What Happened
Isomorphic Labs, the Alphabet-backed drug design company spun out of Google DeepMind, announced on May 12 that it has closed a $2.1 billion Series B funding round, one of the largest single-tranche financings ever for an artificial intelligence biotech. The round was led by Thrive Capital, the same firm that anchored its $600 million Series A in March 2025, with new investors CapitalG, Temasek, MGX, and the United Kingdom's Sovereign AI Fund joining existing backers Alphabet and GV. With this latest infusion, the company has now raised roughly $2.7 billion in external capital across two rounds, a striking concentration of conviction for a five-year-old laboratory that has yet to deliver a marketed therapy.
The capital is earmarked for two purposes that the company describes as inseparable: scaling up its proprietary AI Drug Design Engine, known internally as IsoDDE, and growing its scientific staff across three research hubs in London, Cambridge in Massachusetts, and Lausanne in Switzerland. Chief executive Demis Hassabis, who also leads DeepMind, framed the moment as a pivot from infrastructure-building to direct clinical engagement, saying the team is preparing for "the most consequential phase" of its short history. President Max Jaderberg, formerly a research lead inside DeepMind, has indicated that the company expects to dose its first patients with an AI-designed candidate before the end of 2026, focused initially on oncology and immune-system disorders.
The funding announcement comes roughly fifteen weeks after a separate milestone in late January, when the United States Food and Drug Administration cleared ISM8969 for first-in-human studies. ISM8969 is a small molecule whose structure and target were jointly proposed by IsoDDE working in concert with AlphaFold-derived protein models, and it is widely understood inside the industry as the first concrete test of whether neural networks can compress the multi-year, multi-billion-dollar grind of preclinical development into a faster, cheaper sprint.
Why It Matters
Pharmaceutical research has long been one of the most stubborn productivity puzzles in modern industry. The "Eroom's law" pattern, named as the deliberate inversion of Moore's law, describes how the number of new drugs approved per billion dollars of inflation-adjusted R&D spending has roughly halved every nine years since 1950. The promise that AI could reverse that decline has driven a decade of venture investment, but until now most of the field's headline candidates have been molecules that AI helped optimize at the margin rather than originate from a clean computational sheet.
Isomorphic Labs sits at the most ambitious end of that spectrum. Its founding bet, articulated by Hassabis in 2021, was that AlphaFold's leap in protein structure prediction would unlock not only the static shape of biological targets but also the dynamic surfaces where drug molecules might bind, allowing entire candidate series to be conceived and ranked in silico before a single test tube is touched. The Drug Design Engine, layered on top of AlphaFold's foundations, is designed to handle the messy realities of pharmacology: solubility, off-target effects, metabolic half-life, and the cellular environments that often defeat elegant chemistry. A funding round of this size is, in effect, a public signal from sophisticated capital that the company's models have begun to clear at least the easier of those hurdles.
The $2.1 billion also lands at a moment of considerable structural change in the broader biopharma capital markets. Many traditional venture investors have grown cautious about early-stage biotech after a punishing two-year correction in the sector. Strategic capital from sovereign vehicles like the UK Sovereign AI Fund and MGX, an Abu Dhabi-backed entity focused on AI infrastructure, reflects a different theory of return: that AI-native drug platforms are closer in character to compute companies than to traditional pharma, and that the right comparable is not a single asset but a re-runnable engine. Whether that thesis survives the gauntlet of real clinical readouts is the central question the next two years will answer.
Reaction
Industry analysts greeted the round with a mixture of enthusiasm and measured skepticism. On the bullish side, several biotech investors pointed to the unusual continuity of Thrive Capital leading back-to-back rounds as evidence that insider conviction has hardened rather than softened as ISM8969 moved closer to humans. Hassabis himself struck a deliberately humble note in interviews, acknowledging that the first generation of AI-designed candidates may still fail in trials and emphasizing that the company's value rests on the engine, not on any single molecule. That stance has earned credit from observers who have grown wary of biotech founders making single-asset binary bets.
Scientists outside the company have been more cautious. Computational chemists have noted that AlphaFold's accuracy on the kinds of disordered, membrane-bound, and complex multimer targets that dominate modern oncology pipelines remains uneven, and that benchmarks in published literature do not yet capture how the engine performs on the most pharmacologically interesting targets. A handful of academic critics have also warned that the field needs independent reproduction of the structure-activity relationships IsoDDE proposes, rather than relying on internal validation alone. Hassabis has responded by promising more peer-reviewed disclosures as candidates advance, though Isomorphic has historically guarded the specific architecture of its models more closely than DeepMind's published research arm.
