AI has now become a familiar part of the drug discovery conversation. What is changing in 2026 is the way that technology is being commercialized. Recent deals show pharma companies using a range of structures to access AI-enabled discovery like licensing existing assets alongside future programs, funding multi-target collaborations, validating platforms through pilot projects, and even paying for proprietary datasets that can improve internal AI models.
Five deals announced this year illustrate how quickly the commercial landscape is evolving.
Insilico Medicine and Eli Lilly: Combining Asset Licensing with Future Discovery
Insilico Medicine’s March collaboration with Eli Lilly is one of the clearest examples of an AI biotech combining conventional asset licensing with broader platform access.
Under the agreement, Lilly receives exclusive worldwide rights to a portfolio of preclinical oral therapeutics across multiple therapeutic areas. At the same time, the companies will collaborate on additional R&D programs focused on targets selected by Lilly.
Insilico is eligible for USD 115m upfront, with development, regulatory and commercial milestones taking the potential total value to approximately USD 2.75bn, plus tiered royalties.
The AI sits behind both parts of the transaction. Insilico’s Pharma.AI platform integrates generative AI and other computational approaches across target identification and molecular design. Commercially, however, the more interesting point is that Lilly is not simply buying one AI-generated molecule. It is securing both existing pipeline assets and access to a repeatable discovery capability.
That hybrid structure could become increasingly important as AI biotechs seek to demonstrate that their platforms can generate multiple commercially valuable programs rather than a single success.
Isomorphic Labs and Johnson & Johnson: Partnering Around the Discovery Engine
Isomorphic Labs’ January collaboration with Johnson & Johnson takes a different approach. The companies entered a multi-target, cross-modality research collaboration, with Isomorphic applying its AI-first drug design capabilities across modalities including small molecules, antibodies, peptides and molecular glues. Johnson & Johnson contributes its expertise in target biology, experimental validation and downstream drug development. Financial terms were not disclosed.
Here, there is no headline licensed asset anchoring the partnership. The value lies in the discovery engine itself.
That distinction matters. It suggests that pharma companies are becoming willing to form strategic relationships around AI platforms before a specific therapeutic candidate has emerged. For AI-centered biotechs, this potentially opens a broader business model: rather than waiting until an internally generated asset is sufficiently mature to license, the technology can generate value much earlier in the discovery process.
Iktos and Servier: AI Meets Automated Chemistry
The structure of Iktos’ collaboration with Servier is particularly notable because it combines AI-driven design with automated experimental execution.
Announced in January, the multi-year agreement has a potential value exceeding EUR 1bn, comprising an upfront payment, research funding and milestone payments. Iktos will use its generative AI and AI-orchestrated robotics platform to design, synthesize and optimize small molecules against multiple undisclosed targets, while Servier will conduct proprietary biological assays and take responsibility for preclinical and clinical development of selected programs.
The underlying technology creates a closed loop between computational design and laboratory chemistry. AI proposes molecules, automated systems synthesize them and experimental results can feed back into subsequent design cycles.
From a dealmaking perspective, the division of responsibility is equally interesting. Iktos is effectively being funded to operate the discovery engine, while Servier takes over once programs are sufficiently advanced. It is a structure that gives pharma access to specialized discovery infrastructure without requiring it to build the same capability internally.
Iambic and Takeda: Combining Platform Access with Program-Level Economics
In February 2026, Iambic entered a multi-year AI drug discovery collaboration with Takeda focused initially on oncology and gastrointestinal and inflammation programs. Takeda also receives access to Iambic’s NeuralPLexer technology, which predicts protein-ligand complex structures, alongside its broader AI-enabled discovery and automated wet-lab capabilities.
The commercial structure combines several payment types. Iambic receives upfront payments, research funding and technology-access fees, while also being eligible for success-based payments that could exceed USD 1.7bn, plus royalties on assets arising from the collaboration.
That makes the deal more than a straightforward asset licence. Takeda is paying for access to both a technology platform and collaborative discovery capacity, while much of the economics remain tied to whether individual programs advance successfully.
Relation Therapeutics and GSK: When the Data Become the Asset
The July collaboration between Relation Therapeutics and GSK broadens the definition of what can actually be monetized in an AI drug discovery partnership. Relation may receive up to USD 110m in upfront and success-based milestone payments to generate large-scale human cellular perturbation datasets. That data will be used to improve understanding of disease biology and to develop, validate and train AI foundation models, including Relation’s MORGAN model.
Unlike a conventional licensing deal, however, the immediate output is neither a molecule nor even necessarily a defined drug program. It is a proprietary biological dataset designed to make AI models better at identifying therapeutic opportunities.
As pharma companies develop more sophisticated internal AI capabilities, high-quality proprietary biological data may become every bit as strategically important as access to algorithms themselves.
AI Deal Structures Are Starting to Diverge
These agreements suggest that there is no single emerging template for AI drug discovery partnerships. The common theme is that pharma is increasingly paying not simply for “AI”, but for specific capabilities that can plug into different parts of the R&D process.
That may ultimately be a more important sign of maturity than the size of any individual deal. As AI-enabled drug discovery moves further into mainstream biopharma partnering, the technology is becoming less of a category in itself, and more of another asset around which sophisticated licensing and collaboration structures can be built.