Last updated: September 29, 2026
The ~87% figure depends on the funding tracker’s methodology and should not be interpreted as a complete measure of all AI drug-discovery investment.
Six companies, six very different kinds of evidence. This guide separates what has been shown in patients from what has only been announced.
If you want a broader primer on how to judge AI performance in health, see our review of AI Symptom Checkers: How Accurate Are They? What the Evidence Shows.
Quick Answer
Why 2026 Is a Test Year for AI Drug Discovery
For about a decade, AI drug discovery has been described mostly in terms of speed: faster target identification, fewer molecules to synthesize, shorter time to a clinical candidate. Those claims are now being checked against clinical data.
A survey presented at ASCO in 2026, as summarized by IntuitionLabs, counted 117 AI-enabled therapeutic assets from 63 companies that had entered human trials. Of those, 60 had completed Phase 1 and only 8 had completed Phase 2. In other words, most of the field is still early. Phase 3 is where drugs generally succeed or fail, and very few AI-derived programs have reached it.
AI-enabled assets by development milestone
There is also a funding caveat. The same source cites a tracker estimating that Isomorphic Labs’ single $2.1 billion round accounted for roughly 87% of disclosed AI drug-discovery venture funding tracked in the first half of 2026. That figure depends on the tracker’s methodology and is not a complete measure of all investment in the field, but it shows how a record headline number can say more about one company than about the sector.
How We Selected These Companies (Methodology)
We did not rank companies by valuation or by how much attention they receive. We used four criteria, in this order:
- Clinical evidence: peer-reviewed or registry-listed human data comes first.
- Stage of development: how far a program has moved through trials, and whether that is documented by the company in a filing or press release.
- External validation: paying partners, independent investors, or public-market scrutiny.
- Transparency: whether the company discloses enough (trial identifiers, cash position, timelines) for a reader to check its claims.
We relied on company filings and press releases for facts about the company itself, and on trade press or third-party surveys for sector-wide numbers. We treated statements from the companies as claims, not as independent findings, and we say so where it matters. Prices, valuations and deal terms are given only where they were publicly disclosed.
The Companies at a Glance
| Company | AI role | Clinical stage | Evidence level | Main question |
|---|---|---|---|---|
| Insilico | Target + molecule discovery | Phase 3 | Peer-reviewed clinical data | Does Phase IIa translate to Phase 3? |
| Recursion | Biology/drug discovery platform | Phase 2 | Company-reported | Can the early signal be replicated? |
| Generate:Biomedicines | AI-engineered antibody | Phase 3 | Early clinical + company data | Does engineering translate into better outcomes? |
| Isomorphic Labs | AI drug design | Preclinical / no publicly disclosed human data identified | Partnerships + funding | When will human data appear? |
| Chai Discovery | Molecular/antibody design | Discovery | Platform + partnerships | Will designs reach clinical development? |
| Xaira | Biology/virtual-cell AI | Preclinical | Platform/data | When will a program enter the clinic? |
The six companies do not have the same level of evidence: the table is ordered by company, not by strength of evidence.
The Companies in Detail
1. Insilico Medicine: A Leading Clinical-Stage AI Drug Program
What it is. A biotechnology company with headquarters in Cambridge, Massachusetts and operations in Hong Kong, built around a generative AI platform called Pharma.AI. It listed on the Hong Kong Stock Exchange on December 30, 2025 (stock code 3696).
Why it is on the list. Its lead drug, rentosertib (formerly ISM001-055), is a TNIK inhibitor for idiopathic pulmonary fibrosis (IPF). The published clinical research describes both the target and molecule as products of a generative-AI discovery process. In June 2025, Phase IIa results were published in Nature Medicine (trial NCT05938920). The randomized, double-blind, placebo-controlled study enrolled 71 patients and evaluated three active dosing arms over 12 weeks. The study reported manageable safety and a dose-dependent signal in forced vital capacity, while the authors emphasized the need for larger and longer studies.
