Insitro
Last reviewed
Sources
15 citations
Review status
Source-backed
Revision
v2 ยท 1,832 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Sources
15 citations
Review status
Source-backed
Revision
v2 ยท 1,832 words
Add missing citations, update stale details, or suggest a clearer explanation.
Insitro (styled "insitro") is a machine learning-driven drug discovery and development company founded in 2018 by Daphne Koller, the Stanford machine learning professor and Coursera co-founder, and headquartered in South San Francisco, California. The company combines large in-house biological datasets, including induced pluripotent stem cells, functional genomics, and high-content cellular imaging, with human genetics and modern machine learning to find novel drug targets and design therapeutics. As of 2026 insitro has raised more than $700 million in venture capital, including a $400 million Series C in March 2021, and has signed multibillion-dollar discovery partnerships with Gilead Sciences, Bristol Myers Squibb, and Eli Lilly and Company. [1][2][15]
The company's central premise is that the high failure rate of drug development stems from a poor understanding of disease biology, and that this gap can be narrowed by generating large, purpose-built datasets and applying modern machine learning to them. Insitro combines wet-lab data generation, including induced pluripotent stem cells, functional genomics, and high-content cellular imaging, with analyses of large human cohorts to connect molecular and cellular measurements to clinical outcomes. It describes this strategy as building a "pipeline through platform," meaning it develops its own therapeutic programs and partnered programs on top of a reusable data-and-models engine rather than pursuing one-off drug candidates. [2][3]
Insitro is widely regarded as one of the leading machine-learning-first biotechnology companies, alongside firms such as Recursion Pharmaceuticals, Isomorphic Labs, Xaira Therapeutics, Genesis Therapeutics, and Generate:Biomedicines. Its focus areas span metabolic disease, neurodegeneration, and oncology. [3][4]
Insitro is an American biotechnology company that builds machine learning models of human disease to discover drug targets and design medicines. Rather than treating computation as a downstream analysis step, it couples large-scale data generation in automated wet labs with predictive modeling so that biology guides target selection, modality choice, and molecular design. The name is a play on the laboratory terms "in silico" (computational) and "in vitro" (in cell culture), reflecting the founding thesis that tightly coupling computation with experimental biology produces better predictions than either alone. [1][2]
Insitro was launched publicly in May 2018 by Daphne Koller, who serves as founder and chief executive officer. Koller held the Rajeev Motwani Professorship in computer science at Stanford University for 18 years and remains an adjunct faculty member there. She co-founded the online education company Coursera, where she served as co-chief executive officer and president, and before founding insitro she was chief computing officer at Calico, an Alphabet life-sciences company. Her honors include a MacArthur Fellowship (2004), the ACM Prize in Computing (2008), and election to the National Academy of Sciences (2023); she was named to TIME's list of the 100 most influential people in AI in 2024. [1][5]
Koller has frequently framed insitro's mission around the difficulty of "failing fast" in biopharma, arguing that better predictive models can de-risk programs earlier and reduce the cost of late-stage clinical failures. [1][5]
Insitro's approach integrates two broad data sources. First, it generates large proprietary datasets from human biology in its own automated laboratories, including induced pluripotent stem cell models, CRISPR-based functional genomics, and microscopy-based cellular phenotyping. Second, it draws on large-scale human data, such as genetic, clinical, and medical imaging data from population cohorts, to ground its models in disease as it manifests in patients. Machine learning models trained on these data are used to define disease subtypes, discover therapeutic targets, and predict which patients are most likely to benefit from a given intervention. [2][3]
Over time the company has described several named components of its platform, reflecting its expansion across the drug discovery process. Early target-discovery work was associated with the "insitro Human" (ISH) platform; the company later described a "ChemML" platform for machine-learning-driven small-molecule chemistry and a "Virtual Human" platform for connecting molecular data to patient outcomes. In January 2026 it consolidated and extended these capabilities under a unified platform branded TherML (Therapeutic Machine Learning). [6][7][8]
Insitro's design capabilities span multiple drug modalities. The company has built machine-learning workflows for small molecules and for oligonucleotides, including AI-driven design of siRNA molecules, and through its 2026 acquisition of CombinAbleAI it added physics-informed design of complex biologics such as multi-specific antibodies and T-cell engagers. A stated goal of the unified platform is to optimize for potency and manufacturability at the same time rather than sequentially. [7][8]
Insitro has funded its platform in part through large strategic partnerships with established pharmaceutical companies, in which it typically leads target discovery and early biology while partners contribute development, delivery technology, or commercialization. Reported headline deal values represent the maximum potential including milestones and royalties, and actual amounts depend on undisclosed near-term payments and the achievement of future milestones. [9][10]
