AMD has launched its next-gen AI chips built for large-scale AI operations at an event in San Jose, California. The company unveiled its Instinct MI400 series scheduled to ship sometime later. AMD CEO Lisa Su took the stage to unveil the new chips. OpeAI CEO Sam Altman was also a part of the event confirming that the company will be integrating these new AMD chips. Altman praised the new AMD AI-focused lineup and expressed his excitement for what lies ahead.
The next-gen AMD MI400 chips are built for large-scale operations and can be embedded into complete server racks. The AMD Instinct MI400 Series GPUs are expected to offer up to 432GB of HBM4 memory, 40 petaflops of FP4 performance and 300 gigabytes per second of scale-out bandwidth. These GPUs are modelled to bring rack-scale AI performance leadership for training massive models and running distributed inference at scale.
The AMD chief says that its AI chips are being used by companies like Meta, Microsoft, OpenAI, Tesla, Oracle, and xAI. The company says that the AI chip market is likely to exceed $500 billion by 2028. AMD has confirmed that it will introduce new chips annually and offer cost-effective solutions for large AI deployment.
To put things into perspective, Nvidia currently holds monopoly in big data center GPUs thanks to its early lead in developing CUDA – an AI software. Nvidia’s Blackwell chips offer integration of 72 GPUs for large scale deployment and AMD’s latest lineup looks to topple this monopoly with the new rack-scale line-up. The latest AMD AI chip line-up will begin shipping in 2026.
After this announcement, AMD’s share rose by 9 percent with analysts claiming that they expect a snapback for AMD’s GPU business in the fourth quarter.
