OpenAI, Anthropic and Google in Biotech: The 2026 Deal Tracker

The companies that build the most capable AI models are assembling the parts of a drug company: the reasoning engine, the lab bench, the operations staff and the investment portfolio. This tracker logs every life sciences move by OpenAI, Anthropic and Google that we can verify, with the date, the price where one exists, and what each deal adds. Last updated October 21, 2026.

Presented By Our Partners

The short answer: In 2026, OpenAI bought three health or research companies (Torch, Crixet and Orchestra Bio), launched its GPT-Rosalind life sciences model and kept investing in Chai Discovery. Anthropic bought Coefficient Bio for a reported $400 million in stock, launched Claude Science, opened its own wet lab and started its own early drug programs. Alphabet’s Isomorphic Labs raised $2.1 billion and expects its first human trials by the end of 2026. No AI-discovered drug has reached FDA approval yet.

On October 20, OpenAI bought a four-year-old San Francisco startup that few people outside biotech had heard of. Orchestra Bio did not design molecules. It built software agents that watch over the unglamorous machinery of a drug program: the contract research organizations running studies, the manufacturers producing batches, the timelines that slip when either one stumbles. Six months earlier, Anthropic paid a reported $400 million in stock for Coefficient Bio, a team of fewer than ten people led by former Genentech scientists. Between those two deals sit a Nobel laureate changing employers, a new biology lab in the Bay Area, a $2.1 billion round for Alphabet’s drug designer, and a set of independent companies trying to automate the scientific method itself. Read one at a time, each is a headline. Read together, they look like a blueprint.

The 2026 AI lab biotech tracker at a glance

Entries are listed newest first. “Reported” means a figure came from a news outlet’s sources and was not confirmed by the companies.

OpenAI: buy the workflow, back the molecule makers

OpenAI’s life sciences effort has a name: Rosalind. GPT-Rosalind, introduced April 16 and named for the British scientist Rosalind Franklin, is a reasoning model built for biochemistry, drug discovery and translational medicine. It launched as a research preview for qualified customers through a trusted-access program, alongside a more widely available Life Sciences plugin for Codex that connects researchers to more than 50 scientific tools and data sources. Collaborators named at launch included Amgen, Moderna, Thermo Fisher Scientific, NVIDIA, the Allen Institute and the UCSF School of Pharmacy. On September 11 the model left research preview for eligible organizations worldwide, with published pricing taking effect October 5. Rosalind Workbench connects the model to scientific tools and data so researchers can analyze results and plan experiments, and Orchestra Bio now joins that stack. Our full analysis of the Orchestra Bio deal scores OpenAI’s promise against the evidence.

The acquisitions are small and targeted. In January, OpenAI bought Torch, a year-old app that gathered lab results, medications and visit recordings into one record, reportedly for about $100 million in equity, and moved its four-person team into ChatGPT Health. It also bought Crixet, a LaTeX editing tool it relaunched as Prism for researchers. Orchestra Bio followed in October. Health and science are a small share of a much larger buying spree: the deal database Tracxn counts 27 OpenAI acquisitions in total, 14 of them in 2026, most in developer tools, security, media and consumer software.

OpenAI’s larger biology bets are minority stakes and partnerships. It co-led the 2024 seed round of Chai Discovery, a molecule design company founded by Josh Meier, who previously worked at OpenAI and Facebook, and it has backed Chai in every round since, including a $400 million Series C in July 2026 at a $3.8 billion valuation. Chai now has paid deployments with Eli Lilly, Pfizer, Novartis and argenx. OpenAI also built GPT-4b micro, a protein engineering model, with Retro Biosciences, a longevity company that Sam Altman personally funds; early tests produced Yamanaka factor variants reported to be more than 50 times more effective, a company-reported result that drew conflict-of-interest questions. And since 2024 it has worked with Sanofi and Formation Bio on custom models across drug development.