| Partner | Announced | Disease area | Reported terms |
|---|---|---|---|
| Gilead Sciences | April 2019 | Nonalcoholic steatohepatitis (NASH) | Roughly $15M upfront, near-term payments up to about $35M, and total potential value reported above $1 billion across up to five targets [9] |
| Bristol Myers Squibb | October 2020 | Amyotrophic lateral sclerosis (ALS) and frontotemporal dementia | $50M upfront and more than $2 billion in potential milestones; later extended and expanded [10][11] |
| Eli Lilly and Company | October 2024 | Metabolic disease, including MASLD | Multiple agreements involving siRNA and antibody programs; insitro retains global rights, with Lilly eligible for milestones and royalties (amounts not disclosed) [12] |
The Gilead collaboration, insitro's first pharmaceutical partnership, used the company's platform to build disease models of NASH (also called metabolic dysfunction-associated steatohepatitis) and to identify targets influencing disease progression. The Bristol Myers Squibb collaboration, focused on ALS and frontotemporal dementia, has produced validated targets: Bristol Myers Squibb selected a first ALS target in December 2024, triggering a $25 million milestone payment, and in March 2026 nominated two additional targets (ALS-2 and ALS-3) identified through insitro's platform, triggering a further $10 million payment. The collaboration was extended in October 2025. [10][11] The 2024 agreements with Eli Lilly center on metabolic dysfunction-associated steatotic liver disease (MASLD) and pair insitro-discovered targets with Lilly's delivery technology and development capabilities. [12]
Insitro has raised capital across at least three large venture rounds. Reported figures for the company's cumulative funding vary by source, and the company's own descriptions and third-party aggregators do not fully agree. [13][14]
| Round | Date | Amount | Lead and selected investors |
|---|---|---|---|
| Series A | May 2018 | About $100 million | ARCH Venture Partners, GV, Andreessen Horowitz, Foresite Capital, Third Rock Ventures [13] |
| Series B | May 2020 | About $143 million | Led by Andreessen Horowitz; included Canada Pension Plan Investment Board, T. Rowe Price-advised funds, BlackRock-managed funds, Casdin Capital [14] |
| Series C | March 2021 | About $400 million | Led by Canada Pension Plan Investment Board; included Andreessen Horowitz, T. Rowe Price, BlackRock, ARCH, Foresite, GV, Temasek, SoftBank Vision Fund advisors [15] |
The disclosed amounts from these three rounds sum to roughly $643 million. Insitro's own corporate descriptions have stated that the company has raised "more than $700 million" to date, and some investment-data aggregators report higher cumulative totals approaching $1 billion, which may reflect additional or undisclosed financing. Because these figures are not fully reconciled across public sources, the round-by-round amounts above are the most reliably documented. [2][13][14][15]
On January 12, 2026, insitro announced that it had agreed to acquire CombinAbleAI, an AI biotechnology company specializing in the design of complex biologics, with the transaction expected to close in late January 2026; financial terms were not disclosed. CombinAbleAI's technology uses a physics-informed, AI-driven optimization engine pre-trained on more than 100,000 molecular dynamics surrogates to predict protein structure and flexibility, which insitro intends to apply to multi-specific antibodies, T-cell engagers, and other complex biologics. The CombinAbleAI team, based in Rehovot, Israel, was set to continue as a new insitro research and development center focused on large-molecule design. [6][7][8]
The acquisition was paired with the launch of insitro's TherML (Therapeutic Machine Learning) platform, which the company describes as a full-stack, modality-agnostic AI engine spanning small molecules, oligonucleotides, antibodies, and other biologics within a single system. Insitro's chief scientific officer, Philip Tagari, said the integrated platform allows the company to "treat potency and manufacturability as interdependent design criteria from the outset," rather than discovering manufacturing constraints only after identifying potent candidates. Announcing the deal, Koller said, "We are delighted to welcome the CombinAbleAI team in Israel as fellow insitrocytes, alongside our colleagues in the U.S., Poland and Malaysia." [7][8]
Insitro operates within a competitive field of machine-learning-first drug discovery companies. Direct peers include Recursion Pharmaceuticals, which built a large phenomics platform from automated cell imaging; Isomorphic Labs, the Alphabet drug-discovery company spun out of DeepMind and building on the AlphaFold protein-structure work; and Xaira Therapeutics, a heavily capitalized startup founded around generative protein-design methods. Other competitors include Genesis Therapeutics and Generate:Biomedicines, both of which apply machine learning to molecule and protein design. [3][4]
What distinguishes insitro from several of these peers is its emphasis on generating large in-house human-relevant biological datasets, particularly from induced pluripotent stem cells and human cohorts, and on using those data to discover the targets and patient populations for new medicines, rather than focusing primarily on the chemistry or structural design step. With its 2026 expansion into biologics design through CombinAbleAI and the TherML platform, the company has positioned itself to compete across a broader range of drug modalities while retaining its target-discovery focus. [3][7][8]