Mirendil, an AI startup, has secured a partnership worth over $100 million with Google Cloud to expand its compute infrastructure for research into self-improving AI systems. The deal, exclusively learned by TechCrunch, positions Mirendil to pursue systems that can autonomously enhance their capabilities while accelerating scientific discovery. The timing is strategic for Google, which is carving out its own strategy by backing external research teams as Microsoft deepens its OpenAI alliance and Amazon invests in Anthropic. Self-improving AI is described as the next frontier beyond large language models, with Mirendil chasing algorithms that can iteratively refine themselves, potentially compressing years of AI development into months and unlocking breakthroughs in drug discovery, climate modeling, and materials science. The $100 million-plus commitment gives Mirendil access to TPU clusters and GPU infrastructure. The partnership structure mirrors deals Google has struck with other AI labs, combining cloud credits with technical collaboration and early access to new hardware. Mirendil focuses on scientific applications for researchers in academia and pharmaceutical companies, rather than general-purpose chatbots.