Pomegra
Toronto startup Biossil has raised $70 million co-led by Founders Fund and OpenAI to acquire abandoned drug candidates and advance them through clinical trials using large language models.
- Biossil raised $70M USD across two rounds: a $22M Founders Fund-led seed in 2024 and a $43M co-led raise in 2025, valuing the company above $100M.
- The company has quietly acquired 10 failed drug molecules over three years, with two in late-stage trials and three pending regulatory approval.
- Therapeutic targets span sickle cell disease, idiopathic pulmonary fibrosis, glioblastoma, breast cancer, and Alzheimer's disease.
Lead
Toronto-based Biossil emerged from three years of stealth in mid-2026 with $70 million in total equity financing, co-led by Founders Fund and OpenAI, to commercialize a pipeline built entirely from drug candidates that other companies abandoned. The startup, co-founded by Anthony Mouchantaf and Dr. Alexander Mosa - both University of Toronto alumni - applies large language models to publicly available clinical data, securities filings, and research literature to identify molecules worth reviving. Two of those molecules are now in late-stage trials; three others are seeking regulatory clearance to enter market.
What Does Biossil Actually Do?
The core premise is straightforward: pharmaceutical companies discard drug candidates constantly, often for commercial or organizational reasons rather than irredeemable scientific failure. Biossil uses OpenAI's large language models to parse the paper trail those candidates leave behind - trial results, regulatory communications, academic publications - and identifies situations where a molecule failed in one context but may succeed in another indication, patient population, or dosing regimen.
Once identified, the company acquires or licenses the molecule directly from the original developer. That means Biossil owns the asset rather than collecting a fee for a software service, which makes it a pharma company with an AI screening layer, not an AI company that happens to know about pharma. That distinction matters for how the business scales and where the risk sits.
Why Did Founders Fund and OpenAI Co-Lead?
The round structure tells part of the story. Founders Fund led the initial $22 million seed in 2024, before OpenAI joined as co-lead on the subsequent $43 million raise. OpenAI taking an equity position in a clinical-stage biotech is notable; the company has been selective about where its investment arm participates in non-AI-native businesses. The involvement suggests OpenAI views drug repurposing as a credible near-term validation case for LLMs applied to structured scientific data.
Additional investors include Modern Capital, Staircase Ventures, and Golden Ventures. The round values Biossil above $100 million USD - a figure that implies the post-seed investors assigned meaningful probability to at least some of the existing pipeline reaching approval.
Is "Rescuing Failed Drugs with AI" a Durable Category?
The strategy has precedent but no established playbook. Drug repurposing as a concept is decades old - thalidomide and sildenafil are the canonical examples of molecules reborn in different indications. What Biossil is betting on is that LLMs can systematically surface those opportunities at a scale and speed that human analysts cannot, cutting the time required to identify and diligence candidates by 70 to 80 percent by some internal estimates.
The risk is in the clinical translation. Identifying a candidate algorithmically does not change the underlying biology; molecules still fail in trials at the same rates they always have. Biossil's pipeline spans five distinct disease areas - sickle cell disease, idiopathic pulmonary fibrosis, glioblastoma, breast cancer, and Alzheimer's disease - which creates diversification but also operational complexity for a company that only recently left stealth.
What Comes Next for the Pipeline?
With two molecules in late-stage trials and three at the regulatory submission stage, Biossil is approaching the period where its AI-sourcing thesis either produces approval data or does not. The company has not disclosed trial timelines or primary endpoints publicly, so the near-term clinical read-outs are undisclosed. Any positive Phase 3 result would represent one of the first commercially significant validations of an LLM-first drug repurposing strategy.
The funding will support continued molecule acquisition alongside those existing programs. The company has indicated it will continue scanning the pool of shelved candidates, which by some pharmaceutical industry estimates numbers in the thousands.
Outlook
Biossil's exit from stealth caps a three-year period of quiet deal-making with a credible investor lineup and a pipeline that is already in human trials. The near-term test is clinical, not financial: whether the molecules that LLMs identified as undervalued actually perform in the indications Biossil has chosen. If even one late-stage program succeeds, the model gains evidence that is difficult to argue with. If the trials disappoint, the company will need to defend the thesis on the basis of discovery throughput alone.
