Lilly Just Bet $2.75 Billion That a Robot Can Design Better Drugs

The second course is molecule design. Chemistry42, the generative chemistry engine, doesn't search through existing chemical libraries the way traditional drug discovery does. Instead, it creates brand-new molecules from scratch, like an architect drawing a building that's never existed before. It designs them to bind the right target, dissolve in your gut, avoid toxic side effects, and actually be manufacturable. A foundation model called Nach01 acts as the "chemistry brain" powering predictions across hundreds of tasks.

The third course is clinical prediction. inClinico forecasts which trials are likely to succeed, which patient populations to target, and how to design studies that maximize the odds.

The result? Insilico claims it can go from identifying a target to nominating a drug candidate in 12 to 18 months, testing only 60 to 200 molecules along the way. Traditional drug discovery often takes four to six years and requires screening thousands of compounds.

The Scoreboard So Far

Insilico isn't just selling software and promises. The company has 28 AI-designed drug candidates in its pipeline, with roughly half now in clinical trials.

The headliner is rentosertib, an AI-designed molecule for idiopathic pulmonary fibrosis (IPF), a lung-scarring disease with limited treatment options. In Phase 2a testing, patients on the high dose gained 98.4 mL of lung capacity while placebo patients lost 20.3 mL. That's not a home run yet, but it's a solid double in a disease where any improvement matters.

Another candidate, garutadustat for inflammatory bowel disease, entered Phase 2a in January 2026. The pipeline keeps growing.

Still, there's a critical asterisk: no AI-discovered drug has ever been approved. Across the entire industry, roughly 173 AI-enabled programs were in clinical trials by early 2026. Approvals? Zero. And about 70% of Phase 2 programs fail industry-wide, a stat that applies to AI-designed drugs too. Some high-profile AI players, including BenevolentAI and Recursion, have already stumbled.

Why Lilly Is Willing to Pay Before the Proof

So why would Lilly write a $115 million check for drugs that might never work, with billions more potentially on the line?

Because the math of traditional R&D is brutal. Developing a single drug costs an average of $2 billion and takes over a decade. If AI can shorten timelines, reduce costs, and increase the number of shots on goal, even a modest improvement in success rates could be worth tens of billions.

Lilly is also building its own AI infrastructure. In October 2025, the company partnered with NVIDIA to deploy advanced AI computing capabilities, and in January 2026 announced an expansion into a co-innovation AI lab, with planned investment of up to $1 billion over five years. That's the "build" side of a "build and buy" strategy: invest internally for long-term capability while locking in access to the best external platforms.

The deal structure is smart too. With $115 million upfront and $2.63 billion in milestones, most of Lilly's financial exposure only triggers if the drugs actually work. Insilico keeps massive upside if they succeed. Both sides have skin in the game; neither side is overexposed if things go sideways.

The Bigger Picture: AI Leaves the Sandbox

This deal matters beyond Lilly and Insilico. It's a signal to the entire pharmaceutical industry.

For years, Big Pharma treated AI drug discovery like a promising intern: interesting to have around, worth a small investment, but not trusted with anything critical. The Lilly deal flips that script. A top-five pharma company is now licensing actual drug candidates generated end-to-end by AI, not just using AI tools to speed up one step in the process.

Analysts describe this as the shift from experimentation to scaled deployment. Investors are reading this as a growth lever for Lilly's long-term pipeline and a validation milestone for the entire AI drug discovery sector.

The pattern emerging across the industry is clear: AI-native companies handle the "zero to one" of molecule creation, while pharma giants handle the expensive, complex work of clinical trials, regulatory approval, and global commercialization. It's a division of labor that plays to each side's strengths.

Of course, the ultimate test is still ahead. Until an AI-designed drug gets approved and helps real patients, the skeptics have a point. But with $2.75 billion on the table and clinical data starting to trickle in, Lilly is betting that the question isn't whether AI will transform drug discovery. It's how fast.

The Italian Family That Just Spent $4.1 Billion to Break Into America

A 107-year-old Italian family pharma just dropped $4.1 billion on a Florida rare disease company, paying roughly three times its own annual revenue. The deal gives Angelini instant access to the U.S. market and a portfolio generating nearly $600 million a year.