Merck Just Bet $838 Million on a Startup That Reads Your Immune System
Merck just signed a deal worth up to $838 million with Infinimmune, a four-year-old startup that mines the human immune system for antibodies using AI. It's the latest sign that Big Pharma's antibody shopping spree is far from over.
The Biggest Check a Four-Year-Old Startup Has Ever Seen
Imagine you're running a biotech startup that's barely old enough to attend preschool. You raised $12 million in seed funding. Your team came from genomics companies and research institutes, not Big Pharma corner offices. And then Merck calls with a deal worth up to $838 million.
That's what just happened to Infinimmune, a California-based startup founded in 2022. The collaboration, announced on March 31, will have Infinimmune use its AI-powered platform to discover new antibodies for Merck across multiple therapeutic targets. Merck gets exclusive rights to develop and sell whatever comes out of the partnership.
For a company that started with a skeleton crew of five ex-genomics researchers, this is the biotech equivalent of going from open mic night to headlining Madison Square Garden.
What Infinimmune Actually Does (And Why Merck Cares)
Most antibody drugs are made the old-fashioned way: scientists engineer them in a lab, tweaking proteins until they stick to the right target. Think of it like building a lock from scratch and hoping a door fits it.
Infinimmune flips that approach. Instead of engineering antibodies from scratch, the company mines them directly from the human immune system. Their Anthrobody platform screens millions of single memory B cells (the immune cells that remember past infections) from real people: cancer survivors, individuals with protective genes against Alzheimer's, and healthy donors. They're essentially raiding nature's medicine cabinet.
The logic is elegant. Your immune system has already spent your entire lifetime learning to make antibodies that work safely inside a human body. Why reinvent the wheel when you can copy the answer key?
Once Infinimmune finds promising candidates, their AI model called GLIMPSE (trained exclusively on human antibody sequences) optimizes them for potency, specificity, and the practical stuff that determines whether a drug can actually be manufactured. It's a one-two punch: nature's library plus machine learning's speed.
Get tomorrow's biotech intelligence before your competitors.
Join thousands of biotech professionals who start their day with our free, daily briefing.
Why Now? Two Words: Keytruda Clock
Merck isn't writing big checks to startups because it's feeling generous. The company is staring down one of the scariest cliffs in pharma.
Keytruda, Merck's blockbuster cancer immunotherapy and the world's top-selling drug, will eventually lose patent protection. When that happens, generic competition will erode billions in annual revenue. Every deal Merck signs right now is partly an insurance policy against that future.
And Merck has been busy shopping. In late 2024, the company paid $588 million upfront (with up to $2.7 billion in milestones) to license a bispecific antibody from LaNova Medicines. It took full global rights to an anti-CD30 ligand antibody from Dr. Falk Pharma for $150 million upfront. It's funding an antibody-drug conjugate through a Blackstone partnership.
The Infinimmune deal fits a clear pattern: Merck is diversifying across every flavor of antibody technology it can find. Bispecifics, ADCs, AI-discovered human antibodies; the company is building a buffet, not ordering a single entrée.
The $838 Million Question: Is This Real Money?
Let's pump the brakes for a second. That $838 million headline number deserves some context.
The deal includes an undisclosed upfront payment plus milestone payments. Neither company revealed how much cash actually changed hands on day one. The $838 million figure represents the theoretical maximum if multiple antibody candidates hit every development and commercial milestone.
In biotech deals, milestone-heavy structures are the norm. Think of it like a real estate contract where you put down earnest money, then pay the rest only if the house passes inspection, the appraisal comes in, and the roof doesn't leak. Most milestone-laden deals never pay out their full headline value because drug development is hard and most candidates fail.
That said, even the upfront payment (which remains private) represents a serious vote of confidence in a company that only raised $12 million in seed financing back in December 2022. Playground Global led that round, and the investors are probably feeling pretty good about their bet right now.
The Antibody Gold Rush Is Very Much On
Infinimmune's deal isn't happening in a vacuum. Big pharma has been throwing enormous sums at next-generation antibody platforms throughout 2025 and into 2026.
In January 2026, Sanofi signed a deal worth up to $2.56 billion with Earendil for AI-driven bispecific antibody discovery. That same month, AbbVie committed $5.6 billion for a bispecific antibody from RemeGen. And in mid-2025, BioNTech and Bristol-Myers Squibb struck an $11.1 billion deal (with $1.5 billion upfront) for a bispecific cancer drug.
The broader bispecific antibody market alone hit $9.53 billion in 2025 and is projected to balloon to over $52 billion by 2035. Around 178 antibody therapies are currently in late-stage development.
The pattern is unmistakable: pharma companies are racing to lock up the next generation of antibody technology before their competitors do. Platforms that use AI to discover or optimize antibodies are commanding particular premium valuations, because they don't just deliver one drug candidate; they deliver a machine that can crank out many.
What Makes This Deal Interesting (Beyond the Dollar Signs)
The Infinimmune deal stands out for a specific reason. Most AI-antibody partnerships focus on engineering better versions of known antibody formats. Infinimmune is doing something different: going back to the source.
By screening real human immune cells, the company argues it can find antibodies that are inherently safer and more effective than engineered alternatives. These aren't Frankenstein proteins stitched together in silico. They're antibodies that actual human immune systems produced against actual threats.
The founding team brings credibility to that claim. CEO Wyatt McDonnell and co-founders Katie Pfeiffer, Mike Gibbons, Lance Hepler, and David Jaffe all came from 10x Genomics, Pacific Biosciences, and the Broad Institute. Their backgrounds span single-cell methods, protein engineering, computational biology, and genome assembly. They built the tools that other scientists used to study immune cells, and now they're using those tools to find drugs.
The Bottom Line
Merck is placing a calculated bet that the best antibodies aren't the ones we engineer from scratch; they're the ones evolution already made for us. Infinimmune gets validation (and a lot of money) for a platform that's still in its early days. And the rest of the industry gets another data point confirming that AI-powered antibody discovery is where the smart money is headed.
Whether this deal pays out anywhere close to $838 million depends on whether Infinimmune's immune-cell mining operation can actually produce drug candidates that work in the clinic. That's years away from being answered.
But for a startup that's barely four years old, having Merck as your dance partner is one heck of a way to start the next chapter.
4 min read
A Lung Drug Just Aced a Test It Was Never Built For
United Therapeutics' lung drug Tyvaso just posted blockbuster results in a disease it was never originally designed to treat. The IPF data could double the drug's market, and Wall Street is already doing the math.
