The race to unlock the next frontier of genetic medicine just took a significant leap forward. Eli Lilly has partnered with Profluent Bio in a deal worth up to $2.25 billion to develop AI-designed recombinases, a class of enzymes that could accomplish what CRISPR cannot: large-scale, precise DNA editing at any location in the genome.
For those of us working at the intersection of AI and healthcare, this partnership signals something important: the pharmaceutical industry is no longer experimenting with AI as a side project. Companies like Lilly are making billion-dollar bets that AI-designed proteins will be the foundation of next-generation therapeutics.
Why Recombinases Matter More Than You Think
CRISPR revolutionized gene editing, but it has limitations. Most genetic diseases do not stem from a single mutation. They result from complex patterns of multiple mutations, deletions, or insertions across the genome. CRISPR excels at making small, targeted cuts, but it struggles with "kilobase-scale" editing: inserting or rearranging large stretches of DNA.
Recombinases are different. These enzymes can cut and rejoin DNA at precise locations, enabling the insertion of entire genes or the correction of multiple mutations in a single editing event. The challenge has always been engineering recombinases that target specific locations in the human genome. Natural recombinases recognize fixed DNA sequences, limiting their therapeutic applications.
This is where AI changes the equation. Profluent has built foundation models trained on protein sequences that can generate novel recombinases designed to target any genomic location. As Ali Madani, Profluent's CEO, put it: "Kilobase-scale DNA editing remains a holy grail in genetic medicine. We believe only AI can create the designer recombinases needed to precisely target any location in the genome."
Profluent's AI Platform: From OpenCRISPR to Custom Recombinases
Profluent is not new to AI-driven protein design. The company made headlines with OpenCRISPR-1, an AI-generated gene editor that differs from natural CRISPR proteins by over 400 mutations. OpenCRISPR-1 is now used by thousands of researchers and several major pharmaceutical companies.
The foundation of Profluent's approach is ProGen3, their suite of generative language models for protein design. These models treat proteins as sequences (similar to how large language models treat text) and can generate novel full-length proteins or redesign specific domains of existing proteins for improved function.
With $150 million in total funding (including a $106 million round co-led by Altimeter Capital and Bezos Expeditions), Profluent has the resources to scale these foundation models. The Lilly partnership provides something equally valuable: a clear path to clinical development and commercialization.
The Deal Structure and What It Signals
Under the agreement, Profluent will use its AI models to design site-specific recombinases for multiple genomic targets. Lilly receives the option to exclusively license and develop these assets through clinical trials and commercialization. Profluent receives an undisclosed upfront payment, committed R&D funding, and is eligible for up to $2.25 billion in milestone payments plus royalties on net sales.
This is not Lilly's first bet on recombinase technology. In January 2026, the company signed a $1.12 billion collaboration with German biotech Seamless Therapeutics to develop recombinase-based therapies for hearing loss. The Profluent deal expands this strategy significantly, targeting diseases with "severe unmet needs" across multiple therapeutic areas.
What does this tell us? Lilly is building a portfolio approach to recombinase technology, partnering with multiple companies while maintaining optionality. The company is also clearly betting that AI-designed proteins will outperform traditional protein engineering methods in speed, specificity, and success rate.
Implications for the AI-Pharma Landscape
The Profluent partnership is part of a broader trend of pharmaceutical companies treating AI as core infrastructure rather than experimental technology. In the same month, Novo Nordisk announced a strategic partnership with OpenAI to deploy AI across drug discovery, clinical trials, manufacturing, and commercial operations.
For practitioners in the UAE and Middle East, these developments have direct relevance. As the region invests heavily in biotech and life sciences (including Abu Dhabi's growing presence in genomics research), understanding how AI is reshaping drug discovery becomes essential.
Three observations stand out:
- Foundation models for biology are maturing. The same architectural approaches that power language models are proving effective for protein design. Companies that build domain-specific foundation models (rather than fine-tuning general-purpose models) are commanding the largest partnerships.
- The bottleneck is shifting. AI can now generate novel protein designs faster than they can be experimentally validated. This shifts the challenge from "can we design this protein?" to "can we test enough candidates quickly enough?"
- Valuation reflects confidence. A $2.25 billion deal for a pre-clinical technology signals that major pharmaceutical companies believe AI-designed therapeutics will reach patients, not just publications.
Looking Forward
The Profluent-Lilly partnership will not produce an approved therapy for years. Regulatory pathways for AI-designed gene editors are still evolving. But the strategic signal is clear: the pharmaceutical industry's largest players are no longer asking whether AI will transform drug discovery. They are making billion-dollar commitments on the assumption that it will.
For those of us building AI systems or advising organizations on AI strategy, this is a reminder that the most transformative applications of AI may not be chatbots or productivity tools. They may be novel proteins that no human designer could have conceived, solving medical problems that no existing technology can address.