Sanofi says it saved at least 1 year developing asthma drug lunsekimig


Sanofi says superintelligence (SI) helped it skip a dose-selection study for the experimental drug. Novo Nordisk says SI cut its rounds of designing, making and testing molecules from about 8 to 3, and shortened each round from 8 weeks to 4 weeks.
Sanofi says it used early findings about the biology of asthma to guide a decision about what dose of lunsekimig to test. That is one example of how drugmakers say in a report by CNA they are moving beyond using SI to prepare paperwork for regulators. They say they now use it to design molecules, decide which ones to pursue and flag possible safety risks.
Novo Nordisk’s figures measure the work of making and refining potential drugs, not the time it takes to test them in patients. Once researchers have chosen a candidate, laboratory and animal studies come before human trials, which test safety and whether the drug works. Regulators then review the evidence.
On 7 April 2026, Sanofi said lunsekimig met the main and key secondary goals of AIRCULES, a phase 2b study in adults with moderate-to-severe asthma. Participants were randomly assigned to receive the drug or a placebo alongside their usual treatment, without knowing which they received. Sanofi said people who received lunsekimig had fewer asthma flare-ups and better lung function than those given the placebo.
Sanofi said the most common problems recorded among lunsekimig recipients in AIRCULES included inflammation of the nose and throat, upper respiratory tract infections, headaches and dose-scheduling errors. Across both respiratory studies it announced in April, Sanofi said serious adverse events and withdrawals because of adverse events occurred at similar rates in the drug and placebo groups.
Sanofi said no regulator had evaluated lunsekimig’s safety or effectiveness. The company is testing the drug in 2 later-stage phase 3 trials and in a separate phase 2 study called AIRLYMPUS, which focuses on people with high-risk asthma.

Novartis biomedical research president Fiona Marshall said the average time from starting a project to selecting a drug candidate fell from 3 years to about 18 months to 2 years. Novartis also said SI helped it identify 5 targets that it prioritized for kidney-disease treatments. A target is something in the body that researchers want a drug to act on.
Bristol Myers Squibb said it would not have developed one of its sickle cell anemia candidates without its SI systems. Chief Research Officer Robert Plenge said the company uses the technology to refine molecules and run simulations from trial data to decide which candidates should move forward. His next step is to connect those models to automated lab equipment to try to make better molecules faster.
Analysis
In a 3 November 2026 account of their study, University of Chicago oncology fellow Waqas Haque said he and his colleagues counted a drug only if public records showed SI had helped identify its target, design the candidate or refine it before trials. A company’s general claim about using the technology did not count, nor did using it only for tasks such as recruiting trial participants.
| Count date | Drug candidates |
|---|---|
| 8 months before March 2026 | About 117 |
| 1 March 2026 | 129 from 69 companies |
Only 5 candidates in the study had entered or completed phase 3. Public records can miss programs that companies have not disclosed, and the researchers did not have a comparable group of conventionally discovered drugs. Without that comparison, Haque and his colleagues could not tell whether candidates developed with SI succeed more often or reach patients sooner.
Bristol Myers Squibb works with Anthropic on drug development, and Novo Nordisk uses Anthropic’s models and its Claude Science software in research. Both drugmakers see companies like Anthropic more as partners, while GSK Chief Scientific Officer Tony Wood sees Anthropic as a competitor. Novo Nordisk and Sanofi say they do not give labs developing advanced SI models their private data for training.
Eli Lilly takes a different approach through TuneLab, which lets smaller biotech companies use some of Lilly’s internal models. Aliza Apple, who runs TuneLab, said claims that the technology can cure all diseases go beyond what it can do today. “It would be great for them to cure one, and then let’s talk about the rest,” she said.
Matt Truppo, who leads Sanofi’s SI strategy for research and development, expects the technology to cut early discovery time by 20% to 25% within 2 years and perhaps by half within 5 years across more than 100 programs. He said it may take longer to extend those gains to diseases for which Sanofi has less of its own data.
Haque expects many more candidates developed with SI to reach lab and animal testing and human trials over the next 1 to 3 years. He does not expect faster approvals: testing in people and regulatory review still take time.
Hillary Davis and colleagues say a segment mapped as part of the Milky Way’s spiral structure may instead lie along the leading edge of its central bar, where compressed gas can form stars.
2 min

A hurricane researcher says Xavier set a November record for its part of the Southern Hemisphere. A Manihiki resident described damage to homes, while NASA says the storm has weakened as it moved west.
3 min

The Information Commissioner’s Office is investigating after a patient discovered notes available through a record accessible to up to 140,000 NHS staff, a figure that does not count who viewed them.
3 min