A Palo Alto biotechnology startup has come out of stealth with $50 million in financing and access to computational data from more than 500,000 patients, betting that a foundation model trained on blood samples can spot disease patterns that traditional single-condition tests miss entirely. Network Bio says its approach links tissue, blood and long-term clinical outcomes from major academic medical centers into one system designed to find signals in illnesses it was never specifically trained to detect.
The company, based in Palo Alto, California, launched with the $50 million round, according to Rutland Herald, which reported that the raise comes alongside what the company calls the world's largest patient tissue training dataset. Investors in the round include Section 32, Thiel Bio, Founders Fund, Breyer Capital, Blue Venture Fund, JSL Health Capital and other life science and AI funds, per the same report. The funding will support the expansion of Network Bio's life science platform, according to the outlet.
Central to the pitch is an academic biobank network that includes Duke University, Mass General Brigham, the University of Pennsylvania and the University of Colorado Anschutz, granting computational access to data from more than 500,000 patients spanning oncology, immunology, cardiovascular and metabolic conditions, according to pharmaphorum. That kind of cross-institutional reach addresses a problem that has dogged medical AI for years: individual hospital biobanks have historically operated in silos, making it hard to train models across multiple diseases at once.
An NVIDIA Partnership Tackles a Massive Data Bottleneck
Network Bio also announced a strategic partnership with NVIDIA to build Nexus, a foundation model trained on cell-free RNA using NVIDIA's Parabricks genomic software and BioNeMo Recipes, the company said in an announcement covered by Ion Genomics. Cell-free RNA circulating in the blood reflects active gene expression across the body's tissues in real time, but processing it at scale is a serious computational challenge. A single blood draw yields hundreds of millions of transcript-level cfRNA observations, creating a bottleneck that Network Bio aims to resolve using the Parabricks software for rapid secondary analysis, per FirstWord HealthTech.
The underlying science has a peer-reviewed track record. The generative transformer technology behind Network Bio's RNA foundation models builds on earlier published research, including the Orion model published in Nature Communications in 2024, which detected early-stage lung cancer with 94% sensitivity and 87% specificity, according to bio.rodeo. Rutland Herald separately reported that Network Bio has already published results in respiratory disease, with additional studies underway in ovarian and bone disease, and that its bio-native architecture generates interpretable representations of disease biology intended to generalize across conditions.
Leadership Ties to Tempus AI, nference and the Arc Institute
Network Bio's leadership brings a mix of clinical-data and computational biology backgrounds. CEO and co-founder Asad Ali Ahmad holds a PhD in biomedical engineering from UNC Chapel Hill and previously worked at Tempus AI, while co-founders Raphael Potter and Hani Goodarzi bring experience from biomedical AI firm nference and the Arc Institute, respectively, according to Ion Genomics. Ahmad said, as quoted by Rutland Herald, that the company built a network of leading academic medical-center biobanks alongside an AI platform that finds signals in conditions it was never trained on.
Mike Pellini, managing partner at Section 32 and chairman of Network Bio's board of directors, said the company combines academic medical-center biobanks with AI architecture built specifically for medicine, per Rutland Herald's reporting. The research network applies common selection criteria, sample quality standards and data harmonization across sites, addressing a longstanding weakness in academic biobanking. Kimberly Muller, chief innovation officer of CU Anschutz Innovations, said in August that no single academic medical center can capture the full complexity of human disease, which is why multi-institutional biobank networks are necessary to represent diverse patient populations, according to Ion Genomics. Muller separately said Network Bio is creating a research resource that reflects diverse patient populations and supports discoveries that can translate into clinical practice, per Rutland Herald.
A $30 Million Deal With an Unnamed Healthcare Giant
Beyond the venture round, Network Bio has struck a co-development and licensing agreement valued at more than $30 million with an unnamed Fortune 100, top-10 healthcare enterprise focused on identifying and interpreting novel disease signatures, according to AllSci. The identity of that healthcare partner has not been disclosed. Generated data from the collaboration will be returned to commercial partners and collaborating institutions, per Rutland Herald, and the strategic collaboration is intended to develop next-generation AI models.
The $50 million launch round follows a $17 million seed round the company closed in February 2025, bringing its total disclosed capital to roughly $67 million, AllSci reported. Morgan Cheatham, a partner at Breyer Capital, is among the investors backing the company, and additional peer-reviewed research tied to the platform has appeared in Nature Machine Intelligence, according to Rutland Herald.
Regulators Are Still Writing the Rules
Network Bio frames its long-term ambition as General Medical Intelligence, an AI framework meant to learn transferable biological principles across multiple diseases rather than relying on isolated, single-condition models, per Pulse 2.0. That ambition is arriving just as federal regulators start grappling with how to oversee it. The FDA's Center for Devices and Radiological Health issued a discussion paper this month evaluating regulatory oversight frameworks for generative AI and multiomics models used in liquid biopsy diagnostics, according to a report cited by Rhymes with Haystack.
Several questions remain unresolved as Network Bio moves from launch announcement to clinical execution. The identity of the Fortune 100 healthcare partner behind the $30 million deal has not been made public, nor has it been reported how the company plans to navigate the FDA's emerging guidance on generative AI diagnostics. Whether Nexus can actually deliver on its General Medical Intelligence promise remains untested until formal preprints and benchmark results are released publicly.