On September 10, 2026, Insilico announced that the first patient had been dosed in GENESIS-IPF-3, a Phase 3 study designed to enroll 320 participants across 47 centers in China and follow patients for 52 weeks. The drug remains investigational and has not been approved by a regulatory authority.
What to keep in mind. Phase IIa results do not establish that a drug will ultimately be effective or safe in a larger population. The Phase 3 trial is the next major test of whether the earlier findings translate into a longer, larger study. Insilico has also published exploratory work on biological-age markers from the Phase IIa samples; those findings should be treated separately from the trial’s clinical endpoints.
2. Recursion: A Large Pipeline, With Early and Mixed Signals
What it is. A Nasdaq-listed company (RXRX) based in Salt Lake City that describes its approach as an AI-enabled «operating system» for biology, combining large-scale cell imaging with machine learning.
Why it is on the list. Its most advanced wholly owned program in the latest report we reviewed is REC-4881, a MEK1/2 inhibitor for familial adenomatous polyposis (FAP), a genetic condition with no approved drug therapy. In the Phase 2 part of the TUPELO study, the company reported that 75% of evaluable patients had a reduction in total polyp burden, with a median reduction of 43% after 12 weeks in 12 patients. Safety was consistent with what is known for MEK inhibitors; grade 3 treatment-related events occurred in 15.8% of patients and none reached grade 4 or higher. The company said in May 2026 that it had begun discussions with the FDA about a possible registrational study design, with an update expected in the second half of 2026.
The same report describes early-phase programs in oncology, including REC-617 (CDK7), REC-1245 (RBM39) and REC-4539 (LSD1).
What to keep in mind. Twelve patients is a small sample and the data are company-reported. Recursion has also pruned parts of its pipeline in the past, which is normal in drug development but relevant when reading platform claims. If you follow this company, the FDA update expected this half is the item to check.
3. Generate:Biomedicines: AI-Engineered Antibody in Phase 3
What it is. A protein-therapeutics company that went public on Nasdaq (GENB) on February 27, 2026.
Why it is on the list. Its lead candidate, GB-0895, is a long-acting antibody against TSLP, a signaling molecule involved in airway inflammation. Two global Phase 3 trials, SOLAIRIA-1 and SOLAIRIA-2, are evaluating it in about 1,600 adults and adolescents with severe asthma that remains inadequately controlled. The first patient was dosed in January 2026, the treatment period is 52 weeks, and the goal is dosing about twice a year. Company-reported Phase 1 data in mild-to-moderate asthma (96 participants) described a wide tolerated dose range and a half-life of roughly 89 days.
What to keep in mind. GB-0895 targets a biological pathway that already has an approved drug (tezepelumab). The company describes the antibody as engineered with AI to improve properties such as affinity and duration of action. That is a meaningful use of AI, but it differs from discovering a new target from scratch. Full enrollment is not expected until 2027–2028, so efficacy results are still some way off.
4. Isomorphic Labs: Major Funding, but No Clinical Data Disclosed
What it is. A London-based company founded as a spinout from Alphabet’s DeepMind and led by CEO Sir Demis Hassabis. It develops an AI drug-design engine called IsoDDE.
Why it is on the list. On May 12, 2026, Isomorphic Labs announced a $2.1 billion Series B led by Thrive Capital. The company has also disclosed collaborations with Novartis, Eli Lilly and Johnson & Johnson. According to the company, the new funding is intended to scale its AI drug-design engine and advance its therapeutic programs toward the clinic.
What to keep in mind. The funding and pharmaceutical collaborations demonstrate substantial commercial interest, but they are not substitutes for human clinical data. As of this review, we found no publicly disclosed clinical trial results for an Isomorphic Labs internal drug candidate. Any future clinical milestone should therefore be assessed separately from the company’s financing and partnership announcements.
