US tech firm Nvidia buys stake in French health platform Doctolib

A minority shareholder owns less than a controlling share of a company. Nvidia therefore gains a financial interest in Doctolib, but it does not control the platform’s decisions. Doctolib confirmed that Nvidia will have no seat on its board. The size and value of the stake were not disclosed.

The investment also creates a strategic relationship. Doctolib plans to use Nvidia’s chips, computing capacity and open software for its AI work. Nvidia can therefore support the technical infrastructure behind Doctolib’s planned health assistants, while Doctolib keeps control of its own models and patient data.

This gives Nvidia a foothold in a major European health technology company without full ownership. For Doctolib, the deal brings funding and access to computing resources. The company says the investment will accelerate recruitment and development of in-house AI models, while its governance remains with Doctolib rather than Nvidia.

Doctolib is one of France’s best-known technology companies. Its platform lets patients book medical appointments and store health records. The company was valued at €3.6 billion in its last funding round, giving a clear sense of its financial scale even though Nvidia and Doctolib did not disclose the new stake’s value.

Doctolib also says it has 50 million users. Its research programme uses data from those users to improve patient pathways. That figure describes the platform’s user base, while the article does not specify exactly how many individual health records are stored or how many users’ data are used in every study.

The combination of valuation and user reach explains why the investment matters beyond finance. Doctolib has a large healthcare footprint and access to sensitive information. Its planned AI expansion could affect patients and medical staff across that network, while also increasing scrutiny of how health data is handled.

Nvidia’s contribution combines investment with the technical resources needed for advanced AI work. Doctolib will use Nvidia’s chips, computing capacity and open software. The companies did not disclose the investment’s size or value, so the practical importance of the deal rests partly on access to this infrastructure.

Doctolib’s chief executive, Stanislas Niox-Château, said the investment would accelerate recruitment and development of the company’s own AI models. Those models are intended to power a personal health assistant for patients and assistants for medical staff. The planned tools would cover administrative and clinical tasks.

This arrangement links Nvidia’s hardware and software with Doctolib’s healthcare platform and data. Doctolib says it wants to retain control of its models and patient data. The investment could therefore help it build AI internally, instead of depending entirely on US providers such as Axiom or Paradox.

Doctolib is developing AI assistants for two groups. Patients could receive a personal health assistant, while medical staff could use assistants covering administrative and clinical tasks. The aim is to support work around healthcare and patient pathways through tools built into Doctolib’s platform.

The key mechanism is Doctolib’s planned in-house AI models. The company says Nvidia’s investment will speed recruitment and model development. Doctolib will also use Nvidia chips, computing capacity and open software. Its research programme uses data from its 50 million users to improve patient pathways, although rights groups have criticised that move.

The article does not claim that these assistants are already changing access to care or clinical work. Their future effect will depend on development, deployment and safeguards. Doctolib says it wants to retain control of its models and patient data, while the sensitivity of health information makes privacy and responsible use central concerns.

Doctolib says it wants to develop in-house AI models and retain control of them and its patient data. This approach would let the company shape tools for its own healthcare platform rather than relying fully on outside AI providers. The article names Axiom and Paradox as examples of US providers it wants to avoid depending on.

The company will still use Nvidia’s chips, computing capacity and open software. This separates the infrastructure from the models and data: Nvidia supplies resources for building and running AI, while Doctolib seeks to keep ownership and control of the systems it develops. Its planned tools include patient and medical staff assistants.

The strategy also reflects the sensitivity of healthcare information. Doctolib’s research programme uses data from 50 million users, and rights groups have criticised it. Keeping control may help Doctolib manage its data and models directly, though the article does not claim that this removes privacy risks.

Health data is especially sensitive because it can describe a person’s diagnoses, medical treatments and health history. These details are deeply personal and may affect how someone is viewed or treated. Using such information for AI research therefore raises concerns about privacy, control and possible misuse.

The Ligue des droits de l’Homme warned that health data carries a significant risk of stigmatisation, discrimination and abusive profiling. These risks could arise if information were exposed, used unfairly or linked to conclusions about individuals. The article does not describe a specific breach or claim that such harm has occurred.

Doctolib launched a research programme using data from its 50 million users to improve patient pathways. Rights groups criticised the move, showing the tension between medical innovation and data protection. Doctolib says it wants to retain control of patient data and its AI models, but the article indicates that sensitive-data concerns will remain important as the research develops.

Nvidia is widely known for making graphics processing units, or GPUs. These specialized chips can perform many mathematical operations in parallel. AI systems rely on vast numbers of such operations, so GPUs can provide the processing power needed to train models and run them after training. This explanation uses established background knowledge; the article only says Nvidia will provide chips.

Training AI involves repeatedly adjusting a model through large calculations. Running an AI system also requires calculations whenever it responds to a request. GPUs are designed to handle these workloads efficiently, especially when combined with substantial computing capacity and suitable software. Nvidia’s open software is part of the package Doctolib plans to use.

For Doctolib, access to Nvidia hardware could speed development of its in-house models and planned health assistants. The article says the investment will accelerate recruitment and AI development, but it gives no technical performance figures. Doctolib still intends to retain control of its models and patient data.