Moonshot, which is the use of artificial intelligence used to promote drug development, is back on Earth.
More than $18 billion has driven development to promote AI in over 200 biotech companies, with 75 drugs or vaccines taking part in clinical trials, according to the Boston Consulting Group. Now, investors' trust and fundraising are beginning to waver.
In 2021, venture capital investments in AI pharmaceutical companies reached a peak, with over 40 transactions worth around $1.8 billion. This year there have been less than 20 transactions worth about half that peak amount. Financial Times Reported citing data from Pitchbook.
Some existing companies struggle with challenges. In May, Biotech Company Recussion presented three future drugs in cost-cutting efforts last year following its merger with similar biotech company Exscientia. luck Previously, we reported that the recursive discovered AI compound has not reached the market as an approved drug. After a major restructuring in December 2024, biotechnology company Benevolentai went public in March on the Euronext Austerdam Stock Exchange before merging with Osaka Holdings.
A spokesman for recursion said luck The decision to shelved the drug was “data-driven” and was a planned outcome of the merger with Excientia.
“The 90% failure rate in our industry is unacceptable when patients are waiting. We believe that integrating cutting-edge tools and technology is best positioned for long-term success,” the spokesperson said in a statement.
Benevolentai did not respond to requests for comment.
The industry struggle coincides with a broader conversation about the failure of generative AI, which fails to realize its lofty promise of productivity and efficiency more quickly. Last month, a MIT report found that 95% of corporate generation AI pilots were unable to accelerate their revenues. A survey by the US Census Bureau this month found that AI adoption in large US companies has declined from a peak of 14% this year to 12% in August.
However, the AI technology used to help drug development is very different from the large-scale language models used in most workplace initiatives, so the same standards should not be maintained, according to Scott Schoenhaus, managing director and equity research analyst at Keybanc Capital Markets Inc. Instead, the industry faces its own challenges.
“No matter how much data you have, human biology is still a mystery,” Schoenhouse said. luck.
Political factors that eliminate macros and AI drug development funds
Schoenhaus said slower funding and slower development results could be a number of broader factors rather than limiting the technology itself.
“Everyone admits that the financial environment has dried up,” he said. “The biotech market is heavily affected by low interest rates. Low interest rates equal increased funding for biotechnology, and we have seen biotech funding at record low prices over the past few years as interest rates have risen.”
It wasn't always the case. Using AI in drug development not only increases access to semiconductor chips, but also how technology has enabled rapid and inexpensive ways to map the entire human genome. In 2001, it cost over $100 million to map human genomes. Twenty years later, it cost about $1,000.
Beyond having a pandemic to thank the next interest rate in 2021, Covid has also promoted a partnership between AI drug development startups and major pharma companies. In early 2022, biotech startups Abcellera and Eli Lilly received emergency FDA approval for antibodies used in early Covid vaccines.
But there have been hurdles for other industries since then, Shanehouse said, including the reduction in research and development costs during slowing demand and the uncertainty surrounding President Donald Trump's US and European Union will impose tariffs on drugs over trade contracts. Trump signed a memo this week that threatened to ban direct consumer advertising for prescription drugs and theoretically lowered Pharma's revenue.
AI Limitations
That doesn't mean there were no technical hiccups in the industry.
“There's scrutiny about the technology itself,” Shanehouse said. “Everyone is waiting for these reads to prove it.”
According to Schoenhaus, the next 12 months of emerging data from AI Drug Developments Startups will be important in determining how successful these companies are. Some of the results so far are mixed. For example, Recursion published data from a medium-term clinical trial of a drug to treat neurovascular conditions last September, finding that while the drug is safe, there is little evidence of how effective it is. The company's shares fell double digits after the announcement.
These companies are also limited by their AI-powered methods. The drug development process takes 10 years, and is intentionally bottlenecked to ensure the safety and efficacy of the drug in question. Biotechnology companies using AI to make these processes more efficient are usually only addressing small parts of this bottleneck, such as being able to screen and identify drug-like molecules faster than before.
“There are so many stages that you have to jump over before you actually declare it. [European Medicines Agency]or the FDA, or Health Canada, will designate this as a safe, approved drug to sell to patients around the world, whoever it is,,” Siderovski said. luck. “The early bottleneck of audition compounds is that they are all satisfied shareholders and not to end the issue by announcing that they have approved this compound as a drug.” ”
SMEs also preferred to work with partners with major pharma companies to build their own pipelines instead, even if they had no access to the industry giant's franchise resources.
“They want to pursue their technology and show their platform verification faster,” Shanehouse said. “They aren't going to wait for big pharma to pursue the molecules they've partnered with. They just do it themselves and say, 'Hey, our technology platform will work.' ”
Schoenhaus sees this strategy as a way to perfect the use of AI and try to prove themselves by better understanding the slippery, mystical, yet very unknown frontiers of human biology.
“This is a very complex application of AI,” he said.
