Converge Bio raises $25 million with support from Bessemer and executives from Meta, OpenAI, and Wiz

Artificial intelligence is rapidly moving into the drug discovery field as pharmaceutical and biotech companies look for ways to shave years off research and development timelines and increase the likelihood of success amid rising costs. More than 200 startups are currently competing to integrate AI directly into research workflows, and interest from investors is growing. Converge Bio is the latest company to secure fresh capital and capitalize on the change as competition intensifies in the AI-driven drug discovery space.

The Boston and Tel Aviv-based startup, which uses generative AI trained on molecular data to help pharmaceutical and biotech companies develop drugs faster, has raised an oversubscribed $25 million Series A round led by Bessemer Venture Partners. TLV Partners and Vintage Investment Partners also participated in the round, with additional support from unidentified executives from Meta, OpenAI, and Wiz.

In fact, Converge trains generative models based on DNA, RNA, and protein sequences and incorporates them into pharmaceutical and biotech workflows to accelerate drug development.

“The drug development lifecycle has defined stages, from target identification and discovery to manufacturing, clinical trials, and beyond, and at each stage there are experiments that we can support,” Converge Bio CEO and co-founder Dov Gertz said in an exclusive interview with TechCrunch. “Our platform continues to expand across these stages, helping us bring new medicines to market faster.”

So far, Converge has rolled out a customer-facing system. The startup has already deployed three separate AI systems. One for antibody design, one for protein yield optimization, and one for biomarker and target discovery.

“Take our antibody design system as an example. It’s not just a single model; it’s made up of three integrated components. First, a generative model creates new antibodies. Then, a predictive model filters those antibodies based on molecular properties. Finally, a docking system using a physically-based model simulates the three-dimensional interactions between the antibody and its target,” Gertz continued. According to the CEO, the value lies in the entire system, not a single model. “Our customers don’t have to put the models together themselves; they get a ready-to-use system that connects directly to their workflow.”

The new funding comes about a year and a half after the company raised a $5.5 million seed round in 2024.

Since then, the two-year-old startup has rapidly scaled up. Gertz said Converge has 40 partnerships with pharmaceutical and biotech companies and currently has about 40 programs running on its platform. We work with customers in the US, Canada, Europe, Israel, and are currently expanding to Asia.

The team has also grown rapidly, increasing from just nine employees to 34 in November 2024. Along the way, Converge has started publishing case studies. In one example, the startup helped partners increase protein yield by 4-4.5 times in a single computational iteration. In another example, the platform has generated antibodies with very high binding affinities reaching the single nanomolar range, Gertz said.

There is growing interest in drug discovery using AI. Last year, Eli Lilly partnered with Nvidia to build what the company calls the pharmaceutical industry’s most powerful drug discovery supercomputer. And in October 2024, the developers of Google DeepMind’s AlphaFold project were awarded the Nobel Prize in Chemistry for developing AlphaFold, an AI system that can predict protein structures.

Asked about this momentum and how it is shaping Converge Bio’s growth, Gertz said the company is witnessing the biggest financial opportunity in life sciences history, with the industry moving away from a “trial and error” approach to data-driven molecular design.

“We feel that momentum deeply, especially in the inbox. When we founded the company a year and a half ago, there was a lot of skepticism,” Gertz told TechCrunch. That skepticism has dissipated surprisingly quickly, he added, thanks to success stories from companies like Converge and academia.

Large-scale language models have gained attention in the drug discovery field due to their ability to analyze biological sequences and suggest new molecules, but challenges such as hallucinations and accuracy remain. “With text, hallucinations are usually easy to spot,” the CEO said. “For molecules, validation of a new compound can take weeks, so the costs are much higher.” To address this, Converge combines generative and predictive models to filter new molecules to reduce risk and improve outcomes for partners. “While this filtration is not perfect, it significantly reduces risk and provides better outcomes for our customers,” Gertz added.

TechCrunch also asked about experts like Yann LeCun who remain skeptical about using LLMs. “I’m a big fan of Yann LeCun and completely agree with him. We don’t rely on text-based models for our core scientific understanding. To truly understand biology, we need to train models on DNA, RNA, proteins, and small molecules,” Gertz explained.

Text-based LLMs are used only as a support tool, for example, to help customers navigate the literature regarding generated molecules. “Those are not our core technologies,” Gertz said. “We are not tied to a single architecture; we use LLM, diffusion models, traditional machine learning, and statistical techniques as appropriate.”

“Our vision is for every life sciences organization to use Converge Bio as a generative AI lab. There will always be wet labs, but they will be combined with generative labs that computationally create hypotheses and molecules. We want to be that generative lab for the entire industry,” Gertz said.