Insilico Opens BIO 2026 With a $2.5 Billion CNS Deal, and It Is Heavily Backloaded 🔗

Insilico Medicine signed a research collaboration with Korea's SK Biopharmaceuticals worth up to $2.5 billion to discover AI-enabled drug candidates for neuroimmune disorders of the central nervous system. Insilico's Pharma.AI platform handles target validation, generative chemistry, and molecule optimization, while SK steers late-stage development and commercialization, building on its experience with the epilepsy drug cenobamate. Insilico receives up to $18 million in upfront and near-term payments against the more than $2.5 billion in total potential value, a heavily backloaded arrangement that pairs genuine enthusiasm for AI discovery with real caution about how many of these candidates will survive the clinic. It builds directly on the first-in-human dosing milestone we covered last week.

NVIDIA Ships an Agentic Toolkit for Life Sciences 🔗

At BIO 2026, NVIDIA launched the BioNeMo Agent Toolkit, which packages more than a decade of its life sciences libraries and models into agent-callable tools for protein structure prediction, molecular docking, generative chemistry, and genomic analysis. More than 50 companies are already building on it, including Lilly, Schrödinger, Databricks, Snowflake, Benchling, and Owkin, with Anthropic and OpenAI integrating it into their agents. The pitch is that general-purpose agents waste time finding and operating scientific tools, and that giving them domain-specific skills is what moves them from answering questions to running multi-step research.

Boehringer Ingelheim Buys Into Immunai's Single-Cell AI 🔗

Boehringer Ingelheim signed a discovery collaboration with New York-based Immunai worth up to $15 million through 2027, applying Immunai's single-cell AI platform to thousands of patient samples to find shared patterns of T-cell dysfunction across cancer and autoimmune disease. The first phase is explicitly about building a shared data foundation across both disease areas before any target work begins, and promising candidates move to wet-lab validation in Immunai's New York facility. It is Immunai's third Big Pharma pact of 2026, after Bristol Myers Squibb in January and an expanded AstraZeneca deal in May, and the company now works with eight of the top 20 pharma firms.

Komodo Scales Its AI Platform Across Alnylam's Enterprise 🔗
Komodo Health expanded its partnership with Alnylam to scale Marmot, its analytics AI platform, across the company's commercial and broader enterprise functions. Marmot runs on Komodo's Healthcare Map, a view of more than 330 million de-identified patient journeys, and Alnylam reports cutting reporting cycles from months to hours after integrating its enterprise datasets and building custom AI agents on top.

The signal: This is the data-foundation thesis playing out on the commercial side rather than in discovery. Komodo's CEO framed Marmot as a new operating model rather than a productivity tool, and the value is coming from a single de-identified data substrate that replaces fragmented dashboards. The pattern matches what surfaced at BIO. Organizations that move past AI experimentation are the ones that unified and governed their data first, then put agents on top.

Lilly Takes an Equity Stake in an AI-Designed Antibody 🔗

Absci raised roughly $100 million in a stock offering led by Eli Lilly, which contributed $40 million for equity, with participation from BVF Partners, Adage, and others. The money advances ABS-201, an AI-designed anti-PRLR antibody that Absci is developing for pattern hair loss and endometriosis, on the back of positive early Phase I safety data.

The signal: Pharma is increasingly choosing equity stakes over straight licensing to get inside AI-native discovery platforms, which gives it exposure to the platform rather than a single molecule. The target here is also notable. An AI-designed antibody aimed at large consumer and women's health markets shows generative design moving out of rare oncology niches and toward indications with very different commercial logic.

Europe Moves on Its Own AI Base, in Compute and in Law 🔗

At ISC High Performance in Hamburg, NVIDIA said 35 AI supercomputers are now in development across Europe, spanning 23 countries and around 800 AI exaflops of capacity deployed or announced over the past year, with healthcare and biomedical research among the named use cases and Germany's first AI factory, HammerHAI, in Stuttgart. In parallel, the proposed European Biotech Act is moving through the EU legislative process, aiming to shorten multinational clinical trial timelines, create regulatory sandboxes for AI, and harmonize the legal basis for trial data.

The signal: At BIO, the competitiveness anxiety was about the US and China. The European response is taking shape on two tracks at once, building domestic compute and rewriting the rules that govern biotech and health data. The backdrop is stark. The EEA's share of commercially sponsored clinical trials fell from 22% in 2013 to 12% in 2023, while China's rose from 5% to 18%. For anyone operating in the DACH region or planning European trials, this is the early scaffolding of a market that is being constructed now, with the European Health Data Space sitting underneath it.