#United States #Las Vegas #Google Cloud #Thinking Machines #AI Hypercomputer

Thinking Machines Partners with Google Cloud

In an exciting development for AI research, Thinking Machines Lab recently announced its expanded partnership with Google Cloud. At the Cloud Next '26 conference, Google Cloud revealed that it has signed a new agreement to enhance Thinking Machines' AI infrastructure capabilities. This collaboration aims to significantly boost research, platform development, and model training with the advanced AI Hypercomputer technology offered by Google Cloud.

What This Means for Thinking Machines

This new agreement allows Thinking Machines to utilize the A4X Max virtual machines equipped with NVIDIA's Blackwell architecture through Google Cloud. This positions Thinking Machines as one of the leading users of the NVIDIA GB300 NVL72 GPUs, showcasing their commitment to embracing cutting-edge technology. In early trials, Thinking Machines reported a remarkable 2X increase in training and serving speeds with the new A4X Max VMs, particularly when compared to previous generations of GPUs. This remarkable boost in performance is bolstered by Google Cloud’s Jupiter network, which enables swift weight transfers — a critical capability for TML’s reinforcement learning workloads.

Thinking Machines also benefits from its access to Google Cloud's robust suite of integrated tools and services. The lab currently employs various Google services including Google Kubernetes Engine (GKE) for orchestrating large-scale computations, Spanner for managing transactional data, Cluster Director for resource management, as well as Cloud Storage and Anywhere Cache for optimizing storage retrieval. Together, these tools empower Thinking Machines to develop their frontier models and refine their innovative product, Tinker.

Achieving High Efficiency with Integrated Solutions

Thinking Machines has strategically combined Cloud Storage, Spanner for transaction data management, and a custom caching solution at the node level. This optimal infrastructure ensures a seamless training process, supporting production workloads on a global scale without interruption. “By leveraging A4X Max and the AI Hypercomputer integrated stack, Google Cloud has enabled us to operate at record speeds while maintaining the high reliability we require,” stated Myle Ott, Founding Researcher at TML.

This partnership allows Thinking Machines to focus on developing specialized products and tackling unique challenges within the AI landscape, such as enhancing Tinker and advancing their work in reinforcement learning. According to Mark Lohmeyer, Vice President and General Manager of AI and Computing Infrastructure at Google Cloud, the collaboration underscores the potential of AI to revolutionize how organizations leverage technology for operational efficiency and innovation. He remarked, “The team at Thinking Machines Lab is generating very exciting research and product offerings that will help organizations more effectively utilize AI.”

NVIDIA's Role in the Partnership

NVIDIA also plays a crucial role in this collaboration. Ian Buck, Vice President and General Manager of Hyperscale and HPC at NVIDIA, articulated the importance of system-level optimization as model sizes and complexity of reinforcement learning workflows increase. By using the NVIDIA GB300 NVL72 GPUs, the collaboration aims to reduce existing bottlenecks and enhance throughput. Buck emphasized that the integrated stack from Google Cloud, combined with NVIDIA's technology, strengthens their platform further, rendering it both faster and more sophisticated.

A Promising Future

Thinking Machines’ partnership with Google Cloud began in 2025, and this latest agreement marks a significant milestone in their joint efforts to push the boundaries of what is possible with AI technology. As they continue to collaborate, both companies are poised to make strides in research and application that could reshape industries, enhance organizational efficiencies, and set new standards in artificial intelligence.

In summary, the strengthened alliance between Thinking Machines and Google Cloud, powered by NVIDIA's advanced hardware, heralds a new era for AI innovation. This collaboration not only accelerates research and development but also empowers organizations globally as they seek to harness the power of artificial intelligence in their operations.