Anthropic: where Claude is heading in the life sciences

Anthropic has taken the opposite route: few deals, heavy in-house building. Claude for Life Sciences arrived in October 2025, aimed at biopharma professionals from clinical trial coordinators to regulatory managers. Claude for Healthcare followed in January 2026. In April, Anthropic bought Coefficient Bio in an all-stock deal reported at just over $400 million, for a startup founded in 2025 with fewer than ten employees and no disclosed revenue. Its platform focused on drafting research and development plans, shaping regulatory strategy and identifying drug candidates, and its team joined the life sciences group led by Eric Kauderer-Abrams. We covered that deal in Anthropic Just Bought a Drug Discovery Startup for $400M.

The rest of the year filled in the picture. On April 14, Novartis CEO Vas Narasimhan joined Anthropic’s board, selected by the company’s independent Long-Term Benefit Trust. In June, AlphaFold co-creator John Jumper, who shared the 2024 Nobel Prize in Chemistry, left Google DeepMind for Anthropic. On June 30, Anthropic launched Claude Science, a research workbench with more than 60 curated skills and connectors for genomics, single-cell research, proteomics, structural biology and cheminformatics. It is not a new model; it runs on existing Claude models inside a lab’s own infrastructure. At the same event, the company said it would run its own early drug programs in areas established industry has passed over for financial reasons.

In September, Anthropic confirmed a wet biology lab in the Bay Area. The company says the lab focuses on fundamental biology rather than drug discovery, and that human scientists perform all of the bench work, though CEO Dario Amodei has not ruled out Claude operating lab equipment in the future. Days later it announced the lab’s first headline result: an enzyme system with properties reminiscent of CRISPR, found mostly, though not entirely, by Claude, using roughly 950 agents. It also opened a Life Sciences Verification Program that gives vetted researchers access to its most capable models. Its big pharma partners include Sanofi, Novo Nordisk and AbbVie. The stated ambition, as summarized from Kauderer-Abrams’s remarks at SynBioBeta 2026, is to compress the life sciences research and development timeline tenfold. We score that claim in our Orchestra Bio analysis.

Anthropic’s other acquisitions are not biological: the JavaScript runtime Bun in December 2025 and the computer-use startup Vercept in February 2026. In life sciences, the company is betting on vertical integration: its own models, its own lab and its own programs.

Featured Partner

Invest in the Infrastructure Behind Modern Medicine

As healthcare expands beyond hospital walls, the buildings and campuses supporting that shift are generating compelling returns for investors who move early. The Healthcare Real Estate Fund offers qualified investors direct access to a curated portfolio of medical office, outpatient, and specialty care facilities.

Learn More →

Google DeepMind and Isomorphic Labs: the spinout route

Alphabet took a third path years before the others. Isomorphic Labs spun out of Google DeepMind in 2021 to turn AlphaFold’s protein structure breakthroughs into medicines. It raised $600 million in its first outside round in 2025, then $2.1 billion in May 2026 in a round led by Thrive Capital, with Google Ventures and Alphabet joined by MGX, Temasek and CapitalG. It unveiled its Isomorphic Drug Design Engine in February 2026 and holds milestone partnerships with Eli Lilly and Novartis. Its first human trials are now expected by the end of 2026, a year later than the target CEO Demis Hassabis originally set. For the backstory, see The Nobel Ensemble: Inside the Lab Trying to End Disease and Longevity Escape Velocity, Explained Through Isomorphic Labs.

The independents automating the scientific method

A second category is not owned by any frontier lab, though several share its investors. These companies pair AI models with robotic laboratories to run the full loop of hypothesis, experiment and learning, betting that the next leap in scientific AI will come from new experimental data rather than more internet text.

Lila is the clearest example of the category. It landed at number 25 on CNBC’s 2026 Disruptor 50 list, signed a 235,500 square foot lab lease in Cambridge, and reported in September that its AI-directed lab proposed, synthesized and screened 2,942 green hydrogen catalysts in three months, identifying six high-performing material families. That result is company-reported. “Scientific superintelligence,” Lila’s phrase for its goal, is an ambition rather than a demonstrated capability. On the drug side, the most advanced test of AI-native discovery remains Insilico Medicine’s rentosertib, covered in AI Designed This Lung Drug, and It Just Worked in Real Patients.

