From improved molecular design to AI-driven automation, waves of AI tools promise to accelerate therapeutic development. Still, skeptics ask whether this momentum reflects real transformation or simply hype in drug discovery.
Mike Nally, CEO of Generate:Biomedicines, emphasizes that the promise of AI lies in delivering better medicines to patients faster. What is typically a 10–15-year journey from discovery to clinical approval, he says, could potentially be compressed into an eight-year paradigm with AI technologies.
Founded in 2018, the Flagship Pioneering company recently completed one of the industry’s largest IPOs in years, raising $400 million in gross proceeds toward clinical trials and R&D efforts. Generate’s AI platform includes an optimization stack guided by existing molecules, and a second layer that designs proteins from scratch.
Generate’s optimization approach gives machine learning models an existing therapeutic binder as a starting point. Researchers then computationally learn the functional landscape to improve the molecule for clinical drug properties. The company’s current workflows can complete optimization rounds within a couple of weeks, while three rounds of design optimization, on average, are sufficient to reach the desired criteria.
Nally tempers that new computational tools are not a panacea for every part of drug development.
“If you pick the wrong target, dose, or patient population, no technology will overcome those things,” he said. “If you have a transformational technology, you have to first prove the technology works in the clinic.”
In December, Generate announced that its most advanced program, GB-0895, an anti-thymic stromal lymphopoietin (TSLP) antibody for severe asthma, entered Phase III clinical trials. GB-0895 is the first “AI-derived” antibody to reach this clinical milestone, progressing from discovery to Phase III within five years. The two global studies, SOLAIRIA‑1 and SOLAIRIA‑2, will evaluate GB-0895 in approximately 1,600 adults and adolescents with severe asthma.
More shots on goal
In preclinical discovery, researchers grapple with identifying the right molecule to advance to the clinic. AI promises to accelerate multiple stages of the pipeline—from target discovery and molecular design to clinical validation.
“Have we seen a big impact yet? We are still not there, especially on the research side,” said Sai Jasti, SVP, head of data science and AI at Bayer, when describing the role of AI platforms. He says Bayer has an internal goal to increase R&D productivity by 40% in 2030.
To achieve this aim, Bayer has entered a three-year strategic collaboration with Cradle, an AI-based protein design company, to accelerate protein optimization across Bayer’s therapeutic antibody pipeline. By reducing the number of optimization cycles and improving developability properties, including potency, safety, and manufacturability, Cradle’s platform will expand Bayer’s biologics portfolio.
Anastasia Hager, PhD, head of drug discovery sciences, SVP, pharma R&D at Bayer, says the company has been on a journey to replenish its early pipeline given growth across multiple indications, including cardiovascular, immunology, oncology, and neurodegenerative disease.
“The most exciting piece is enabling creativity to test different sequences while unlocking target and structural space,” said Hager when describing the value-add of Cradle’s platform. “We’re committed to enhancing our pipeline with data science and AI as key tools for the biologics portfolio.”
When moving molecules through the clinic, Stef van Grieken, co-founder and CEO at Cradle, says translation is not a solved problem. Yet, the ability to more quickly identify candidates early in the development pipeline while increasing biological understanding of the target will “provide more shots on goal” to support informed decisions during drug development.
In that vein, Jasti emphasizes that the partnership is “not just a software deal,” but a collaboration on the scientific and machine learning level, where scientists from both organizations will exchange ideas and expertise on research direction. Cradle’s platform will also be embedded in Bayer’s workflows to improve accessibility for scientists in the lab.
Year of deployment
While traditional tech-pharma collaborations focus on a small number of drug targets, 2026 kicked off with a stream of AI platform deals across pharma, signaling a cultural shift of investing in AI infrastructure for broad discovery.
“If 2025 was the year of breakthrough research, we believe 2026 will become the year of deployment,” said Jack Dent, co-founder at Chai Discovery, an AI-driven biologics company developing therapeutics against undruggable targets.
Chai’s core technology centers on Chai-2, a de novo antibody design model capable of generating full-length antibodies with therapeutic attributes. The model speeds up workflows by reducing reliance on labor-intensive and time-consuming experimental screens.
Earlier this year, Chai announced a collaboration with Eli Lilly to deploy Chai’s technology to design novel biologics for multiple targets. Chai will also develop an exclusive AI model for Lilly that is trained on the pharma giant’s proprietary data and tailored to Lilly’s discovery workflows.
According to Aliza Apple, PhD, vice president of Lilly Catalyze360 AI/ML and global head of Lilly TuneLab, Lilly’s tech philosophy centers on being an early adopter of promising tools. She emphasizes that models must be trained on quality data and undergo rigorous testing to design better molecules.
“We want to lean in early to the tools that look truly differentiated and put Lilly’s weight behind them, not just rely on what we’ve already built,” said Apple. Rather than outsourcing therapeutic design, the collaboration with Chai gives Lilly scientists direct access to Chai’s generative AI capabilities.
From molecules to humans
Other biotechs have focused their efforts entirely on building platforms, rather than developing drugs internally. Boltz, an AI research and product company, launched in January with $28 million, with a mission to advance open science for drug discovery.
The public benefit corporation (PBC) is co-founded by MIT researchers—Gabriele Corso, PhD, Jeremy Wohlwend, PhD, and Saro Passaro, known as the developers of the widely adopted Boltz series of models. The Boltz team first made waves in November 2024, with the launch of the co-folding model, Boltz-1, a fully commercially available AI model to achieve AlphaFold 3-level accuracy in predicting the 3D structure of biomolecular complexes.
Boltz has already solidified a multi-year collaboration with Pfizer to build exclusive models that improve target selection for structure prediction, small-molecule affinity, and biologics design. Boltz scientists will also build custom models and workflows with Pfizer for a number of target programs to enhance preclinical decision-making.
Corso, who leads Boltz as CEO, said two realizations drove the decision to turn the Boltz models into an enterprise.
“First, continuing to push the frontier of biomolecular AI requires sustained investment in talent, compute, and data at a scale not attainable within academic environments,” Corso explained. “Second, truly democratizing the technology, and maximizing its impact, means going beyond publishing models and building reliable, well-designed products that scientists could integrate directly into their daily work.”
While much of the industry is focused on molecular design, San Francisco-based Noetik is building biological foundation models trained on human data.
Ron Alfa, MD, PhD, CEO of Noetik, emphasizes that a huge gap remains for large-scale translational data, preventing drugs from succeeding in the clinic. As an answer, the company generates multimodal data from human tissue samples with an intact in vivo context. These data fuel Noetik’s foundation models, which predict clinical outcomes in cancer.
Earlier this month, Noetik announced a five-year licensing partnership with GSK, which gives the pharma giant access to Noetik’s non-small cell lung cancer and colorectal cancer models. The deal includes a $50-million upfront payment and will follow a subscription-based framework.
Alfa describes the GSK partnership as one of the first true foundation model licensing deals in biotech. “For years, the sector has looked for a way to commercialize AI as infrastructure rather than the standard R&D collaborations,” he said. “Now, we have a template.”
Whether AI-driven drug discovery is reality or hype, rising pharmaceutical investment points to the former. The ultimate payoff, however, will be decided in the clinic.
The post Pharma’s AI Investment Signals New Drug Discovery Paradigm appeared first on GEN – Genetic Engineering and Biotechnology News.
