Mirendil’s Bold $200M Bet on Autonomous AI for Science

A new startup called Mirendil just closed a massive $200 million seed round. This funding values the company at $1 billion, marking one of the largest seed rounds in AI history. The round was led by top investors Andreessen Horowitz and Kleiner Perkins, with NVIDIA also joining as a strategic partner.

Mirendil was founded by four former researchers from Anthropic: Behnam Neyshabur, Harsh Mehta, Shayan Salehian, and Tara Rezaei. The team now includes about 20 engineers and scientists from top AI labs like OpenAI, Google DeepMind, and xAI. Their shared goal is to create AI systems that can work independently to speed up scientific research.

Unlike many startups that launch early products, Mirendil has not revealed any specific tools or revenue yet. They are still in the early stages of development. But their vision is clear: build AI agents that can think, write code, and solve complex problems on their own.

Reimagining Scientific Discovery with AI

Behnam Neyshabur, Mirendil’s CEO, says the core skill scientists and engineers have is deep domain expertise that grows over time. Mirendil aims to replicate that with AI that learns and masters specific scientific fields. Their technology will include neural networks, attention mechanisms, and reinforcement learning sandboxes.

The company wants its AI tools to serve fields like chemistry, medicine, and robotics. They plan to help scientists build customized AI models for their own work. However, Mirendil has not yet announced which fields it will target first or how it will distribute its technology.

Backing from Heavy Hitters and Industry Trends

Andreessen Horowitz called Mirendil’s mission “to democratize frontier AI capabilities that have traditionally remained with a limited number of research laboratories globally.” Mamoon Hamid from a16z added the team has been achieving remarkable results quickly.

NVIDIA’s involvement signals their interest in boosting demand for high-end GPUs, which are essential for training large AI models. This move fits a larger trend of huge investments in AI research startups like Safe Superintelligence and Thinking Machines Lab.

Investors acquired about 20% ownership in Mirendil based on the $1 billion valuation. This $200 million seed funding dwarfs typical AI seed rounds, which usually range between $5 million and $30 million. It shows increased confidence in ambitious AI projects focused on science.

The founders bring strong backgrounds in large-scale machine learning and optimization. Their experience helps them push the boundaries of what AI can do in research. They believe autonomous AI agents will soon transform how complex scientific problems get solved.

Mirendil’s journey is just beginning. As they build out their team and technology, the AI community is watching closely. Their success could change how scientists work and how AI powers discovery across many fields.