Thinking Machines Lab, an AI research company founded by former OpenAI CTO Mira Murati, announced plans to deploy large-scale AI infrastructure built on NVIDIA’s accelerated computing platform to support development of next-generation AI models.
The company said the infrastructure will be designed to support frontier AI training workloads requiring extremely large GPU clusters and high-performance networking. The systems will incorporate NVIDIA’s accelerated computing architecture and are intended to enable large-scale distributed training environments used to develop advanced AI models.
The deployment reflects the growing infrastructure requirements of frontier AI development as model sizes and training datasets continue to expand. Training these models requires massive compute clusters capable of coordinating thousands of GPUs operating simultaneously across tightly coupled networking fabrics.
Key Points
• Thinking Machines Lab plans large-scale AI training infrastructure
• Systems expected to be built on NVIDIA accelerated computing platforms
• Infrastructure designed for frontier AI model training
• Deployment will support distributed training across large GPU clusters
• Reflects increasing infrastructure requirements for advanced AI models
“Advancing the frontier of artificial intelligence requires a new generation of infrastructure designed specifically for large-scale AI workloads,” said Mira Murati, founder and CEO of Thinking Machines Lab. “Our goal is to build systems that allow researchers and developers to train increasingly capable AI models while maintaining efficiency and reliability at unprecedented scales.”
