Key takeaways
- OpenAI signs six-year agreement for 457 MW capacity
- AMD MI455X GPUs will power multiple US facilities
- Deal expands AMD's presence in Nvidia-dominated AI infrastructure
OpenAI has signed a six-year agreement with AI cloud provider TensorWave for 457 megawatts of AMD-powered data center capacity, marking another major expansion of its computing infrastructure as competition with Nvidia intensifies.
According to TensorWave's November 16 announcement, the agreement covers dedicated computing capacity across multiple US locations equipped with AMD Instinct MI455X graphics processors.
The deal strengthens AMD's position in the AI hardware market as OpenAI diversifies its suppliers beyond Nvidia. Neither company disclosed the financial value of the agreement.
OpenAI Secures 457 MW of AMD-Powered Computing Capacity
TensorWave, an AI cloud provider backed by AMD Ventures, will deliver infrastructure designed for demanding artificial intelligence workloads across several US data centers.
The agreement involves AMD's Instinct MI455X accelerators, which are built for large-scale AI training and inference using the company's Helios rack-scale platform.
According to AMD's official product specifications, the MI455X uses its fifth-generation CDNA architecture and targets high-performance AI computing.
The 457 MW figure represents contracted IT power capacity rather than electricity already being consumed by operating GPUs. Actual deployment will depend on infrastructure construction, hardware availability and the delivery schedule.
The announcement adds another significant project to the expanding AI data center financing market, where cloud providers increasingly build infrastructure around long-term customer commitments.
AMD Challenges Nvidia With OpenAI's Expanding Infrastructure
The TensorWave agreement follows the AMD–OpenAI strategic partnership announced in October 2025, which outlined plans to deploy 6 gigawatts of AMD GPUs over several years.
That agreement also granted OpenAI warrants to acquire up to 160 million AMD shares, subject to deployment milestones, share-price conditions and other requirements.
The latest contract gives AMD another commercial foothold as investors reassess Nvidia's dominance in AI chips.
AMD has been expanding its Instinct accelerator portfolio and Helios computing systems to compete more directly with Nvidia's integrated hardware and software ecosystem.
Recent AMD stock volatility has also highlighted growing investor attention to alternative AI chip suppliers.
However, the OpenAI agreement does not mean Nvidia is being replaced. OpenAI separately announced a strategic partnership targeting at least 10 gigawatts of Nvidia-powered systems, although that arrangement was initially outlined through a letter of intent.
AI Infrastructure Spending Remains Under Scrutiny
The agreement arrives as investors question how AI developers and cloud providers will finance enormous computing commitments.
Long-term infrastructure contracts can support project financing, but they also leave providers exposed to construction expenses, electricity requirements and future utilization rates.
Those concerns have fueled scrutiny of AI infrastructure borrowing and the ability of customers to sustain spending as computing capacity expands.