Tokyo, Feb 10, 2026, 02:23 JST
- Takeda signed a multi-year AI drug discovery partnership with privately held Iambic with potential payments above $1.7 billion plus royalties
- Collaboration will start with small-molecule programs in oncology and gastrointestinal and inflammation diseases
- Takeda will get access to Iambic’s NeuralPLexer model for predicting how drug molecules bind to protein targets
Takeda Pharmaceutical (4502.T) has struck a multi-year partnership with privately held Iambic to use artificial intelligence in small-molecule drug discovery, a deal that could exceed $1.7 billion in milestone payments and includes royalties on any product sales. Takeda will also get access to Iambic’s NeuralPLexer software under the agreement. Reuters
The agreement lands as drugmakers spend heavily on AI tools meant to speed early research and cut costs. These systems aim to narrow down which compounds are worth making and testing, an early step that can burn years and cash.
Small molecules are traditional chemical medicines, distinct from protein-based biologic drugs such as antibodies. Iambic has argued that pairing AI predictions with automated lab work can shorten the time it takes to get a compound ready for clinical trials, where most drugs still fail.
Iambic, which has operations in San Diego and Cambridge, Massachusetts, said the work will begin with Takeda programs in oncology and gastrointestinal and inflammation diseases. Chief executive Tom Miller said the company was “excited to partner with their team” to advance drug candidates. Business Wire
Takeda will gain access to NeuralPLexer, which is designed to predict how a drug molecule fits onto a protein target — the kind of molecular “lock and key” problem chemists wrestle with. Miller told Reuters that without a clear view of the target’s structure, it is like trying to “make a sculpture in the dark”.
Iambic has said it can run fast design cycles by combining computation with high-throughput, automated laboratory work. Miller told Reuters the point is not just speed, but making medicines that “couldn’t have been done before”.
Takeda has been pushing to embed AI across research and has previously signed an agreement with Nabla Bio focused on protein-based drugs. Takeda Chief Scientific Officer Christopher Arendt said adding AI to small-molecule work means “you can go faster”, while warning that faster does not matter much if the molecules are low quality.
In its statement, Arendt said the Iambic platform could “de-risk candidate selection” and “improve probability of success” as projects move toward an IND. An IND, short for Investigational New Drug application, is the package filed with regulators before human trials can begin.
In an interview with Fierce Biotech, Miller said Iambic’s pitch is “not about AI hype” and that the company is “showing in patients” its approach can translate beyond computer models. Fierce said Iambic has partnered with groups including Revolution Medicines, Jazz Pharmaceuticals and Lundbeck. Fierce Biotech
The risk is that better predictions do not guarantee better drugs. AI models can still miss toxicity and other failure points that only show up in animals or people, and much of the headline value depends on milestones that may never be reached.
The companies did not disclose the size of any upfront payment. Takeda will pay Iambic royalties — a percentage of net sales — if any products from the collaboration reach the market.
