Mirendil Raised $1 Billion in a Series A, and Agentic AI Just Got Expensive
Mirendil, a San Francisco agentic AI startup, closed a $1 billion Series A on November 12, co-led by Kleiner Perkins and Clearlake Capital. It was the biggest round of the week.

The Mirendil Series A closed on November 12, and the number is hard to ignore: $1 billion, in a Series A, co-led by Kleiner Perkins and Clearlake Capital. A Series A is supposed to be the round where a young company proves it has something real. A billion dollars is the round where investors decide a company is already a category leader. Mirendil got the second kind of round at the first kind of stage.
The Mirendil Series A, in numbers
Start with the context. The ten biggest disclosed startup rounds worldwide between November 9 and 15 totaled $4.88 billion, and six of them went to AI-focused companies. Mirendil led the list. Behind it sat Mercor with a $750 million Series D and UFORCE with a $700 million Series A. Mirendil alone took more than a fifth of the week's top-ten total.
That is not normal for a Series A. The classic Series A is tens of millions, raised to hire a team and find product-market fit. A $1 billion Series A says the investors are not funding an experiment. They are funding a land grab, and they expect the company to spend like a much older business: hiring at scale, buying compute, and signing customers before rivals can.
The co-leads matter too. Kleiner Perkins is one of the oldest names in venture capital, and Clearlake Capital is a large private equity firm. A venture firm and a private equity firm co-leading a Series A is unusual, and it signals that this round is as much about financial firepower as it is about startup mentorship. Mirendil is being capitalized like an infrastructure company, not a software startup.
What Mirendil actually builds
Mirendil is an agentic AI company based in San Francisco. Agentic AI is the industry's current obsession: systems that do not just answer questions but carry out multi-step tasks on their own, using tools, calling APIs and checking their own work. The pitch is that a company can hand an agent a goal, like reconciling invoices or triaging a support queue, and the agent does the job instead of a person doing it with a chatbot open beside them.
The details of Mirendil's product are still thin in public coverage, which is normal for a company that has just raised its first big round. What is clear is the category. Agentic AI is where the money is going right now: LangChain raised $300 million at a $3 billion valuation in the same month for agent tooling, and the biggest rounds of the last few weeks have clustered around infrastructure for agents, not chatbots.
That clustering is the real story. Investors are no longer funding AI companies that make a cleverer demo. They are funding the plumbing that lets agents run reliably in production: tooling, evaluation, memory, orchestration. Mirendil sits in that plumbing layer, and a billion dollars is the price of building it at the scale enterprises demand.
A week of billion-dollar rounds
Mirendil did not raise in a vacuum. The same week brought Mercor's $750 million Series D for AI infrastructure and UFORCE's $700 million Series A for defense technology. The week before, Fractile raised $720 million at a $7 billion valuation for AI inference chips, and Agile Robots raised $500 million for industrial robotics. The pattern is consistent: the biggest checks are going to companies that sell picks and shovels to the AI boom, not the AI applications themselves.
There is also a valuation lesson here. Boltz raised $110 million for AI drug discovery, and Humyn Labs raised $47 million to film people doing chores so robots can learn them. Those are serious rounds, and they are a tenth or a twentieth of Mirendil's. The gap between the top of the market and the middle is widening, and it is widening fast.
Why investors are paying this much
Three forces are pushing checks up. First, compute is expensive, and agentic systems burn more of it than chatbots because they run many model calls per task. Second, enterprise buyers want vendors with staying power, and a billion-dollar balance sheet is a sales tool in itself. Third, the fear of missing out is real: if agentic AI becomes the next platform shift, the winners will be decided in the next two years, and investors would rather overpay now than watch from the sidelines.
None of that guarantees Mirendil succeeds. A billion dollars buys runway and talent, but it also buys expectations. The company now has to grow into a valuation that was set before it has a long public track record. The history of mega-rounds is mixed: some become category leaders, and some become cautionary tales about burning cash on a market that arrived slower than the pitch deck promised.
What could go wrong
The risks are concrete. Agentic AI still fails in embarrassing ways: agents loop, hallucinate tool calls, or take actions nobody asked for. Enterprise buyers are cautious, and many are still in pilot mode. If adoption stalls, a company capitalized for hypergrowth has to cut hard, and a $1 billion Series A becomes a very public problem.
There is also the competitive angle. Mirendil is not the only company building agent infrastructure, and the field is crowded with well-funded rivals. A billion dollars helps, but it does not buy a moat on its own.
Still, the signal is clear. Investors just put a billion dollars into a Series A for agentic AI, and they did it in a week when the top ten rounds totaled nearly $5 billion. The agentic AI land grab is on, and Mirendil just became one of its best-funded players.


