Here is your reality check on the AI in oncology market

AI in oncology is gaining ground, with stronger evidence in clinics and new uses in research.

We've found a lot of interesting signals. For example, in October 2025, Google and Yale said one of Google's AI models had suggested a new way to make some tumors easier for the immune system to spot, and lab tests on cells backed it up. Then in January 2026, The Lancet published the MASAI trial from Sweden: among about 106,000 women, AI-supported breast screening led to 12% fewer cancers diagnosed between screening rounds. AI is starting to deliver results both in cancer research and in real screening programs.

In this section, we look at why cancer teams are trusting AI more now, which uses already have strong evidence, and what still has to be proven in treatment decisions and trials.

Cancer care runs on scans, slides, genetic tests and long patient records, and oncologists rarely have time to read it all.

We compare six kinds of oncology AI companies, from screening and imaging tools to treatment planning, biomarker and trial matching software. Their buyers include radiologists, oncologists, cancer centers, screening programs and drug makers. Then we rank their biggest pains, like cancers missed on busy screening days, treatment choices made with incomplete data and trials that can't find enough patients.

This shows which problems cancer teams will pay to solve first, and which ones still live mostly in research papers.

Software licenses for screening programs, per-test fees, radiotherapy planning subscriptions, data deals and pharma contracts for trials: oncology AI companies make money in several ways.

We compare the payers and the costs behind each model. A screening tool can be paid for by a health system that saves reading time, but it rarely gets its own insurance payment. A prognostic test billed per patient needs coverage and a place in guidelines. Pharma pays well for help with trials and patient selection, but those deals rise and fall with drug pipelines. And every model needs costly data and clinical studies to stay credible.

You'll see which models bring in enough to pay for the evidence they need, and which ones depend on one pharma partner.

To understand the growth rate, you need the forces behind it. We take them one at a time.

Some drivers are easy to see: more cancer cases, too few radiologists and oncologists, and more targeted drugs that require a test first. Others reinforce each other. Large screening trials give health systems the proof they need to buy. Each tool in use creates data for the next version. And drug makers that use AI to pick trial patients can turn the same tool into a test doctors rely on later.

Before building or investing, you should know what evidence makes cancer teams change how they work, and who ends up paying for that change.

Flatiron Health and Owkin come up first when people talk about AI and cancer data, but the market goes well beyond them.

Many interesting players work on one narrow step. Some only plan radiotherapy. Others predict how aggressive a tumor is, read lung CT scans in screening programs, or sort through medical records to find patients for a trial.

They're hard to compare because each one works on a different cancer, at a different stage of care, with evidence from different studies. Many also sell through drug makers, imaging vendors or labs, so patients and even doctors rarely see their name.

Below, you'll find the market grids and ecosystem maps we've built to show who works where, from screening to treatment and trials, and where newer companies can still get in.

Artera, ScreenPoint Medical, Massive Bio… a few oncology AI companies have made it into real cancer decisions. They got there in very different ways.

Artera won FDA De Novo authorization in 2025 for a tool that estimates a prostate cancer patient's risk from biopsy images. ScreenPoint Medical's Transpara was the AI used in the Swedish MASAI screening trial. Massive Bio helps cancer patients find clinical trials and works with the drug makers running them. None of these paths has been proven across every cancer yet.

We'll look at how these teams chose one cancer or one step of care, built the evidence doctors and payers ask for, and found someone willing to pay.