Mirendil Raises $200M at $1B Valuation With No Product Yet

Published by James Harris on June 25, 2026June 25, 2026

A team of 20 researchers, most recruited from Anthropic, OpenAI, Google DeepMind, and xAI, has closed a $200 million seed round at a $1 billion valuation before shipping a single commercial product. Andreessen Horowitz and Kleiner Perkins co-led the round, with NVIDIA participating alongside other investors.

The sharpest context the source article glosses over is what this round implies about ownership economics at the seed stage. Two hundred million dollars at a $1 billion post-money valuation means incoming investors bought roughly 20% of a company with no revenue, no product, and a stated mission to automate AI research and development itself. That is a price structure historically associated with Series B or C rounds, compressed into a seed check. The comparison that matters here is Safe Superintelligence, the lab Ilya Sutskever launched after leaving OpenAI, which raised approximately $1 billion at a valuation near $5 billion in 2024. Mira Murati’s Thinking Machines Lab secured $2 billion in commitments despite disclosing almost nothing about its product direction. Mirendil’s round is smaller than both, but the pattern is identical: frontier pedigree substitutes for traction, and investors are pricing the team rather than the business.

That is not a criticism. It is a structural observation about where AI capital is concentrating right now, and it tells you more about the funding environment than about Mirendil specifically.

Co-founder and CEO Behnam Neyshabur, formerly of Anthropic, has described the company’s aim as building AI systems that can accumulate deep domain expertise the way a scientist or engineer does, then offering those capabilities to third parties. Kleiner Perkins’ Mamoon Hamid characterized the team’s early results as remarkable, and a16z’s Matt Bornstein said the team is working on a major hyperscale problem in AI without elaborating further. NVIDIA’s participation is consistent with its broader pattern of investing across the AI startup ecosystem, where early stakes in compute-intensive labs serve both financial and strategic purposes. For NVIDIA, every large AI lab that scales is also a large GPU customer.

The broader signal here is that the window for raising at these valuations without product evidence appears to still be open, but it is narrowing to a specific profile: researchers with direct frontier lab credentials, working on foundational rather than applied problems, backed by top-tier lead investors who can credibly signal quality to the market. Mirendil fits that profile cleanly. Whether the underlying research produces something commercially meaningful is a question the round does not answer and, for now, investors are not requiring it to.