Health and AI startups in Brazil: Real promise or bubble?

Brazil’s healthtech market is in 2026 at its most mature moment yet. With more than 1,900 active companies, the country already accounts for nearly 65% of all health startups in Latin America, ahead of Mexico, Argentina, Colombia and Chile combined, according to the Yearbook of the Associação Brasileira de Startups de Saúde e Healthtechs (ABSS), the Brazilian association of health startups and healthtechs.

In the first half of 2026, startups in the sector raised more than R$ 520 million. Much of that capital went to artificial intelligence solutions, especially those designed to speed up imaging exams and act as clinical copilots, supporting medical decision-making without replacing it.

Compared with 2025, when the sector raised R$ 1.214 billion, the pace of investment suggests less euphoria. Far from signaling a pullback, the shift points to a course correction. Today, the benchmark for a strong healthtech is no longer the size of the checks, but the ability to prove real cost reductions and efficiency gains for hospitals and health plan operators.

Given this new context, the question is unavoidable: Are we looking at a promise that is finally beginning to materialize, or at expectations still inflated by the recent investment cycle? 

To understand how this shift plays out in practice, Prime Health Report heard from two perspectives that meet in the same market but start from different logics: that of someone building an AI healthtech from the inside and that of someone who decides, every day, where to allocate capital in the sector.

 Has AI really become indispensable for a healthtech to survive?

For Pedro Batista Jr., a physician and CEO of Horuss AI, a Brazilian healthtech that unifies health data, the answer is yes, but the progress comes with a relationship between bets and returns that he describes as “uncomfortable.” "In 2025, AI accounted for about 46% of all health investment worldwide," he notes.

The mismatch becomes clear when you look at the US market: the amount invested is ten to one relative to actual returns. In other words, ten dollars put in on one side for just one dollar of real return on the other.

Despite the mismatch, the progress is undeniable, at least on the practical side. In the United States, the share of hospitals and health systems using AI jumped from 3% to 22% in just two years, driven mainly by tasks such as visit documentation and billing management. Even so, the pace of adoption continues to lag behind the volume of capital invested. 

In Brazil, the picture is even more down-to-earth. With high interest rates, investors have lost patience with long-term promises and started demanding immediate returns. It’s no coincidence that financial and accounting management systems, the less “glamorous” side that delivers quick savings, attracted the most money, while the boldest medical applications remain in the testing phase.

What is already a proven result, and what is still a promise?

In practice, Batista divides the sector into three layers: what is real operation, what is in transition and what hasn’t yet moved beyond talk. Today, proven results are concentrated in the back office, such as billing, procedure authorization and telemedicine, and in AI for image analysis, which already has regulatory approval.

So-called predictive tools, capable of forecasting patients’ health risk and performing smart triage, are starting to deliver results. However, that success depends less on the algorithm itself and more on good management and the maturity of medical teams.

But the real problem lies where the science is still nothing more than a promise: diagnoses without a physician, symptom chatbots in place of nurses and autonomous virtual health assistants. “The closer to the patient and the more autonomous the decision, the less evidence is available today,” Batista sums up. And he concludes: “The concrete results are in the back office; the promise is in the stethoscope.”

What real challenges does an AI healthtech face in Brazil?

For Batista, the biggest barrier isn’t regulatory, it’s data. "Validating AI requires longitudinal, structured and interoperable data, and the Brazilian system is an archipelago," he says, citing medical records fragmented by hospital and the still-nascent National Health Data Network (Rede Nacional de Dados em Saúde, RNDS). 

A model trained on data from a private hospital in São Paulo, he warns, is not validated for the interior of the Northeast within the SUS (Brazil’s public health system). "It’s not a statistical detail; it’s a risk of amplifying inequality." Added to this is real regulatory fragmentation: the same solution answers simultaneously to Anvisa (Brazil’s health regulatory agency), Brazil’s Federal Council of Medicine and the LGPD, Brazil’s General Data Protection Law. "What’s missing isn’t regulation, it’s coherence among them," he sums up.

But the "cruelest" obstacle, according to Batista, is the so-called eternal pilot: "Brazilian hospitals are happy to run a pilot for free, but the gap between the pilot and the contract is an abyss." In the United States, about 80% of AI projects in health care never get past the pilot phase, and in Brazil, in the executive’s assessment, the figure is even worse.

Do investors see this volume of capital as sustainable, or as a bubble?

Gustavo Cavenaghi, a partner at KX Ventures, a venture capital firm specializing in health, avoids simplistically labeling the moment as either a "bubble" or "sustainable." "Whenever relevant new technologies emerge, it’s natural that the market doesn’t yet understand exactly how to price them," he reflects.

In his view, the sector is going through a cycle similar to that of telemedicine in 2020 and 2021: the technology transformed care, but the initial euphoria faded. "Few companies managed to build a profitable business on that technological proposition alone, and they had to reinvent themselves as they grew," he compares.

When it’s time to sign the check, Cavenaghi looks for startups that deliver more than impressive technology. To attract serious capital, a company needs to combine solid technology, a clear benefit for the customer and a team that deeply understands the sector. "Being an entrepreneur in health is a challenge of its own; having a good product isn’t enough," he says.

For the investor, it’s precisely in the team that the main warning sign lies. Because AI has made building software increasingly cheap and accessible, the technical barrier has fallen. As a result, the dividing line between those who attract investment and those who are left out has become the team’s ability to navigate the bureaucracy of hospitals, health plan operators and regulators.

Comparing Brazil with the US market, Cavenaghi highlights local particularities. Here, the sector is focused on B2B (sales to companies and hospitals) and split between the SUS and private health insurance plans. Solutions created in the United States can rarely just be "imported" without adaptation, which opens valuable room for Brazilian startups to solve local pain points without international competition.

What do the sector’s biggest failures reveal?

The sector’s most instructive failures, according to Batista, didn’t come from small companies, but from some of the biggest startups in the sector. The most emblematic case was that of British company Babylon Health. The startup reached a valuation of US$ 4.2 billion at its IPO in 2021, boasting contracts with the UK’s public health system (NHS) and the promise of an "AI doctor" for everyone. The company went bankrupt in 2023, after former employees revealed that the supposed "artificial intelligence" was nothing more than a set of manual rules organized by junior doctors in spreadsheets. Another billion-dollar collapse was that of US-based Olive AI, valued at US$ 4 billion after raising more than US$ 800 million for hospital billing automation. The company grew too fast, promised more than it delivered and shut down operations at the end of 2023.