced AI applications. Currently, seven of the ten largest AI customers are deploying AMD Instinct Accelerators, underscoring the company's growing leadership in the space.
A central highlight was the introduction of the AMD Instinct MI350 Series, which includes the MI350X and MI355X GPUs. These are built to drive transformative AI performance and are supported by the 5th Gen AMD EPYC processors and AMD Pensando Pollara NICs, all working together as part of a new open-standards rack-scale infrastructure. Broad availability of this infrastructure is expected in the second half of 2025.
AMD offered a glimpse into the future with its "Helios" AI Rack, a fully integrated AI platform set for release in 2026. Powered by next-gen AMD Instinct MI400 Series GPUs, "Zen 6"-based AMD EPYC "Venice" CPUs, and AMD Pensando "Vulcano" NICs, Helios aims to redefine AI performance standards and support ever-larger and more complex models with efficiency at scale.
On the software front, AMD launched ROCm 7.0, its latest AI software stack. The update brings enhanced support for industry-standard frameworks, expanded hardware compatibility, and new APIs, drivers, and libraries to accelerate AI development.
Additionally, the company announced broader access to the AMD Developer Cloud, offering developers and the open-source community access to next-gen compute platforms.
AMD also emphasised sustainability with a significant energy efficiency milestone. The MI350 Series has surpassed AMD's earlier five-year goal of a 30x energy efficiency gain achieving a 38x improvement in AI training and high-performance computing nodes.
AMD is now setting a new 2030 goal: a 20x increase in rack-scale energy efficiency from 2024 levels, aiming to reduce power consumption by 95 per cent.
