AMD MI350: 35X Faster AI Inferencing
- — AMD (Advanced Micro devices) has launched its extensive AI platform, featuring an open, scalable rack-scale AI infrastructure built on industry standards.
- The new AMD Instinct MI350 Series accelerators promise a fourfold increase in AI compute speed and a 35-fold increase in inferencing capabilities compared to previous generations, according to...
- OpenAI CEO Sam Altman, appearing onstage with Su, expressed excitement about the MI350 specifications and appreciation for AMD's responsiveness to his company's feedback.
AMD just unveiled its new AI platform, highlighted by the groundbreaking Instinct MI350 Series, and it’s shaking up the industry. This innovative platform promises a staggering 35X increase in AI inferencing capabilities, with a fourfold surge in AI compute speed, directly challenging Nvidia’s current market dominance. News Directory 3 has the scoop on AMD’s open, scalable rack-scale AI infrastructure, featuring the latest accelerators, and the company’s strategic moves to foster open collaboration. With deployments already underway in hyperscale environments, AMD is poised to capture meaningful market share. Discover what’s next as AMD continues to expand its AI ecosystem beyond 2027!
AMD Unveils New AI Platform, Aims to Challenge Nvidia’s Dominance
Updated June 12, 2025
Santa Clara, Calif. — AMD (Advanced Micro devices) has launched its extensive AI platform, featuring an open, scalable rack-scale AI infrastructure built on industry standards. The announcement, made at the company’s annual Advancing AI event, signals AMD’s intent to compete directly with Nvidia in the rapidly expanding artificial intelligence market.
The new AMD Instinct MI350 Series accelerators promise a fourfold increase in AI compute speed and a 35-fold increase in inferencing capabilities compared to previous generations, according to AMD. CEO Lisa Su emphasized the importance of open collaboration in shaping the future of AI, contrasting it with closed systems.
OpenAI CEO Sam Altman, appearing onstage with Su, expressed excitement about the MI350 specifications and appreciation for AMD’s responsiveness to his company’s feedback.
AMD’s rack-scale AI infrastructure,incorporating Instinct MI350 Series accelerators,5th Gen AMD Epyc processors,and AMD Pensando Pollara network interface cards,is already being deployed in hyperscale environments like Oracle Cloud Infrastructure (OCI). Broader availability is expected in the second half of 2025. The company also previewed its next-generation AI rack, Helios, built on future MI400 Series GPUs, Zen 6-based Epyc Venice CPUs, and Pensando Vulcano NICs.
Creative Strategies analyst ben Bajarin noted AMD’s potential to target a different customer base than Nvidia, particularly “neocloud” opportunities, tier two and three clouds, and on-premise enterprise deployments.Bajarin added that AMD is well-positioned to compete as the market shifts to inference, offering a compelling total cost of ownership (TCO) proposition for customers were nvidia’s solutions might be overkill.
The latest version of AMD’s open-source AI software stack, ROCm 7, is designed to meet the growing demands of generative AI and high-performance computing workloads. ROCm 7 features improved support for industry-standard frameworks, expanded hardware compatibility, and new development tools to accelerate AI development and deployment.
AMD also announced the broad availability of the AMD Developer Cloud, providing developers and open-source communities with a managed cloud environment for rapid AI development. Strategic collaborations with companies like Hugging Face, OpenAI, and Grok underscore the power of co-developed, open solutions.
Customers like Meta and Oracle are already leveraging AMD AI solutions. Meta is deploying Instinct MI300X for Llama 3 and Llama 4 inference, while OCI is adopting AMD’s open rack-scale AI infrastructure with Instinct MI355X GPUs to offer zettascale AI clusters.
Microsoft announced that Instinct MI300X is powering both proprietary and open-source models in production on Azure. Other companies, including HUMAIN and Cohere, are also partnering with AMD to build AI infrastructure and deploy models on Instinct MI300X.
Red Hat highlighted its expanded collaboration with AMD to enable production-ready AI environments, with AMD Instinct GPUs on Red Hat OpenShift AI delivering powerful AI processing across hybrid cloud environments.
What’s next
AMD plans to continue expanding its AI platform with next-generation hardware and software, focusing on open standards and collaboration to drive innovation in the AI industry. The company’s roadmap extends beyond 2027,with a focus on rack-scale AI performance and energy efficiency.