The pattern: five layers of a drug company

Each lab is choosing which layers to own and which to rent. Anthropic is integrating vertically, with its own lab and drug programs. OpenAI is building an ecosystem of tools, partnerships, minority stakes and targeted tuck-in acquisitions. Alphabet put its drug ambitions into a standalone company. For startups, that creates a new kind of exposure: the fear that a lab will ship a feature that replaces them. In Orchestra’s case, the company was absorbed rather than flattened. For a deeper look at how capital is organizing around these layers, see Generative Biology: The New Healthcare Venture Capital Stack. For the investor’s view of these deals, see Frontier AI Labs Are the New Strategic Acquirers in Healthcare on Healthcare Venture Capital Fund.

What has not been proven yet

Funding is not efficacy, and the evidence on AI-discovered drugs is still early. A 2024 analysis of AI-native biotech pipelines by Boston Consulting Group authors, published in Drug Discovery Today, found that AI-discovered molecules succeeded in phase 1 trials 80 to 90 percent of the time, well above historical averages, but about 40 percent of the time in phase 2, roughly in line with history and on a small sample (DOI). The most advanced randomized test, Insilico’s rentosertib for idiopathic pulmonary fibrosis, enrolled 71 patients for 12 weeks in a phase 2a trial published in Nature Medicine; its primary endpoint was safety, and the highest dose showed a lung function signal that the authors said warrants larger and longer trials (DOI). For scale, a JAMA study of drugs approved from 2009 to 2018 estimated a median capitalized research and development cost of $985 million per new medicine (DOI). Peer-reviewed sources were retrieved via PubMed.

Timelines also slip. Isomorphic moved its first trials back a year. Periodic Labs’ reported valuation round had not been confirmed as closed by mid-2026. Industry forecasts put the first FDA approval of an AI-discovered drug in 2027 or 2028, and those forecasts depend on clinical data still being generated. More context on where AI-designed candidates stand is in The First Wave of AI-Designed Drugs Hits Human Trials and Why the Breakthrough You Read About Never Became a Treatment.

How to read the next AI lab biotech headline

1. Which layer did they buy? A molecule company and an operations company solve different problems, and only one of them changes whether a drug works.
2. Was a price disclosed, or only a team? Many of these deals are acqui-hires, and the acquired product often shuts down or disappears into a larger platform.
3. Is the promise a goal or a measurement? A measurement names the programs, the time or money saved, and what it was compared against.
4. What would verify it? Our HYPE Index asks this of every claim: a clinical outcome, a regulatory decision, peer-reviewed data, or nothing yet.
5. How far is it from a patient? Discovery news is usually years from a prescription. Development news is closer, but it still runs through trials.

What This Means for You

If you are a patient, nothing on this tracker changes what your doctor can prescribe this year. The earliest AI-designed candidates are in early and mid-stage trials, and even optimistic forecasts put the first approval in 2027 or later. Watch phase 2 and phase 3 results, not funding rounds.

If you work in research or at a biotech, these tools are arriving in your workflow now through GPT-Rosalind, Claude Science and research plugins. Ask what each tool was validated on, in which tasks, and who reviews an agent’s output before it touches a regulated decision. Our coverage of who decides which AI is good enough for medicine explains why benchmark scores are not the same thing.

And while the labs race, the bridge to whatever they deliver is still built from foundations: sleep, movement, nutrition, breath and mindset, the habits that keep you healthy long enough to benefit from the breakthroughs.

Frequently Asked Questions

What companies has OpenAI acquired in health and life sciences?

In 2026 OpenAI acquired Torch, a personal medical records app, in January for a reported $100 million in equity; Crixet, relaunched as the research writing tool Prism, in January; and Orchestra Bio, a drug development operations platform, on October 20 for an undisclosed price. It also invests in Chai Discovery and partners with Sanofi and Formation Bio.

What has Anthropic done in the life sciences?

