Atlas Bio raises $12.5 million to simulate drug trials on computers, with no public validation so far

Atlas Bio, a company founded in 2024 and based in Basel, Switzerland, announced total funding of $12.5 million on October 27, 2026. It includes a $10 million seed round led by Creandum, whose partner Katharina Wilhelm joins the board, and $2.5 million in previously unannounced pre-seed funding led by First Momentum Ventures and LEA Partners (Tech.eu).

The product is a foundation model meant to run clinical trials simulated on a computer rather than conducted on patients, what the company calls "in silico" trials. According to the company, it was trained on about 2 million RNA profiles from human tissue, which measure gene activity. It simulates each patient's molecular response within a virtual cohort, and can test several treatments on the same patient profile, something a real trial cannot do (Tech.eu, TechFundingNews).

According to Atlas Bio's presentation, as relayed by Tech.eu, a simulation project would take about 12 weeks, whereas a phase II trial typically takes two to three years. The company positions its tool ahead of phase II, the stage where, it says, about 70% of drug candidates fail and which costs at least $100 million. These figures are the company's own and are not sourced in its announcement. One outside benchmark exists: an analysis by BIO of 2006-2015 data, cited by TechFundingNews, finds that only 30.7% of phase II candidates advance to phase III.

On the commercial side, Atlas Bio says it works with one of the world's ten largest pharmaceutical companies and is in talks with several others. It also claims that a Basel biotech skipped animal studies for one part of a program and went straight to human trials on the strength of its simulations. None of these clients is named, which rules out any outside verification (European Biotechnology Magazine).

The regulatory context favors this type of approach. In the United States, the FDA Modernization Act 2.0, passed in 2022, removed the legal requirement to test on animals before human trials. European Biotechnology Magazine also notes that in September 2026 the US Food and Drug Administration finalized a rule replacing "animal testing" with "nonclinical testing" in its regulations, effective February 4, 2027.

One worked example is circulating in the specialist press. TechFundingNews cites a simulated asthma trial of a bispecific antibody. It reportedly predicted the responses of 78 patients and showed a larger drop in eosinophils, a type of white blood cell, than the standard treatment. This is a result presented by the company, with no scientific publication to back it in the sources consulted.

Atlas Bio is not alone in this field. TechFundingNews mentions Unlearn, which raised $50 million in a Series C. It also cites QuantHealth, which raised $45 million in a Series B and says it has simulated more than 600 trials with predictive accuracy of up to 90%. Mordor Intelligence, a market research firm cited by the same outlet, estimates the simulated clinical trial market at $4.16 billion in 2026. This can be read as a sign that investors are betting on simulation, but it says nothing about how reliable it is.

The stakes may extend beyond the startup. If simulations came to be accepted as evidence, pharmaceutical companies and contract research organizations would lose part of the monopoly conferred by their ability to run long, costly trials. That is our own inference, which the company expresses in its own way by speaking of the "end of phase II." Nothing indicates at this stage that any regulator has accepted a simulation in place of a phase II result.

That is where the main cost of the bet lies, for investors and authorities alike. Atlas Bio itself says the funding will go in part toward deepening the validation of its platform, which means that validation is not complete. None of the sources reviewed reports independent or peer-reviewed validation. TechFundingNews notes that it is unknown whether regulators will accept a simulation in place of a phase II result. It also points out that a systematic review of simulated trials concludes that most models complement clinical data rather than replace it.

A test is coming soon: according to TechFundingNews, the first trial results are expected about six weeks after the announcement, and they will be the key test of the approach.