Topic overview
In brief
- Mirendil has signed a multi-year partnership with Google Cloud to enhance its compute infrastructure for AI research.
- The deal, valued at over $100 million, provides access to advanced computing resources necessary for developing self-improving AI.
- This partnership positions Mirendil to automate significant scientific research processes and advance AI technology.
Summary
In a significant development in the AI sector, the startup Mirendil has entered into a multi-year partnership with Google Cloud, valued at over $100 million. This agreement aims to enhance Mirendil's compute infrastructure, which is crucial for its research into self-improving AI systems. The partnership reflects a growing trend where cloud service providers are making substantial commitments to startups, while AI companies are aggressively securing compute resources to support their scaling efforts. Mirendil's co-founder and CEO, Behnam Neyshabur, indicated that this deal represents a substantial portion of the funding raised by the company, which was valued at $1 billion in late June.
The collaboration will provide Mirendil with access to advanced computing resources, including Google’s Tensor Processing Units (TPUs) and Nvidia GPUs, as well as managed training clusters. These resources are essential for developing self-improving AI, a concept that involves AI systems that can iteratively enhance their own capabilities. Neyshabur expressed optimism about the potential of self-improving AI to automate significant portions of scientific research, particularly in fields such as medicine and biology. He emphasized the ability of AI to learn and improve over time, akin to human scientists.