Here is your reality check on the AI protein design market

AI protein design is hot right now, and it has moved from academic papers to very big checks.

We've found a lot of interesting signals. Chai Discovery raised $400M at a $3.8B valuation in July 2026, working with Pfizer, Lilly and Novartis. And Generate Biomedicines raised $400M in a Nasdaq IPO in February. Public and private investors are funding companies that design proteins with AI.

Here, we unpack why protein design is taking off now, which kinds of proteins draw the most money, and what rounds this size mean for smaller teams.

Designing a protein the old way means months of lab work, and most candidates still fail.

We compare six types of protein design companies, from AI models and data providers to drug and enzyme developers. Then we rank what scientists and buyers most need fixed: slow experiments, expensive labs and proteins that fail at scale.

This shows which problems buyers will pay to solve first, and where they still prefer to rely on their own scientists.

Software licenses, paid design partnerships with milestones and royalties, in-house drug pipelines, and enzymes sold to industry: protein design companies make money in very different ways.

We compare software subscriptions, pharma partnerships and companies developing their own drugs. Subscriptions bring repeat revenue. Partnerships may pay upfront but depend on future drug results. Owning a pipeline keeps more upside and more trial costs.

You'll see which models bring in cash early, and which ones only pay off if a drug or a product reaches the market.

A growth rate tells you little until you know what drives it. So we go through the forces one by one.

Some drivers are plain to see: better models, cheaper DNA synthesis, and pharma's need for new drugs as big patents expire. Others reinforce each other. Each lab test creates data that improves the next design. Faster automated labs make that loop shorter. And every AI-designed protein that does well in trials makes the next partnership easier to sign.

Before building or investing, you should know what turns a good design into a protein that works in the body or the factory, and who pays for it along the way.

EvolutionaryScale and Profluent are the model makers people name first, but they're only part of the AI protein design market.

Many interesting players work out of sight. Some design enzymes that make detergents or food ingredients cheaper to produce. Others focus on one kind of protein, like antibodies, miniproteins or vaccines, or run the automated labs that test thousands of designs a week.

They're hard to compare because their results stay private. A pharma partner rarely says how well a design performed, and a model's score on a benchmark says little about what happens in a real cell. Many also work under a partner's name for years before a drug becomes public.

Below, you'll find the market grids and ecosystem maps we've built to show who designs what, for whom, and where newer teams are getting in.

Cradle, Arzeda, Absci… a few protein design companies have found paying customers or reached the clinic. Their business models are still being tested.

Cradle sells protein design software to biotech teams. Arzeda designs enzymes and ingredients for industry. Absci combines AI with its lab to make antibodies, including candidates in human trials. Each chose a different route, and none has shown that one approach wins everywhere.

We'll look at how these teams chose a first customer, paired their models with real lab data, and kept enough cash to reach proof.