Skip to content
Developer312

Novo Nordisk Says AI Cut Its Molecule Design Cycles From Eight to Three

Drugmakers told Reuters AI is shortening early discovery. Novo Nordisk says design-make-test cycles fell from about eight to three, each from eight weeks to four. The figures are company-reported, and AI drugs in clinical testing have risen only modestly.

Published Report an error

Big drugmakers are putting numbers on what AI has done for early discovery. In a Reuters report published November 20, scientists at GSK, Bristol Myers Squibb, Novartis, Eli Lilly, Novo Nordisk, AstraZeneca and Sanofi said AI has cut by as much as half the time it takes to produce new molecules and move them into the clinic.

Key Takeaways: What Matters Most?

  • Novo Nordisk told Reuters AI cut the number of design, manufacture and test cycles for new molecules from around eight to three, and each cycle from eight weeks to four.
  • Novartis says the average time from project start to a drug candidate went from three years to about 18 months to two years; Sanofi says it saved at least a year on asthma treatment lunsekimig.
  • These are company-reported figures. Reuters notes AI-enabled drug candidates in clinical testing rose only from around 117 to 129 over eight months.
  • Sanofi's Matt Truppo told Reuters models break down quickly without training data, and Novo and Sanofi say they do not share proprietary data with frontier labs.

Takeaways like these, every weekday morning. Get the briefing

What Happened

The most specific figures came from Novo Nordisk, which told Reuters that AI cut the number of design, manufacture and test cycles for new molecules from around eight to three, and the time of each cycle from eight weeks to four.

Others gave their own examples, per Reuters:

  • Novartis: biomedical research president Fiona Marshall said average time from project start to a drug candidate went from three years to about 18 months to two years. Novartis also said AI helped it find five targets it prioritized for kidney disease.
  • Sanofi: said it saved at least a year on asthma treatment lunsekimig by skipping a dose-selection study. Sanofi's Matt Truppo expects AI to cut early-discovery timelines by 20% to 25% within two years.
  • Bristol Myers Squibb: said one sickle cell candidate would not have been developed without its AI systems.

AstraZeneca's Sharon Barr told Reuters there is "no part of the discovery through development pipeline where we are not using AI."

Read the Fine Print

These speedups are the companies' own figures, reported to Reuters, not audited results. Reuters itself supplies the counterweight: AI-enabled drugs in clinical testing have climbed only incrementally. In March, researchers identified 129 candidates across 69 companies, up from around 117 eight months earlier, and most remain early-stage. Dr. Waqas Haque of the University of Chicago, who helped compile that data, told Reuters he does not know whether those drugs will be more successful, and does not expect quicker approvals because of trial and regulatory bottlenecks.

Lilly's Aliza Apple told Reuters that claims AI can cure all disease overstate current capabilities. "It would be great for them to cure one, and then let's talk about the rest," she said.

Why It Matters for Builders

The story is about data as much as models. Truppo told Reuters that without training data, a model "breaks down very quickly," and that gains may take time to spread beyond areas where Sanofi has the most proprietary data. Novo and Sanofi say they do not give proprietary data to frontier labs for training. Reuters also notes that Anthropic, which listed this week, is positioning itself as a rival, partner or both to drugmakers.

Practical Takeaways

  • Measure cycle time, not just accuracy. Novo's claim is about iterations and weeks per iteration. That is a metric any team running an experiment loop can track.
  • Treat proprietary data as the moat. The drugmakers credit their own data for results. Builders selling into regulated industries should expect customers to keep that data in-house.
  • Discount single-company speedups. Look for downstream evidence, such as candidates entering trials, before treating a timeline claim as industry-wide.
  • Watch the lab partnerships. Who supplies models and who keeps the data will shape where AI value lands in pharma.

Developer312 covers the AI business signals builders actually need to act on. Get the weekday briefing at developer312.com.

Sources

  1. [1]Big Pharma, with eye on Anthropic, says AI is shrinking drug discovery timelines (opens in a new tab) — Reuters (2026-11-20)

Frequently Asked Questions

Get the next briefing

The lead story each weekday morning, with the key takeaways and a link to the full analysis.

One lead story each weekday morning: what happened, the key takeaways, and a link to the full analysis.

New posts daily, briefing weekday mornings. We use your email only to send the briefing. Unsubscribe anytime. Privacy policy.

Related Articles