Alphabet spin-off raises $2.1B to develop drugs using AI
Demis Hassabis's project aims to revolutionize pharma with AI. Investors include Thrive Capital and Google Ventures.
Isomorphic Labs and $2.1B: Why 'curing all diseases' is not megalomania but a business model
The gist: Betting not on a drug, but on a drug factory
On May 12, 2026, Isomorphic Labs announced the close of a $2.1 billion Series B round led by Thrive Capital, with participation from Alphabet, GV, MGX, Temasek, CapitalG, and the UK Sovereign AI Fund. A year ago, the Series A was $600 million. In less than 18 months, the company's valuation, by indirect estimates, surpassed $10 billion.
On the surface, it's a giant check for an AI biotech. In essence, it's the largest bet in history that drug discovery can be turned into a computational problem. Not accelerated, not optimized, but turned. The difference is fundamental.
Isomorphic Labs is not looking for a cure for cancer or Alzheimer's. It is building a machine that will find cures for everything. This is not a pharmaceutical company with AI tools. It is an AI company that views biology as a domain for computation.
Timeline and context: From Nobel Prize to record check in two years
The story of Isomorphic Labs is a rare case where scientific recognition and market timing aligned to the year.
2020: DeepMind unveils AlphaFold 2, solving the '50-year problem'—predicting a protein's 3D structure from its amino acid sequence. Accuracy approaches experimental methods, but speed is orders of magnitude higher.
2021: Alphabet spins off Isomorphic Labs as a separate company. CEO is Demis Hassabis—founder of DeepMind and one of the architects of AlphaFold.
2024: Hassabis and John Jumper receive the Nobel Prize in Chemistry for AlphaFold. This is the moment the scientific establishment formally recognizes: AI can solve problems once deemed inaccessible.
March 2025: Series A of $600 million—first external money.
May 2026: Series B of $2.1 billion—largest single round in AI pharma history.
The speed of check escalation reflects not just hype. It reflects a shift in risk perception. Traditional pharma is a portfolio of dozens of programs, where 90% fail in clinical stages. Isomorphic promises to radically reduce the attrition rate through AI prediction accuracy. If it works, each dollar invested yields disproportionately higher expected returns than in the classic model.
Who wins and who loses
Alphabet wins—in its unique style.
Alphabet does not absorb Isomorphic or demand immediate returns. Instead, it retains a controlling stake, allows the company to raise external capital, build valuation, and develop independent expertise. This is the same strategy as with Waymo: fund through external investors what could become the next trillion-dollar business without burdening the parent company's balance sheet.
Thrive Capital wins—and its founder Joshua Kushner.
Thrive Capital led the Series A and has now doubled down in Series B. For Kushner, this is not just a portfolio investment but a strategic asset in the race for AI dominance. Thrive has already invested billions in OpenAI—betting on Isomorphic diversifies the AI portfolio toward life sciences.
Sovereign funds win—and this is an underappreciated signal.
MGX (UAE), Temasek (Singapore), UK Sovereign AI Fund—the new investor lineup is impressive. These are not venture funds chasing multiples. These are states buying a seat in what they see as future infrastructure. When a sovereign fund from Abu Dhabi, which also partners with Microsoft and BlackRock, invests in AI biotech, it means AI drug discovery is recognized as a matter of national security and economic sovereignty.
Classic biotech startups lose.
If Isomorphic's model works, the traditional approach—'synthesize thousands of molecules and see which binds to the target'—becomes morally obsolete. Competitors, including Recursion, Meta Flux, and Exscientia, now compete not just with a company but with a platform that has raised $2.7 billion in total capital.
Patients lose—for now.
No AI-designed molecule has yet passed clinical trials. None has reached the market. Hassabis first promised clinical trials by end of 2025, then pushed to end of 2026. This is not a failure, but it's a calibration of expectations. The last mile—from computer model to human-tested drug—has proven longer than optimists predicted.
What the media isn't saying
First insight: IsoDDE is not an improved AlphaFold. It's a fundamentally different product.
