IQVIA Has 150 AI Agents in Production — Targeting 500 by 2027 🔗

IQVIA confirmed this week that it has deployed over 150 AI agents across its operations, with a target of more than 500 by 2027. Recent acquisitions have extended its drug discovery capabilities, and productivity and automation initiatives are driving margin expansion across clinical and commercial workflows. IQVIA is not a startup signalling ambition — it is the world's largest CRO, embedded in the operations of nearly every major pharma company. Agents in production at IQVIA means autonomous AI systems running on real client data, in regulated workflows, at commercial scale. That is the clearest enterprise signal yet that agentic AI in life sciences has moved from pilot to production.

Medidata Launches Dot — AI Companion for Clinical Trials 🔗

At NEXT New York, Medidata unveiled Dot, its commercial AI companion for clinical trial management. Dot integrates across trial design, site selection, protocol optimisation, and execution monitoring, drawing on data from a large library of historical studies. The headline number: study build timelines compressed from weeks to hours. Medidata also launched capabilities for real-time trial simulation and continuous plan updates as new data arrives. For sponsors and CROs on the Medidata platform, Dot is positioned as the central intelligence layer for how trials are designed and run — not an optional feature but the default way the platform operates going forward.

NVIDIA Commits $2 Billion to Life Sciences AI Infrastructure 🔗

NVIDIA made a $2 billion direct investment in Nebius, an AI cloud provider with explicit focus on life sciences applications including molecular modelling and drug screening. The plan calls for up to 5 gigawatts of NVIDIA-powered compute capacity by 2030 across a global network of AI factory sites. This is not a partnership announcement — it is capital commitment to build the physical substrate on which the next decade of life sciences AI will run. NVIDIA has named life sciences as a primary strategic vertical and is now actively shaping which compute infrastructure the sector standardises on. Decisions made in the next 18 months about which cloud providers and compute stacks to build on will be difficult to reverse.

Waiv Raises $33 Million to Scale AI Precision Testing in Oncology 🔗

Waiv, spun out from Owkin this week, closed a $33 million round co-led by OTB Ventures and Alpha Intelligence Capital. The company delivers clinical-grade AI diagnostic tools for oncology: a breast cancer recurrence risk profiling test, an AI-based BRCA mutation pre-screen, and Destra, a digital pathology platform for integrating AI diagnostics into existing lab workflows. Active pharma partnerships with AstraZeneca and MSD are already operational. What distinguishes Waiv's model is continuous validation — AI models are updated through real-world clinical deployment, not static training sets.

The Signal: Three separate outlets covered the same Waiv fundraise this week — Sifted, Pulse 2.0, and Ventureburn. That level of coordinated coverage across European tech and life sciences media indicates institutional momentum, not just a press release. AI diagnostics is maturing from research adjacency into regulated clinical infrastructure, and capital is beginning to follow at meaningful scale.

Radial Launches With $500 Million to Fix Scientific Infrastructure for the AI Era 🔗
A new nonprofit called Radial, backed by the Astera Institute with over $500 million in funding, launched this week with a precise mission: rebuild the underlying data generation and sharing systems of scientific research so that AI can actually deliver on its potential. CEO Becky Pferdehirt and Astera Institute founder Seemay Chou argue that most AI ventures in life sciences target specific applications, but the infrastructure underneath those applications has not been redesigned for the AI era. Radial will publish all results publicly, including failures.

The Signal: Half a billion dollars committed to the unglamorous layer of scientific infrastructure is a serious bet. The public failure-sharing commitment is rarer than it should be. If Radial succeeds in shifting norms around data transparency and scientific process design, it changes the foundation on which every AI tool in this newsletter is built. Worth tracking as a long-duration story rather than a single funding announcement.

Shadow AI in the Lab — Governance Is Becoming a Product Category 🔗

A detailed piece in Technology Networks this week from Sapio Sciences named a problem becoming acute across life sciences R&D: scientists are adopting AI tools faster than IT and compliance can govern them. The term is "shadow AI" — ungoverned use of AI models in research workflows, without audit trails, without validation, and without documentation of what the AI contributed to a result. The proposed solution is embedding AI directly into electronic laboratory notebooks with validated, traceable computational tools, so AI contributions are captured, auditable, and regulatorily defensible.

The Signal: This connects directly to what IQVIA and Medidata are doing at the enterprise level. As AI agents are deployed across clinical and research operations, the question of what the AI did, when, and under whose authority becomes a regulatory requirement, not a best practice. The ELN is the natural control point — the system of record for scientific reasoning. Vendors who solve governance at that layer, rather than bolting compliance onto separate AI tools, have a structural advantage as regulators begin requiring traceability for AI-contributed scientific evidence.