Exscientia

An AI drug-discovery company built around molecule design

Exscientia was one of the companies that treated molecule design itself as a computational problem. In the conventional process, chemists propose compounds a few at a time and each is made and tested in sequence; in the computational version, algorithms generate candidate structures against a target and only a small number of the most promising are synthesised and assayed. Coverage of its combination with Recursion described the two firms as pioneers of AI drug discovery^[1]. What Exscientia brought to the merged business was its design capability together with its own pipeline assets; its most advanced oncology asset was REC-617, a CDK7 inhibitor developed by Exscientia^[1]. CDK7 is a kinase involved in cell-cycle control and transcription, and inhibiting it has been pursued as an oncology strategy. Before the combination, the company's sites included London and Oxford, both of which the combined company inherited^[1].

The combination with Recursion

The deal was announced in August 2024. Reuters and CNBC reported its value at 688 million dollars, although the companies never formally disclosed a figure^[1]. It completed on 20 November 2024, when Exscientia became a wholly-owned subsidiary of the combined company^[1].

Leadership, sites and cost base

Dave Hallett moved from Exscientia to become chief scientific officer of the combined company, while Chris Gibson remained chief executive^[1]. The combined company was headquartered in Salt Lake City, with offices in Toronto, Montreal, Milpitas, New York, London and Oxford, the last two inherited from Exscientia^[1]. Headcount fell to roughly 800 from about 900 before the combination, and the company expected around 100 million dollars of annual savings from the integration^[1]. It reported about 576.6 million dollars in cash and equivalents as of 30 September 2024^[1].

What the combined company does

Recursion supplied the experimental end of the business. It describes its automated laboratories as running up to 2 million experiments a week and as having produced a proprietary dataset of more than 50 petabytes^[2]. Its platform, Recursion OS, is described as spanning target identification through to clinical trial enrolment^[2]. On the research side, the combined company reported more than 10 clinical and preclinical programmes, 10 programmes in advanced discovery and more than 10 partnered with biopharma^[1].

Where it sits in the field

The transaction brought together two companies associated with the first wave of AI drug discovery^[1]. Much of the case for combinations of this kind rests on scale: experiments are the expensive part of drug discovery, and a platform that runs them automatically and records every result is meant to make each later programme cheaper to start. Recursion's own figures for its laboratories and dataset are the most concrete statement of that argument^[2]. The financial side was equally explicit: the reduction from about 900 staff to roughly 800, and the expected 100 million dollars of annual savings, were presented as integration effects^[1]. On the research side, the numbers published with the deal — more than 10 clinical and preclinical programmes, 10 in advanced discovery and more than 10 partnered with biopharma — are the reference point against which the combined company's progress will be read, with REC-617 named as the furthest-along oncology asset^[1].