In the rapidly advancing world of AI-driven drug discovery, French company Aqemia is trying to chart a unique path by teaching physics to artificial intelligence.
Last week, the company announced a $38 million funding round led by Cathay Innovation, bringing its total funding to over $100 million - a milestone that underscores growing confidence in its innovative approach to pharmaceutical research.
What sets Aqemia apart in the crowded field of AI drug discovery is its novel approach. Rather than training its AI on vast databases of existing molecules, the company teaches its generative AI theoretical physics principles. This allows it to predict how potential drug molecules will interact with disease targets at the atomic level, without needing experimental data as a starting point.
According to CEO and Co-Founder Maximilien Levesque, this approach allows the company to innovate more freely, creating entirely new molecules rather than variations on existing ones.
"We do not put a lot of data into a Gen AI that will then invent molecules that look like what it's been trained on," Levesque said. "Instead, we teach physics. In fact, atomic scale physics to GenAI."
Background
In 2017, during the Start-Ulm competition—a platform for start-ups emerging from the École Normale Supérieure (ENS-PSL) community—Levesque, then a group leader in the ENS-PSL chemistry department – met business angel Olivier Vaury. Vaury introduced Levesque to Emmanuelle Martiano, who had spent a decade in the corporate world at Boston Consulting Group (BCG).
This introduction was pivotal. Martiano's business acumen complemented Levesque's scientific expertise. Their collaboration led to the founding of Aqemia in early 2019, with support from PSL Valorisation and investors like Elaia and the PSL Innovation Fund.
Levesque said the idea behind the company was to leverage the technology he had developed. His main work involved finding an efficient way to test molecules virtually on a computer. With the addition of GenAI, he was able to massively scale that work.
"Generative AI invents many, many innovative molecules," he said. "When I say many, it's in the order of billions of molecules. And they are tested in the computer with theoretical physics at the atomic scale. Because we don't need data to train on, we can invent molecules that have never been invented before.
Business Model
The business model is deliberately different from many AI drug discovery companies. Rather than selling software or services, Aqemia focuses primarily (80%) on developing its internal pipeline of drug candidates.
The company's approach has already yielded promising results. Aqemia currently has three advanced programs targeting different forms of cancer, including solid tumors in the head, neck, thorax, and breast cancers. These programs are currently being tested in animal studies, with clinical trials expected to begin by the end of 2025.
The company has more than 50 programs in various stages of development, an unusually large number that Levesque attributes to their approach: "Since we do not need experimental data to start the program, we are extremely capital efficient."
The goal is to get the original drugs through the initial phases of clinical testing and then license or sell them to pharmaceutical companies that can complete testing, get regulatory approval, and then manufacture them.
Aqemia hopes to advance its most promising candidates through Phase II clinical trials before partnering with larger pharmaceutical companies for Phase III trials and commercialization. Levesque noted that successful Phase II trials typically result in deals worth around $1 billion, split between upfront payments and milestone-based compensation. The company already has a $140 million partnership with Sanofi.
"The only VC-compatible model is to sell IP at some point to a pharma company," Levesque said.
What's Next
The latest round of funding is a €30 million extension of its Series A round, bringing the total to €60 million. This investment was led by Wendel Growth and included money from historic investors Bpifrance, Eurazeo, and Elaia.
The new funding will support several key initiatives. First, Aqemia plans to double its current team of 60 people, creating new medical and regulatory departments to prepare for clinical trials. The company is also expanding globally, starting with a new office in London's Kings Cross area, scheduled to open in January 2025.
The company faces several challenges as it moves forward. Perhaps the most significant is the transition from computer-based drug discovery to clinical trials - a complex and highly regulated process that requires different expertise than the company's current focus on physics and AI. That is part of the motivation for the London expansion.
"The ecosystem in the UK is much more mature than in Paris," Levesque said. "When I say ecosystem, I mean, pharma, biotech, venture capital... So first, the ecosystem, second, the talents to recruit, and third to learn to operate at the global scale."
As the company sets its sights on clinical trials and global expansion, the question now is whether the company prove that its physics-based approach can consistently produce better results than traditional methods or competing AI approaches. That will determine just how profoundly Aqemia's technology can transform drug discovery.
