Mirendil is in advanced discussions on a new financing that would peg the company's value at $5 billion, including the new funding, said people with knowledge of the talks. Those people say Kleiner Perkins is in line to lead, with the round potentially delivering as much as $1 billion in new capital. Andreessen Horowitz is also weighing participation. The same two firms led a $200 million seed only three months ago that valued Mirendil at $1 billion.
Started earlier this year, Mirendil was founded by former Anthropic PBC researchers and now has more than 20 employees. The company aims to make broadly accessible AI models that can largely teach and refine themselves with minimal human input, a strategy often called recursive self-improvement. Some see that path as a catalyst for much faster progress, while others worry it could introduce major security risks if achieved.
The fresh funding would support Mirendil's first model release and the training of subsequent systems. The company is planning to roll out a frontier model geared to help with engineering and research tasks at the start of next year, according to two people.
Running and training AI systems is expensive, and Mirendil possesses far less money and infrastructure than giants like OpenAI and Anthropic. To keep costs in check, it is pursuing a cheaper, more automated approach to model development. Last month Mirendil announced a partnership with Google that includes using Google Cloud TPUs to scale its model building. It also relies on Nvidia Corp.'s compute resources, and the chipmaker is one of its investors.
Investor interest in neo-labs - companies prioritizing ambitious research over quick commercialization - remains strong. Others raising money include Thinking Machines Lab, started by the person who previously served as chief technology officer at OpenAI, and Periodic Labs, which focuses on scientific discovery. There is also growing momentum among software startups to build their own in-house models as open-source AI improves.
Recent weeks have brought louder warnings from both insiders and outsiders that AI could eventually slip beyond human control, raising significant security concerns. Those worries are fueling calls to slow development or to establish global guardrails for building the technology. Mirendil has told stakeholders that a key part of its mission is to get advanced AI into more hands while solving the technical challenges of robust safety.
Big headlines can feel urgent, but a steady strategy protects long term goals. Join Briefs Finance CEO Jaspreet Singh on September 29th for a FREE live investor workshop, How to Profit From A Dollar That's Losing its Value, where he shows how we're spotting investment opportunities as the dollar falls. Save your spot.
Mirendil's website puts it plainly: "Today, any lab trying to use AI in drug discovery, chemistry, biology, or robotics must also become a frontier AI lab - a process that is expensive and requires expertise concentrated in a small number of labs," and, " Our goal is to democratize frontier AI R&D and make it widely accessible." The company declined to comment.
For your money, here is the takeaway: serious capital is still flowing to labs chasing big leaps in capability, not just quick revenue. If Mirendil hits its near-term model milestones while keeping costs and safeguards in check, expect more startups to follow the same build-your-own-model playbook.
When news arrives, thoughtful planning helps you preserve and grow your savings. Our CEO Jaspreet Singh is hosting a FREE live investor workshop, How to Profit From A Dollar That's Losing its Value, on September 29th. Sign up free to join him live.