5. Chai Discovery: Expanding Partnerships and Design Models
What it is. A San Francisco company building generative AI models and computer-aided design tools for molecular discovery, including antibody design.
Why it is on the list. During 2026, Chai expanded its commercial activity, including a $400 million Series C announced in July 2026. It also announced collaborations or agreements involving Eli Lilly, Novartis, Pfizer, argenx and Bristol Myers Squibb during 2026. These partnerships give pharmaceutical companies access to Chai’s AI design capabilities, but they do not by themselves establish clinical efficacy or safety.
The company has described Chai-2 as a zero-shot antibody-design system capable of producing double-digit experimental hit rates in its reported testing. That is a company-reported laboratory metric, not a clinical outcome.
What to keep in mind. Chai remains a discovery-stage example in this article. A successful laboratory design or binding result is an early step in drug development. It does not demonstrate that a resulting molecule will become a safe and effective medicine in people.
6. Xaira Therapeutics: An AI Platform Still Before the Clinical Stage
What it is. A South San Francisco biotechnology company launched in 2024. It is led by CEO Marc Tessier-Lavigne and has a scientific advisory group that includes protein-design researcher David Baker. The company is developing AI models across the drug-discovery process, including models for biology, protein design and virtual-cell research.
Why it is on the list. In March 2026, Xaira launched X-Cell, its first virtual-cell model, trained on the X-Atlas/Pisces genome-wide perturbation dataset. The company has continued expanding its translational and early-clinical-development team during 2026.
What to keep in mind. Xaira’s public materials describe a pipeline moving toward clinical development, but we found no disclosed human clinical program in the sources reviewed for this article. That makes Xaira different from companies such as Insilico and Generate:Biomedicines: the evidence currently available is primarily platform, data and preclinical rather than patient-level clinical evidence.
Summary of the Evidence (and How Strong It Is)
| Evidence item | What it shows | Independent of the company? | Limitation |
|---|---|---|---|
| Rentosertib Phase IIa, Nature Medicine (2025) | Safety and a dose-dependent efficacy trend over 12 weeks | Peer-reviewed, but company-led | Small, short trial; trend, not proof |
| REC-4881 Phase 2 (TUPELO), company reports | 43% median reduction at 12 weeks; 53% at week 25 | No | Small sample; company-reported |
| GB-0895 Phase 1, company reports (posters, filings) | Tolerability and half-life in mild-to-moderate asthma | No | Not an efficacy trial; Phase 3 pending |
| ASCO 2026 survey of AI-enabled assets (via IntuitionLabs) | 117 assets across 63 companies; 8 had completed Phase 2 | Third-party summary | Secondary source; definitions of «AI-enabled» vary |
| Pharma partnerships (Lilly, Novartis, J&J, Pfizer, etc.) | Large pharmaceutical companies have licensed or accessed AI platforms | Partly | Commercial interest is not clinical evidence |
Limitations of AI Drug Discovery (and of This Article)
Limitations of this article
- It covers six companies out of a much larger field. The list is not exhaustive, and leaving a company out is not a judgment about it.
- Most company data are self-reported. We link to primary sources wherever we could.
- The situation changes quickly. Trial results, financing and partnerships can shift within weeks.
What to Watch for the Rest of 2026
- Recursion’s regulatory update following its FDA engagement around REC-4881.
- Enrollment and eventual results from Insilico’s Phase 3 GENESIS-IPF-3 trial.
- Whether Isomorphic Labs moves an internal drug candidate into human clinical testing.
- Whether Chai’s growing pharmaceutical partnerships produce candidates that progress from discovery into clinical development.
- Whether Xaira discloses its first clinical-stage program.
The same evidence-first approach applies elsewhere in digital health. If you are interested in how AI performs against clinicians in imaging, see AI in Radiology in 2026: Leading Tools, Real-World Uses and What Hospitals Should Know.
Frequently Asked Questions
Has any AI-discovered drug been approved by the FDA?
