AI startup Mirendil raises $200M to boost AI development

The biggest names in artificial intelligence operate with a quiet, shared belief: the most efficient path to smarter models is to let AI help design itself. They run this self-improving loop internally, and their terms of service explicitly prevent outsiders from doing the same. Now, two former researchers from one of those labs have raised $200 million to break that lock.

The startup is called Mirendil, and on June 24, it announced what stands as one of the largest seed rounds the industry has ever seen. At a $1 billion valuation, the company has no product to sell yet. The round was co-led by Andreessen Horowitz and Kleiner Perkins, with Nvidia also participating.

Mirendil’s founders are Behnam Neyshabur and Harsh Mehta. They met at Google in 2019, moved to Anthropic in late 2024, and left in December 2025, shortly after the launch of Claude Opus 4.5. Neyshabur, now chief executive, previously spent more than five years at Alphabet co-leading reasoning research for Gemini.

Plenty of alumni from top labs have started their own companies. Mirendil is aiming at a different layer entirely. The goal is to build an AI that does the work of an AI researcher: designing experiments, searching for optimal settings, evaluating models, and running the next round of training. The plan is to package that capability as a platform other organizations can use on their own problems.

The more significant shift is who that platform is for. Neyshabur describes the concept as “AI for AI for science,” not simply AI for science. A university biology lab, for instance, could use it to build a drug-target model without needing a dedicated machine-learning team. He points to a model that predicts a person’s risk of Alzheimer’s as the kind of thing a customer might create. The pitch is that work which currently takes labs months could be compressed into days.

What gives this thesis its edge is the lock it tries to pick. As of May, Anthropic stated that its Claude model wrote more than 80% of the company’s own code. Yet its terms of service forbid using its tools to build competing services. Anthropic told the Wall Street Journal that the policy is standard among model providers and helps keep frontier AI out of foreign adversaries’ hands.

That gap is the business opportunity. As Matt Bornstein of Andreessen Horowitz told the Journal, the labs are being “rational economic actors” when they deny customers the means to supercharge their own models. “Structurally, there has to be an independent company,” he said. Mirendil wants to be that company.

The technical name for this loop is recursive self-improvement, and it remains contested territory. Anthropic has flagged it as a potential danger, theorizing that a model rewriting its own code without oversight could slip beyond human control. The founders disagree. They call it the “shortest path” to faster science and argue it is a problem that can be supervised rather than avoided.

That argument lands at a tense moment for Anthropic itself. The company recently pulled access to its most powerful Mythos and Fable models after the Trump administration imposed export controls. The same week, critics accused it of quietly degrading answers about AI development. Into that backdrop steps a startup whose entire reason to exist is handing that capability to others.

Mirendil’s team is small and senior, running on about 20 researchers and engineers drawn from Anthropic, xAI, Google DeepMind, and OpenAI. The founding group also includes Shayan Salehian, an early member of xAI, and Tara Rezaei, a 23-year-old MIT graduate. There is a neat irony in the lineup. Mehta built the first version of Anthropic’s internal AI-research platform, at times as a team of one. Now he is rebuilding that idea to sell it. The Information first reported some details of the round.

The valuation makes sense only against the flood of capital around it. AI took close to half of all global venture funding in 2025, some $202 billion, up more than 75% on the year, according to Crunchbase. The AI infrastructure market alone ended 2025 near $337 billion in revenue and is forecast to reach $1.2 trillion by 2030.

Mirendil also sits inside a specific cluster of lab spinouts, and the comparison flatters it. Ilya Sutskever’s Safe Superintelligence has raised $6 billion at a $32 billion valuation. Mira Murati’s Thinking Machines Lab took $2 billion at $12 billion. Both, like Mirendil, launched with no shipped product, on the strength of their founders alone.

The closer rhyme is Periodic Labs, another Andreessen Horowitz bet that raised $200 million to aim AI at materials science. Mirendil is pitching the layer beneath that: the research-automation engine such companies would themselves run on. It is a harder thesis to prove, and a bigger prize if it holds. Venture firms have poured money into the field for two years, and this is their next structural wager.

There is an ideological pull, too. The founders talk about prying AI research out of a few labs and handing it to thousands, the same democratizing argument that has followed every open challenge to Silicon Valley’s frontier. Whether that vision survives contact with a real product is another matter, and it lands amid a steady run of nine-figure AI infrastructure rounds.

For now, Mirendil has a name from Tolkien, a billion-dollar valuation, and a model and product it says will arrive in the coming months. If AI can truly automate its own research, the advantages today’s labs hold, thousands of staff and years of accumulated knowledge, start to look less permanent. Whether Mirendil is on time or five years early is exactly what the $200 million is there to find out.