NVIDIA and Thinking Machines Lab today announced the establishment of a long-term strategic partnership, planning to deploy at least one gigawatt capacity of NVIDIA's next-generation Vera Rubin systems. This collaboration will support frontier AI model training at Thinking Machines Lab and develop training and inference systems based on NVIDIA architecture. System deployment is expected to begin in 2027. Additionally, NVIDIA will make a strategic investment in Thinking Machines Lab, though the specific amount remains undisclosed, to aid its long-term development. Both parties also commit to jointly designing training and inference systems built on NVIDIA architecture, further expanding the application scope of cutting-edge artificial intelligence and open-source models. Jensen Huang, founder and CEO of NVIDIA, stated that "artificial intelligence is humanity's most powerful tool for knowledge discovery." He emphasized that Thinking Machines Lab brings together world-class teams, and this partnership will realize their grand vision for future AI. Mira Murati, co-founder and CEO of Thinking Machines Lab, noted that NVIDIA technologies form the cornerstone of the entire field, and this collaboration will accelerate the creation of human-shapable AI, thereby reshaping human potential. Building robust, understandable, and collaborative AI systems requires breakthroughs in R&D, design, and infrastructure scale. This strategic partnership aims precisely to lay such foundations, ensuring collectively that this transformative technology can maximize human capabilities. The formation of this alliance marks a critical step forward in advancing large-scale AI applications, promising joint efforts to popularize and advance cutting-edge AI technologies globally.