Not as far as we could verify as of September 29, 2026. Several are in clinical trials, and the most advanced are in Phase 3. Check the FDA’s approvals database and company announcements for updates.
Which companies have the most advanced clinical-stage programs covered here?
Insilico Medicine’s rentosertib and Generate:Biomedicines’ GB-0895 are both in Phase 3. They are not directly comparable: rentosertib is described in the published literature as a generative-AI-discovered small molecule targeting a target identified with generative AI, while GB-0895 is an AI-engineered antibody directed against the known target TSLP.
Does «AI-designed» mean a drug is safer or more effective?
No. How a molecule was discovered does not by itself say anything about how it behaves in patients. That is decided in clinical trials, the same way as for any other drug.
Are AI drug discovery companies good investments?
Searches for «AI drug discovery companies to invest in» often mix two different questions: which companies have promising technology and which companies may generate attractive investment returns. Those are not the same thing. This article does not answer the second one. Several of these companies are private, and the public ones have volatile share prices and long timelines to potential revenue. Please consult a licensed financial adviser before making investment decisions.
Why are there large funding rounds if clinical approvals are still limited?
Funding reflects expectations about the potential of AI-enabled discovery platforms, their ability to generate multiple candidates and the commercial value of partnerships. Financing is not the same as clinical validation, so funding announcements should be considered separately from human efficacy and safety data.
How often is this article updated?
We review it as significant news appears and update the date at the top when we do. The current review date is September 29, 2026.
Conclusion
Related Reading
Sources
Company and primary sources
- Isomorphic Labs. Isomorphic Labs secures $2.1 Billion funding to scale its AI drug design engine. May 12, 2026. isomorphiclabs.com
- Insilico Medicine. Nature Medicine publication of Phase IIa results of rentosertib. June 3, 2025. insilico.com
- Insilico Medicine. First patient dosed in GENESIS-IPF-3 (Phase 3). insilico.com
- Insilico Medicine. Proteomic aging clocks study. September 7, 2026. insilico.com
- Insilico Medicine. Rentosertib inhalation solution IND clearance. April 29, 2026. prnewswire.com
- ClinicalTrials.gov. NCT05938920 (rentosertib Phase IIa protocol). clinicaltrials.gov
- Recursion. First-quarter 2026 results and business update. May 6, 2026. ir.recursion.com
- Recursion. Form 8-K business highlights. February 25, 2026. sec.gov
- Generate:Biomedicines. Announcement of Phase 3 SOLAIRIA studies. December 1, 2025. generatebiomedicines.com
- Chai Discovery. 2026 company news and Series C announcement. chaidiscovery.com
- Xaira Therapeutics. X-Cell announcement. March 17, 2026. xaira.com
- Xaira Therapeutics. 2026 company news. xaira.com
- FDA and EMA. Guiding Principles of Good AI Practice in Drug Development. January 2026. fda.gov
- FDA. List of Artificial Intelligence-Enabled Medical Devices. fda.gov
Trade press and secondary sources
- pharmaphorum. Insilico ends 2025 with $293m Hong Kong IPO. December 31, 2025. pharmaphorum.com
- Fierce Biotech. Isomorphic Labs Series B coverage. May 12, 2026. fiercebiotech.com
- Xtalks. Generate Biomedicines IPO. March 5, 2026. xtalks.com
- pharmaphorum. Generate Biomedicines IPO filing. February 5, 2026. pharmaphorum.com
- BioSpace. Secretive AI drug hunter Xaira dips toe into dealmaking. July 8, 2026. biospace.com
- IntuitionLabs. AI-Discovered Drugs in Clinical Trials 2026: Full Pipeline. July 31, 2026. intuitionlabs.ai
- Drug Target Review. AI in drug discovery: predictions for 2026. April 15, 2026. drugtargetreview.com
- AIM Media House. 2026 Is the Year AI Drug Discovery Meets Clinical Reality. July 3, 2026. aimmediahouse.com