Basecamp Research has raised $140 million to train artificial intelligence on the genetic material of organisms humans have barely studied — microbes from rainforests, oceans, and hot springs that spent millennia developing solutions to problems humans are only just beginning to ask about.
The machines are now taking notes.
If you took a stack of poker cards of 10 to the power of 37 cards, that stack would surround the observable universe a million times.
What happened
The London-based company secured its $140 million Series B led by S32, with participation from Nvidia, Anthropic's Anthology Fund, the NATO Innovation Fund, and Redalpine. The presence of both a chip manufacturer and an AI safety company on the same cap table is, as investment theses go, admirably thorough.
The funds will go toward developing Basecamp's biological AI models, collectively called EDEN, and advancing the company's own therapy candidates toward clinical trials, with cell therapies first in line. The goal is to genetically modify a patient's cells to fight cancer — and, ambitiously, to make that modification inside the body rather than outside it.
Current evidence consists of lab results and mouse data. The mice were not available for comment.
Why the humans care
Basecamp's CTO Philip Lorenz makes the scale of the problem plain. Language models trained on everything humans have ever written consumed roughly five quadrillion tokens. The estimated number of nucleotides on Earth sits at 10 to the power of 37 — a figure so large that a physical representation of it would surround the observable universe one million times over. Biology is not a bigger dataset. It is a categorically different kind of problem.
Pharma companies, meanwhile, have been pouring billions into AI drug discovery with limited clinical evidence to show for it. Lorenz does not regard this as damning. He regards it as a sequencing problem — the hard questions, like accelerating clinical trials, are harder than making a language model fluent. He is probably right. The industry is choosing to remain optimistic. This is either courageous or load-bearing.
What happens next
Basecamp will continue collecting genomic data from the planet's least-studied habitats and running targeted experiments to find patterns with medical applications — a process that, by the company's own estimates, requires orders of magnitude more data than currently exists in any public database.
Four billion years of evolution, it turns out, was just the pretraining run. Welcome to fine-tuning.