RESOURCE · AI-NATIVE BIOTECH · October 2026 · work in progress

Same shortage, different labs

Four AI-first drug discovery companies name the same thing as scarce: accuracy in designing molecules. They own very different labs, and they can find out they were wrong at very different speeds.

How to read this. Placement is a reading of what each company’s own people said in interviews, talks and earnings calls. It is not a measurement and has not yet been checked by a human reviewer. Two to three sources per company. Quotes come from machine-made transcripts that have not yet been checked against the audio. Research and commentary, not investment advice.

The picture

Hollow dot = low-confidence reading. Recursion is shown for reference: its executive names a different limit (picking the right target), so it is not in the group.

The table

Company	What they say is scarce	Who owns the lab	How fast they can check their own claim	One line in their words	Sources read
BigHat Biosciences	Design accuracy	Integrated, automated lab (medium)	Days to weeks (high)	“We have a lab that makes these molecules and generates data about them. We have AI tech that can use that lab to make molecules and test its predictions and learn from that.”	Two interviews (YouTube).
Genesis Therapeutics	Design accuracy	Large in-house lab (high)	Not stated in sources (low)	“there's numerous advantages to having in-house wet lab, data generation clearly, dog fooding the models, getting really direct human feedback from real med chemists”	Latent Space podcast (2026) and a 2023 interview.
Absci Corporation	Design accuracy	Small validation lab (low)	Months (high)	“We'll have a 13-week interim readout on that later this year with the 26 efficacy readout, first part of 27.”	Three conference and earnings transcripts.
Manas AI	Design accuracy	No lab described (medium)	Months (medium)	“Right now our focus is mainly on making molecular generators. So we have partnerships that do these other aspects of drug development.”	Two interviews.
Recursion Pharmaceuticals (reference)	Target choice	Integrated, automated lab (high)	Months (medium)	“We do about two point two million experiments a week.”	Three sources plus an earnings call.

Confidence in brackets is the reading’s confidence, not the company’s. Genesis’s sources do not say how fast it can check; the page leaves that as “not stated” rather than guessing.

What stands out

The same sentence, different clocks. Genesis: “if your accuracy is not sufficient, it will therefore not be useful for the downstream things that you care about, which is both potency prediction, but also prospective design.” BigHat describes learning in days: “We wanted a situation where we could do data science and machine learning today, and make predictions tomorrow, and know the truth of those the next day after.” Absci points to a dated clinical readout: “We'll have a 13-week interim readout on that later this year with the 26 efficacy readout, first part of 27.”
Owning a lab is not the same as naming the same shortage. Recursion says it runs a very large lab, yet its executive says “building the best drug isn't generally the number one limitation. It's picking the right target.”
A different arrangement. Manas AI: “Right now our focus is mainly on making molecular generators. So we have partnerships that do these other aspects of drug development.”

Sources

BigHat Biosciences: AI in Action E274, Mark DePristo; Founder Stories, Mark DePristo.
Genesis Therapeutics: Latent Space podcast (2026); Supercharging Discovery with Evan Feinberg (2023).
Absci, Manas AI, Recursion: conference, earnings-call and interview transcripts; page-level references are kept in the project record.

Method: each company’s quoted statements were read together and sorted into a small set of categories by an automated reader, with a confidence level. Companies whose sources conflicted were placed at their most common position. Speaker identity in the transcripts is unconfirmed. This page will be updated as more sources are read and human review is added.