Mirendil lands $100M+ Google Cloud pact for self-improving AI work

AI lab Mirendil has entered a multi-year partnership with Google Cloud to secure computing power for its self-improving AI research, according to TechCrunch. The deal is valued at over $100 million, as confirmed by Mirendil's co-founder and CEO Behnam Neyshabur. That figure is roughly half of the seed funding Mirendil raised at a $1 billion valuation in late June.

Under the agreement, Mirendil gains access to Google's TPUs and Nvidia GPUs, plus managed training clusters for its work on recursive self-improvement - AI systems that iteratively enhance themselves. The startup's co-founders previously worked at Anthropic, a lab also exploring this concept. A few other startups, including Recursive Superintelligence and Ricursive Intelligence, have recently launched with similar goals.

Mirendil aims to automate scientific and AI research, potentially accelerating progress in medicine, biology, and materials science. Neyshabur described the vision as pointing an AI at a problem and having it improve over time, such as tackling Alzheimer's research. Co-founder Harsh Mehta noted that training these models requires matching workloads to the right chips, and Google's variety of accelerators allows flexible and cost-effective assignments.

Google's Amin Vahdat, SVP and chief technologist for AI and infrastructure, said advancement now depends on orchestrating entire systems of intelligence, not just chip performance. Mirendil's software layer reportedly helps customers get more from Google's hardware, while Google gains a strategic partner in frontier recursive AI that could later be offered to enterprise clients.

Source: TechCrunch