Ortet launches as a frontier AI lab for health with $500M from Thoreau
Ortet, a frontier artificial intelligence lab for health, said Tuesday it launched with a $500 million commitment from Thoreau to build compute, data infrastructure, research, and health-focused models that learn across biology, clinical care, operations, and finance around each patient.
Most health AI tools are built for one narrow task, on one narrow slice of data. Ortet is putting a $500 million commitment into the computers, the data, the research, and the models together, so a system built around one patient can learn across biology, care, operations, and money, and carry that learning into the next encounter.
On Tuesday, 29 September 2026, Ortet announced its launch with a $500 million commitment from Thoreau. Ortet calls itself a frontier artificial intelligence lab for health. A frontier lab, in that phrase, is a company trying to build new models at the edge of what the field can do. Thoreau, in the same release, is building the foundational compute, rules, and research infrastructure platform for health. Compute is the machines that train and run models. Business Wire carried the announcement. The open reprint of that wire stamps it September 29, 2026, 8:00 a.m. Eastern. Those lines are Ortet’s.
What Ortet says it will put together. The lab will bring compute and data infrastructure, frontier research, and model development and deployment into one effort. Data infrastructure is the system that holds the records those models learn from. Deployment is putting a trained model into real use. Ortet says the point is to address health across biological, clinical, operational, and financial dimensions at the same time. Biological is the science of the body. Clinical is the care a patient receives. Operational is how a clinic or hospital runs the work. Financial is the money side of that care. Those lines are the release.
The gap Ortet names. Most healthcare AI tools, the release says, are built for a single task and a narrow set of data. Ortet says that limit makes it hard to learn how a patient’s biology, treatment, outcomes, and dealings with the health system fit together. Its answer is a shared intelligence foundation: one base that can keep improving and be adapted across scientific discovery, clinical care, health operations, and financial administration. The wire’s subhead calls that a patient-centered foundation model, supported by research, data, and compute together. A foundation model is a large model built so many jobs can reuse it. Designed around the individual patient, Ortet says, what the model learns in one of those areas can strengthen the others. Those lines are the company’s.
Who is in charge. The co-founder and chief executive is Kyunghyun Cho, Ph.D. The release says he co-developed the attention mechanism and the gated recurrent unit, shortened to GRU. Attention is a way for a model to weigh which parts of an input matter most. A GRU is a small memory unit for a sequence, such as words or a string of measurements over time. The release says those two advances laid important groundwork for the transformer architecture behind today’s large language models. A transformer is that architecture, the design those models are built on. Cho is the Glen de Vries Professor of Health Statistics at New York University. The release says he spent five years building an AI platform for lab-in-the-loop molecular design, co-founding Prescient Design and, after that company was acquired, establishing Genentech’s frontier research team. Lab-in-the-loop means the model and the laboratory experiment take turns, each one informing the next. The release does not name the buyer of Prescient Design. Those lines are the release.
The people the release names beside him. Keunwoo Choi, Ph.D., is co-founder and chief AI officer. He previously developed AI systems at Upstage, Spotify, ByteDance, and Genentech. Henri Dwyer is co-founder and chief technology officer. He built and managed what the release calls the largest GPU cluster in life sciences at Genentech, after engineering roles at Instagram and Lyft. A GPU cluster is a bank of graphics chips used to train models. Jeff Hammerbacher is co-founder and chairman. He is founder and chief executive of Open Athena. He was previously principal investigator at Hammer Lab, founder and chief scientist at Cloudera, and manager of the data team at Facebook. Elman Mansimov, Ph.D., is co-founder and chief science officer. The release calls him a pioneer of neural text-to-image generation, software that turns a written description into a picture, and says he worked on post-training and agentic AI platforms at Amazon Web Services. Post-training is the extra teaching after a base model already exists. Agentic, here, means software that takes a series of steps. Claudia Shi, Ph.D., is a founding member of frontier research. The release does not list her as a co-founder. She worked on research and development of AI at Meta Superintelligence Labs. Those names and titles are the release.
Cho’s line in the announcement. “Patient health should improve with every interaction, but today its data, knowledge, and infrastructure remain too fragmented for that learning to happen at scale,” he said. He said Ortet is building at every layer of the AI stack so advances in research can become systems that understand a patient’s full context and support better decisions across the continuum of care. A layer of the stack, in that sentence, is one part of the system, from the computers up through the models. The continuum of care is the path from one visit to the next. He said Thoreau’s commitment gives the resources and the long-term foundation that mission needs. That quotation is his, in the release.
Where the $500 million goes, as the release states it. The commitment will support Ortet’s compute and data infrastructure, the training and deployment of health-focused models, and continued investment in research and talent. Ortet says it has secured an initial GPU cluster and is building the technical foundation to move research into production at scale. Production, here, means systems running real work, past a lab demo. The company expects to work with customers across the health ecosystem from launch, so the models and products sit on real-world needs. The release does not name those customers. Goldman Sachs advised on the transaction. A transaction, in that sentence, is the commitment the wire is announcing. The release does not print a valuation, an ownership percentage, or a contract with a named hospital or insurer. Those lines are the release.
How the two companies describe themselves on the same wire. Ortet says it is a frontier AI lab building a platform that spans compute and data infrastructure, frontier research, and model development and deployment. The about box calls that a full-stack platform, meaning those layers sit in one place. Centered on the individual patient, Ortet says it connects the biological, clinical, operational, and financial dimensions so the system can learn from every patient, for every patient, and deliver predictably better care. “Predictably better” is the company’s phrase. The about box calls Ortet a founding Thoreau investment. Thoreau says it creates, invests in, and acquires platforms across provider, payer, employer, life sciences, and government markets. A provider is a hospital or clinic. A payer is an insurer. Thoreau’s initial platform partnerships, as it lists them, include Ensemble, which it says manages over $55 billion in net patient revenue, Penelope Health, a real-time health policy intelligence platform, and Ortet. Net patient revenue is money from patient care after discounts. More than $55 billion is Thoreau’s figure for Ensemble, not a figure for Ortet. Thoreau’s mission, on the page, is to simplify the administration of health and to improve and extend patients’ lives in the United States and globally. Those lines are the release.
In plain terms, a new health AI lab led by Kyunghyun Cho said on Tuesday that Thoreau is committing $500 million so Ortet can build computers, data systems, research, and models around the whole patient. The wire names the founding team, an initial GPU cluster, and Goldman Sachs as adviser. It names a commitment. It does not name a valuation or a customer.
The picture is a photograph of Kyunghyun Cho, Ortet’s co-founder and chief executive, at a conference. He is seated at a table with a microphone, in a dark jacket and a light shirt. A thin teal bar runs across the top of the frame. The photograph is by Xuthoria on Wikimedia Commons, under the Creative Commons license CC BY-SA 4.0. That license allows reuse when the photographer is credited and changes are shared on the same terms. It is a portrait of the person named in the release.