Open-source AI models are fundamentally changing how researchers approach drug discovery and genetic research, moving the field away from proprietary black boxes toward collaborative, transparent science. At the International Conference on Machine Learning (ICML) 2026, the trend became unmistakable: hundreds of papers now build on open AI infrastructure rather than developing isolated tools, signaling a shift in how modern biomedical research gets done.

Why Are Open Models Becoming the Foundation for Biology Research?

The numbers tell the story. NVIDIA reported that approximately 2,000 accepted papers at ICML 2026 cite NVIDIA GPUs, while 145 papers specifically use NVIDIA BioNeMo, an open model family designed for biomedical research. This isn't just about having more tools available; it's about democratizing access to cutting-edge AI infrastructure that would otherwise cost millions to develop independently.

Open models function as a research stack rather than a single product. Researchers gain access to open weights (the underlying mathematical parameters), open datasets for training, and open recipes for reasoning, tool use, and data curation. This transparency allows scientists to understand how models work, adapt them to their specific problems, and build on each other's progress rather than reinventing solutions.

The ecosystem is already producing real-world applications. Basecamp Research developed EDEN, a DNA foundation model that helps researchers interpret and design genetic sequences. Merck & Co. uses KERMT, another open BioNeMo model, to predict how potential drug molecules behave in the body, including whether they're likely to be effective and safe. These aren't theoretical exercises; they're being deployed in actual pharmaceutical research pipelines.

How Are Pharma Companies Scaling AI-Driven Drug Discovery?

The pharmaceutical industry is investing billions to embed AI into its core operations. In May 2026, Isomorphic Labs announced a $2.1 billion funding round, with backing from Thrive Capital and partnerships with Novartis, Eli Lilly, and Johnson & Johnson. The company's platform, called IsoDD (Isomorphic Labs Drug Design Engine), expands the druggable landscape by identifying cryptic binding pockets that remain hidden under normal conditions but open when a specific molecule interacts with a protein.

This represents a fundamental shift in how drugs are discovered. Traditional approaches focus on known binding sites revealed by structural biology, but AI-driven platforms can probe previously inaccessible biology and predict induced-fit interactions where proteins change shape upon ligand binding. The platform also works across multiple drug modalities, including de novo antibodies and other large biologics.

Beyond Isomorphic, the wave of partnerships reflects industry-wide conviction in AI's potential. Genesis Molecular AI and Incyte announced an expanded collaboration worth potentially over $1 billion to apply the GEMS platform for protein-ligand structure prediction across difficult targets in Incyte's pipeline. Chai Discovery secured a licensing agreement with Pfizer for early access to Chai-3, its AI model for de novo antibody design, plus a custom model trained on Pfizer's proprietary data. Eli Lilly selected Tamarind Bio to host inference infrastructure for TuneLab 2.0, a federated AI/ML drug discovery platform that gives biotech partners access to models trained on Lilly's proprietary data.

Steps to Understand How AI Models Are Advancing Genomics Research

• Foundation Models for Biology: Open models like BioNeMo and EDEN provide researchers with pre-trained AI systems that understand protein structures, genetic sequences, and molecular properties, eliminating the need to build these capabilities from scratch.
• Synthetic Data Generation at Scale: Tools powered by open models enable researchers to create high-quality training datasets without relying solely on human-labeled data, accelerating research timelines that would have been impractical just years ago.
• Collaborative Research Infrastructure: Open datasets, recipes for data curation, and reproducible training methods give researchers a transparent foundation for developing and validating new models across institutions.
• Multi-Modal Drug Design: AI platforms can now predict protein behavior, identify hidden binding sites, design antibodies, and model how molecules interact with the body, addressing multiple stages of drug discovery simultaneously.

The shift toward open models is also driving innovation in specialized applications. Inceptive, backed by Obvious Ventures, is developing foundation models for sequence-based medicines like RNA interference therapies that silence disease-causing genes. The company recently announced a collaboration with Alnylam Pharmaceuticals worth up to $2 billion to advance small interfering RNA design by modeling target messenger RNAs while exploring novel chemical modifications.

Another Obvious-backed company, Inductive Bio, combines AI chemistry assistants with predictive models for absorption, distribution, metabolism, excretion, and toxicity (ADMET) and pharmacokinetics (PK), plus human-relevant digital organ technologies to surface risks earlier. The platform gained external validation in February 2026 when it placed first in the OpenADMET-ExpansionRx blind challenge, a benchmarking competition predicting properties of previously unseen compounds from real-world drug programs.

"A model that's accurate but doesn't change the pace or probability of success in the clinic is meaningless," said Rohan Ganesh, a partner at Obvious Ventures.

Rohan Ganesh, Partner at Obvious Ventures

This emphasis on clinical outcomes reflects a broader maturation in how investors evaluate AI biotech companies. Rather than betting on computational promise alone, leading venture firms are backing companies that can demonstrate tangible improvements in drug discovery timelines and success rates.

The infrastructure moment for AI-driven drug discovery continues to accelerate, with billion-dollar investments flowing into end-to-end platforms driven by models and compute rather than single drug assets. Underpinning this trend is the proliferation of AI reasoning workflows that accelerate biomedical research and large integrated datasets spanning genomics, transcriptomics, proteomics, and metabolomics, enabling more powerful models of biological complexity for a new era of programmable therapeutics guided by prediction and rational design.

As the AI biology ecosystem grows increasingly crowded, differentiation comes from pursuing problems that others cannot tackle and owning business outcomes that matter in the clinic. The companies winning investment are those that combine cutting-edge AI with deep domain expertise in biology, regulatory pathways, and clinical validation.