Bessemer’s biotech roadmap

Bessemer Venture Partners says the next biotech winners will be built less on standalone models and more on the data systems that feed them. (bvp.com) In an April 12, 2026 post, Bessemer’s biotech team argued that falling compute costs and maturing models have shifted the bottleneck in drug discovery toward data infrastructure, not access to algorithms. The firm said “biology-native data,” agentic workflows, and lab automation feedback loops will define the next generation of biotech companies. (bvp.com) The basic problem is speed and attrition. Bessemer wrote that moving from target identification to a clinical candidate still often takes more than five years, that nearly 90% of drugs entering clinical trials fail, and that research and development costs per approved therapy continue to double every nine years. (bvp.com) Its roadmap centers on four building blocks: data stacks designed for biological experiments, multi-modal datasets that combine formats like sequences, images, and lab readouts, software agents that automate research tasks, and closed-loop labs where experiment results flow back into the model for the next round. Bessemer framed that setup as the operating system for artificial-intelligence-led drug discovery. (bvp.com) Bessemer tied that argument to recent drug-development results. The firm pointed to Insilico Medicine’s June 2025 Nature Medicine paper on rentosertib in idiopathic pulmonary fibrosis and said it marked the first clinical proof of concept for a drug whose target and molecule were both generated with artificial intelligence. (bvp.com; nature.com) Bessemer wrote that Insilico nominated a preclinical candidate after screening 78 molecules rather than thousands and did so in 18 months at less than 10% of the average cost per approved drug. Insilico said the Phase 2a results were published on June 3, 2025 and presented at the American Thoracic Society 2025 meeting. (bvp.com; insilico.com) The market is also moving from experiments to commercial deals. Bessemer cited early-2026 agreements in which GlaxoSmithKline committed $50 million in upfront capital and near-term milestones for Noetik’s oncology foundation models, while Eli Lilly partnered with Chai Discovery on biologics design. (bvp.com; businesswire.com; biospace.com) That fits a broader shift in pharma spending. In a separate March 11, 2026 roadmap, Bessemer said pharmaceutical companies collectively spend more than $150 billion across service providers and software for drug discovery, clinical trials, operations, regulatory affairs, and commercialization, and argued that artificial intelligence agents are starting to automate parts of that outsourced work. (bvp.com) Other analysts have described the field as moving past early hype but still uneven in execution. A 2023 Nature feature said investment and partnerships had surged around artificial-intelligence drug discovery, while warning that companies still faced operational and scientific challenges as the sector tried to convert platform claims into measurable impact. (nature.com) Bessemer’s post lands as the firm has been publicly arguing that 2026 is an inflection year for artificial intelligence in biotech. Axios reported on January 26, 2026 that partner Andrew Hedin said recent deals were fueling market enthusiasm and shaping Bessemer’s investment strategy. (axios.com) The thread running through Bessemer’s roadmap is that the durable advantage may sit below the model layer. If that view holds, the companies that own the best experimental data, the cleanest feedback loops, and the fastest lab-to-model cycle will have the strongest claim on the next wave of biotech spending. (bvp.com)