Nvidia still leads AI chip performance, but the benchmark contest has become more meaningful as inference workloads grow in cost and importance. MLPerf Inference v6.0 drew 24 submitters, and AMD’s MI355X delivered credible results, while buyers increasingly evaluate tokens per second, latency, power use, rack-level throughput, software support and availability.
Nvidia’s position remains strong. Its GB300 NVL72 rack-scale system beat the GB200 platform by 45% on DeepSeek R1 inference tests and led workloads including Llama 3.1 405B, Llama 3.1 8B and Whisper. The company also highlighted upgraded tensor cores, NVFP4 quantization and a 130 TB/s NVLink fabric across the 72-GPU rack.
AMD is gaining leverage through major customer commitments. OpenAI’s October 2025 agreement covers 6 gigawatts of AMD Instinct GPUs, beginning with a 1 gigawatt MI450 deployment in the second half of 2026, while Meta followed with its own 6 gigawatt AMD deal. Meta’s arrangement is worth more than $100 billion and includes warrants that could give Meta up to 10% of AMD if milestones are met.
Custom silicon adds more pressure. OpenAI announced a 10 gigawatt custom accelerator agreement with Broadcom, with deployments beginning in the second half of 2026 and running through 2029. Broadcom, Apollo and Blackstone also launched a $35 billion AI XPV Platform intended to support more than 20 gigawatts of compute capacity through 2028.
