Share with your CTO
Verily is betting it can become the data infrastructure layer that healthcare AI runs on top of, not just another model vendor. The company secured follow-on investment from NVIDIA and existing backer CU Healthcare Innovation Fund II, extending a March fundraising round at an undisclosed size. The strategic weight here is NVIDIA converting a technology partnership into an equity position, giving Verily access to NeMo, Parabricks, RAPIDS, and B200 GPUs through its Verily Pre platform while tightening NVIDIA’s grip on the healthcare compute stack.
What this means for your business
The story that matters here isn’t the funding, it’s the pattern: NVIDIA taking equity stakes in companies building vertical AI platforms on top of its hardware. Healthcare CTOs evaluating AI infrastructure vendors should treat this as a signal about which platforms will get preferential access to next-generation compute and optimized software libraries. Organizations already running workloads on NVIDIA infrastructure, or planning to, will find Verily Pre an easier integration path. Those who’ve standardized on AMD or alternative GPU stacks are looking at a widening capability gap, not just a pricing one.
Forecast 1.0, Verily’s foundation model combining genomics data with electronic health records (EHRs, the structured clinical data systems hospitals use to track patient history), is the clearest proof point for what Verily Pre is actually selling. The model uses multimodal data, meaning it draws from multiple data types simultaneously rather than a single source, to predict long-term health risks. That’s architecturally significant because most healthcare AI tools today operate on a single data type. Building on a platform where that multimodal capability is native infrastructure, not a custom integration, is a meaningful architectural decision for any health system planning AI at scale.
The risk Verily is running is the same one every would-be infrastructure layer runs: the customers it needs to win are also potential competitors. Health systems large enough to generate the data volume Verily Pre is designed to handle are exactly the organizations building internal AI platforms of their own, with their own NVIDIA relationships. NVIDIA investing in Verily doesn’t preclude NVIDIA selling directly to those same health systems. If Verily Pre doesn’t demonstrate measurable lift over what a well-resourced health system can build internally within the next 18 months, the infrastructure layer story collapses into a point-solution story, and point solutions don’t command infrastructure pricing.
Concept deep-dive: Multimodal health data
Multimodal health data means combining fundamentally different types of medical information, like genomic sequences, EHR records, medical imaging, and wearable signals, into a single model input. Think of it as the difference between diagnosing a patient from lab results alone versus synthesizing labs, imaging, and family history together. The business relevance is that predictive accuracy improves sharply when models can draw across data types, but so does the infrastructure complexity required to clean, align, and process those inputs at scale.
Based on reporting from Verily Health Secures Investment From NVIDIA And CU Healthcare Innovation Fund To Expand Precision Health AI Platform, originally published 2026-09-20 09:03:00.