AlphaFold predicts a protein's static structure. It solves the first, visual task: 'what the target looks like.' But drug design requires answers to questions AlphaFold doesn't address: how strongly does a molecule bind to the target? Does it bind to anything else in the body? How is it metabolized?
IsoDDE (Isomorphic Drug Design Engine) answers precisely these questions. In recent tests, the platform's accuracy in predicting protein-ligand interactions more than doubled that of AlphaFold 3. It predicts binding affinity with accuracy comparable to the 'gold standard'—physical modeling—but orders of magnitude faster. It finds 'hidden pockets' on protein surfaces—drug binding sites that scientists missed for decades.
This is not a cosmetic improvement. It's a shift from 'we see the target' to 'we can design the perfect key for this lock in days, not years.'
Second insight: Isomorphic does not make drugs. It makes a repeatable process for creating them.
The key word in the company's rhetoric is 'engine.' President Max Jaderberg says: 'Our drug design engine works, and it gives us a repeatable way to create new drugs.' 'Repeatable' is what matters to investors.
Traditional drug discovery is a mix of science, intuition, and luck. Repeatability is low. If Isomorphic has truly built a system that outputs candidates with predictable quality, this is not an improvement of pharma. It's a complete overhaul. Pharma becomes like chip design: set specifications, get a predictable result.
This—not the promise to 'cure all diseases'—is what convinced investors to write a $2.1 billion check.
Third insight: Big Pharma partnerships are not technology validation; they are insurance against failure.
Novartis, Johnson & Johnson, and Eli Lilly are strategic partners of Isomorphic. The total partnership contracts are estimated at nearly $3 billion. It sounds like validation. But look at the structure: these are collaboration agreements, not purchases of specific molecules. Big Pharma pays for access to the platform and first right of refusal on results.
For Big Pharma, $3 billion spread over several years is cheap insurance. If the AI model works, they get the best molecules ahead of competitors. If not, they lose an amount comparable to the budget of a single Phase 3 clinical program. It's an asymmetric bet with limited downside.
For Isomorphic, these partnerships are a revenue stream not tied to clinical trial success. The company earns from the process, not just the outcome.
Forecast: Next 30 days and 90 days
30 days (through mid-June 2026).
Expect hiring announcements. $2.1 billion is money for scaling. The company will aggressively recruit clinical trial, toxicology, and regulatory specialists. This signals a shift from 'we are an AI lab' to 'we are a pharma company with AI.'
Also possible is an announcement of a specific clinical program—indication, phase, timeline. Hassabis pushed deadlines to end of 2026, but investors will want to see progress sooner. A concrete announcement—e.g., 'IND filing for an oncology program in Q3 2026'—would reassure the market.
90 days (through mid-August 2026).
The key indicator is progress from preclinical to clinical studies. By August, the company should show at least one program ready for IND submission. This will be the moment of truth: it's one thing to publish impressive benchmark results, another to convince the FDA that an AI-designed molecule is safe for human administration.
Also during this period, an expansion of geography is likely. The company already has offices in London, Cambridge, and Lausanne. A logical next step is opening a research center in the US—closer to major clinical sites and regulators. The UK Sovereign AI Fund, which participated in the round, may offer tax incentives to keep headquarters in Britain, but for clinical operations, the US is indispensable.
Bottom line.
Isomorphic Labs raised $2.1 billion not for a promise to find a cure for a specific disease. It raised it for a promise to turn the drug discovery process into an engineering problem. If the promise is fulfilled, it will be the biggest breakthrough in medicine since the genome was decoded. If not, it will be the most expensive AI experiment in history.
Demis Hassabis likes to say the company's goal is to 'solve all diseases.' It sounds like megalomania. But when backed by a Nobel Prize, AlphaFold, IsoDDE, and $2.7 billion in total investment, it's no longer mania. It's a business plan. And 2026 is the year that plan begins to face reality.
— Editorial Team
No comments yet.