Within Alphabet, the round is also read as a vote of corporate confidence at a time when several large technology firms are wrestling with how to translate frontier AI capability into businesses outside advertising and cloud. Treating Isomorphic Labs as a stand-alone company with outside investors, rather than as a captive lab, gives Alphabet a way to validate the asset through external capital and clinical milestones rather than internal accounting. If the trials succeed, the company becomes a credible counter-example to the argument that today's AI labs cannot generate non-software value. If they fail, the structure also contains the reputational risk.
What's Next
The immediate watch-list is short and consequential. The first patient dosing for ISM8969, expected before the end of 2026, will be the moment the engine collides with biology in a way that no benchmark can mediate. Phase 1 oncology trials of this kind are designed primarily to measure safety and pharmacokinetics rather than efficacy, but observers will pore over even the earliest readouts for evidence that the molecule behaves in humans the way IsoDDE predicted in silico. A second candidate, focused on an undisclosed immunology indication, is expected to follow within twelve months.
Beyond the company's own pipeline, Isomorphic Labs is also expanding its external research collaborations. It has existing deals with Eli Lilly and Novartis, both signed in 2024, that effectively license the engine for specific therapeutic areas, and analysts expect at least one additional large-pharma collaboration to be announced before year-end. These partnerships matter because they distribute clinical risk across multiple readouts and create independent data points on whether IsoDDE's predictions translate consistently across disease areas. Hiring will also accelerate sharply, with the company reportedly targeting several hundred new scientists, engineers, and clinical operations staff over the next eighteen months across its three sites.
Closing Thoughts
It is tempting, particularly for those who have lived through a decade of AI hype cycles, to treat $2.1 billion as the story. But the more interesting question Isomorphic Labs poses is structural rather than financial. Drug discovery is one of the few domains where a computational system, however clever, must eventually be judged by a binary biological outcome: the patient gets better or they do not. There is no benchmark to game, no leaderboard to top, no synthetic evaluation that will substitute for the truth of a clinical readout. That brute simplicity is what makes the next two years interesting.
If IsoDDE delivers a candidate that works, the implications extend well beyond oncology. It would suggest that the same general approach, foundation models plus domain-specific design engines plus tight loops with experimental data, can be ported to materials, agriculture, and any other field where molecular interactions govern the outcome. If it fails, the lesson will be equally clarifying: not that AI cannot help with drug discovery, which the literature has already settled, but that the leap from useful tool to autonomous designer is steeper than the most optimistic accounts suggested. Either result will reshape how much capital and how many careers flow into AI-for-science in the second half of the decade.
For now, Isomorphic Labs has bought itself something more valuable than the headline number suggests: roughly thirty months of runway in which to let science, rather than narrative, do the talking. That is a luxury few biotech companies, AI-native or otherwise, ever receive. What they do with it will be one of the defining experiments of the current era of artificial intelligence.
Further reading: Isomorphic Labs Series B announcement, Fierce Biotech coverage, MIT Technology Review on AlphaFold's road ahead, and Clinical Trials Arena on the upcoming trials.
한글 요약
알파벳 자회사이자 구글 딥마인드의 스핀오프인 Isomorphic Labs가 5월 12일 21억 달러 규모의 시리즈 B 투자를 유치했다고 발표했다. 2025년 3월 6억 달러 규모 시리즈 A를 주도한 Thrive Capital이 이번에도 리드 투자자로 나섰고, CapitalG·Temasek·MGX·영국 정부 주권 AI 펀드 등 신규 투자자가 합류했다. 누적 외부 자금은 약 27억 달러에 이른다.
이번 자금은 자체 신약 설계 엔진 IsoDDE를 확장하고 런던·매사추세츠 케임브리지·스위스 로잔 연구 거점의 인력을 늘리는 데 투입된다. 회사는 1월 미국 식품의약국(FDA)으로부터 첫 번째 AI 설계 항암 후보물질 ISM8969의 임상 1상 진입 승인을 받았으며, 데미스 하사비스 CEO는 2026년 말까지 첫 환자 투여를 목표로 한다고 밝혔다. 종양학과 면역질환이 우선 적응증으로 설정됐다.
업계는 이번 라운드를 "AI가 신약 후보를 처음부터 설계할 수 있는가"라는 가장 어려운 질문에 대한 본격적인 자본 시장의 검증으로 보고 있다. 임상 1상은 안전성과 약동학을 평가하지만, IsoDDE의 인실리코 예측이 실제 사람의 몸에서 어떻게 작동하는지 보여줄 결정적 데이터 포인트가 될 전망이다. 향후 12개월간 추가 면역 분야 후보 임상 진입과 대형 제약사와의 신규 제휴 발표가 예상된다.