Anthropic launched Claude for Life Sciences in October 2025 and Claude for Healthcare in January 2026, acquired Coefficient Bio in April 2026 for a reported $400 million in stock, added Novartis CEO Vas Narasimhan to its board, hired AlphaFold co-creator John Jumper, launched Claude Science on June 30, 2026, and confirmed its own Bay Area wet lab in September 2026.

What is Lila Sciences?

Lila Sciences is a Cambridge, Massachusetts company founded by Flagship Pioneering in 2023. It pairs AI models with robotic labs it calls AI Science Factories to run the scientific method from hypothesis to experiment. It has disclosed $550 million in funding and was reported in June 2026 to be in talks to raise about $2 billion.

Is Orchestra Bio the same company as Orchestra BioMed?

No. Orchestra Bio is the private San Francisco software startup OpenAI acquired in October 2026. Orchestra BioMed is an unrelated Nasdaq-listed medical device company, ticker OBIO, developing cardiovascular therapies.

Has an AI-discovered drug been approved yet?

Not as of October 2026. AI-discovered molecules have performed well in early safety trials, and Insilico’s rentosertib showed a lung function signal in a 71-patient phase 2a trial, but industry forecasts place the first FDA approval in 2027 or later.

When will Isomorphic Labs start human trials?

Isomorphic Labs expects its first clinical trials by the end of 2026, a year later than the end of 2025 target its CEO set earlier.

Sources

• OpenAI: OpenAI acquires Orchestra Bio
• TechCrunch: OpenAI acquires startup Orchestra Bio
• OpenAI: Introducing GPT-Rosalind for life sciences research
• Fierce Healthcare: OpenAI acquires Torch
• Tracxn: Acquisitions by OpenAI
• TechCrunch: Chai Discovery Series B
• PharmaVoice: Anthropic and Coefficient Bio
• BioSpace: Novartis CEO joins Anthropic board
• TechCrunch: Anthropic is operating a lab that conducts biology experiments
• TechCrunch: Anthropic says its biology lab has already found something big
• SynBioBeta: Anthropic is hiring biologists and building wet labs
• Anthropic: Anthropic acquires Vercept
• Reuters via WHTC: Isomorphic raises $2.1 billion
• Reuters: Lila Sciences tops $1.3 billion valuation
• Bloomberg: Lila Sciences in talks for funds
• Forbes: Periodic Labs funding talks
• Tech Funding News: Edison Scientific seed round
• HLTH: Sanofi, Formation Bio and OpenAI partnership
• Jayatunga M, et al. How successful are AI-discovered drugs in clinical trials? Drug Discovery Today, 2024. doi.org/10.1016/j.drudis.2024.104009
• Xu Z, et al. A generative AI-discovered TNIK inhibitor for idiopathic pulmonary fibrosis: a randomized phase 2a trial. Nature Medicine, 2025. doi.org/10.1038/s41591-025-03743-2
• Wouters OJ, McKee M, Luyten J. Estimated research and development investment needed to bring a new medicine to market, 2009-2018. JAMA, 2020. doi.org/10.1001/jama.2020.1166

How this tracker is maintained

Entries require a company announcement or reporting by a named outlet. Prices marked as reported come from a single outlet’s sources and have not been confirmed by the companies, and funding talks stay labeled as talks until a close is announced. Healthcare Discovery uses AI research tools from several companies on this page, including OpenAI and Anthropic models, in its editorial workflow; no company listed reviewed or paid for this tracker.

Update log: October 21, 2026, tracker published with entries from May 2024 through October 20, 2026.

Free Daily Briefing

The Latest Longevity Science.
Delivered Every Morning.

Join researchers, physicians, and health professionals getting daily breakthroughs in AI-driven medicine, epigenetics, and longevity research.

Support the research that powers this editorial

Subscribe Free

No spam. Unsubscribe anytime. We respect your inbox.

Keep Reading on Healthcare Discovery

• Senolytics in 2026: Inside the Clinical Trials Targeting Zombie Cells to Extend Healthy Lifespan
• Biological Aging Reversal in 2026: Gut Bacteria, NAD+ Balance, and Epigenetic Clocks Point the Way
• What the World's Oldest Person Reveals About the Biology of Extreme Longevity