Think Machines Lab, an artificial intelligence (AI) startup created by former OpenAI executive Mira Murati, has reached a multibillion-dollar agreement with Google. Under the contract, the lab will intensively utilize Google’s AI infrastructure on the cloud and use the most sophisticated computers using the cutting-edge GPUs produced by NVIDIA.
The deal amounts to several billion dollars. In addition to that, it includes utilization of Google’s latest AI systems, which have been created using NVIDIA’s GB300 processors. The new GPUs enable to work with massive datasets in order to train large models. Indeed, modern AI technology relies heavily on scalable computing, which requires powerful hardware.
It should be noted that Google has already tried to build similar partnerships before. The corporation seeks to develop its cloud computing capabilities through collaboration with the developers of AI technologies. Moreover, according to Google’s strategy, cloud computing is combined with other cloud-related services such as storage, Kubernetes Engine, and Spanner (database).
This move fits into a broader trend. AI labs now need massive computing power to stay competitive. Cloud providers like Google offer that scale without requiring companies to build their own infrastructure. In return, they secure long-term customers who use large amounts of computing resources.
Thinking Machines Lab Diversifies Compute with Multi-Gigawatt Google Deal
Earlier this month, another AI company, Anthropic, signed a deal with Google and Broadcom. That agreement involved several gigawatts of tensor processing unit capacity. TPUs are custom chips built by Google for machine learning. They are optimized for training and running AI models. Deals like these show how demand for AI compute continues to rise.
Thinking Machines Lab has not limited itself to one partner. Before this deal, the company worked with NVIDIA in a separate agreement that included an investment. The new deal with Google is not exclusive. This means the start-up can work with multiple cloud providers at the same time. This approach gives it flexibility. It can choose the best tools for each task and avoid relying on a single vendor.
The partnership also shows Google’s intent to work early with emerging AI labs. By forming ties at an early stage, Google can build long-term relationships. If these companies grow, they may continue to rely on Google’s infrastructure.
Mira Murati founded Thinking Machines Lab in February 2025 after leaving her role as OpenAI’s chief technologist. Soon after its launch, the company raised a $2 billion seed round. That funding valued the company at $12 billion. Despite this strong start, the lab kept a low profile for several months.
It stepped into the public eye in October with the launch of Tinker. This tool helps automate the creation of custom frontier AI models. Instead of building models from scratch, users can tailor systems to their specific needs. This can save time and reduce the complexity of AI development.
Google and Thinking Machines Partner to Scale AI Development
The new deal with Google will support the growth of Tinker. Large-scale infrastructure is key for tools like this. Custom model creation requires heavy computation, especially when working with advanced architectures.
The agreement also highlights the type of work Thinking Machines is doing. Google confirmed that its systems will support the start-up’s reinforcement learning workloads. Reinforcement learning is a training method where models learn by trial and error. They improve by receiving feedback based on their actions.
This method has driven major progress in AI. Labs like DeepMind and OpenAI have used it to achieve strong results in areas such as game playing and reasoning tasks. However, reinforcement learning is resource-intensive. It requires repeated simulations and large amounts of data. This makes access to powerful hardware critical.
The magnitude of this deal is a testament to how expensive the research into AI has become. The training of today’s AI involves thousands of GPUs that must be utilized for prolonged periods of time, and only a select few companies can do so. One such important company is the cloud services provider.
The collaboration between Google and Thinking Machines Lab is beneficial to both parties. In exchange for access to their cloud services, the latter gets more powerful resources for the development of their technology. Thus, both get a lot out of it.
This deal highlights an emerging trend within the industry. It is clear that today, the competition for AI also includes the ability to access the necessary infrastructure. This partnership is one step towards gaining such an advantage.
Comments are closed.