Product1 publisher2 min readPublished
Basecamp banks $140M on a gene atlas it built with Anthropic and Nvidia
The Series C funds a 28-billion-parameter model trained on more than 100 billion genes. Two of the partners who helped assemble that dataset, Nvidia and Anthropic's joint fund, are also in the round.
The Product Desk · Product desk
What happened
- Basecamp Research said it raised $140 million in a Series C round led by S32, the fund affiliated with Google co-founder Bill Maris.
- More than a dozen other investors joined, among them NATO, Nvidia and the Anthology Fund, a $100 million joint venture between Anthropic and Menlo Ventures.
- The company's model, EDEN, has 28 billion parameters and was trained on the Trillion Gene Atlas, a dataset Basecamp built with Anthropic, Nvidia and other partners.
- Its first application is in vivo cell therapies, with therapeutic DNA delivered using large serine recombinases taken from bacteriophages.
Compiled by The Product DeskSomething wrong?How this is made
Why it matters
- decision A lab weighing access has to price two concrete outputs, the recombinase designs and the immune-response predictions, because a parameter count never appears on a protocol.
- contradiction The dataset's name promises a trillion genes while the company discloses more than 100 billion, so anyone negotiating for exclusive data has to pin down which scale is under contract.
- constraint Anthropic and Nvidia helped assemble the atlas, so Basecamp shares the data advantage it is funded on. That limits how far exclusivity can be pushed in a pharma deal.
The hard part of an in vivo cell therapy is delivery. A human cell holds several double helices about seven feet long and keeps them rolled into tiny coils with significantly reduced surface area, so a designed sequence has to reach a target that is mostly folded away [9]. Basecamp's answer is large serine recombinases, DNA delivery mechanisms derived from bacteriophages [10].
The buyer here is a translational group that already has a target and cannot get its cargo into the cell.
The dataset carries the name Trillion Gene Atlas, and Basecamp describes it as holding information on more than 100 billion genes [6][7]. A trillion is ten times 100 billion [15]. The headline asset is named at one scale and disclosed at another, and a partner buying access should establish which figure the contract covers.
Two of the atlas's named collaborators, Anthropic and Nvidia, also turn up in the money [6][3][4]. That complicates a straightforward data-moat story, because the work of assembling the atlas was shared with partners whose compute and models much of the field already rents. The announcement did not disclose who outside Basecamp can use the atlas, or name a pharmaceutical partner [16].
"We believe the future of medicine lies in reprogramming the body to repair itself," said Glen Gowers, Basecamp's co-founder and chief executive. "We design the models and the medicines to teach it how" [12][13]. The capital goes to accelerating drug development and to signing more pharmaceutical partnerships [14].
A 2x2 separates the two questions a licensee has to answer. One axis is whether a well-funded rival could rebuild an equivalent dataset inside your program's timeline. The other is whether a model output replaces a step your lab already pays for. If a rival could rebuild the data and no output removes a wet-lab step, the 28 billion parameters are the only thing being sold [5]. EDEN has two outputs a lab can check against its own failure log before signing anything: prediction of immune responses to a newly developed therapy, and the design of whole cells equipped with therapeutic cargo [11].
What to watch
- A named pharmaceutical partnership, which Basecamp says it plans to sign, would be the first outside check on EDEN's designs.
- Published access terms for the Trillion Gene Atlas would show whether the data advantage belongs to Basecamp alone.
- A benchmark comparing recombinase-based delivery designed by EDEN against the delivery methods labs use now.