Infinimmune Is Betting the Best Antibody Has Already Been Made

Somewhere in the long archive of the human immune system, Infinimmune thinks, a useful antibody may already exist. The trick is to find it before the body forgets where it put it. That sounds like a whimsical plan until you meet the machinery: single human memory B cells, their naturally paired antibody chains, a database of sequences, a model that proposes edits, and a laboratory that must prove those edits work.

The pitch is both modest and audacious. Modest, because nature has done much of the molecular drafting. Audacious, because an antibody found in a person is still a very long way from a medicine that reliably helps another person. Infinimmune’s work is the expensive stretch between those two facts.

The library has a pulse

Antibodies are Y-shaped proteins whose two chains work together to recognize a target. Conventional discovery can immunize animals, search display libraries, or generate candidates computationally. Infinimmune starts with human B cells and tries to preserve the antibody pair as it existed in a person. That choice gives researchers a record of biology that has already operated in humans, even though it gives no guarantee of clinical safety or efficacy.

Its Complete Human sequencing and Anthrobody platform are meant to capture that record at scale. The company says it can screen up to 300 million B cells across more than 350 targets, recover naturally paired sequences, and select candidates for deeper testing. It has described a path from screening to a development candidate in as little as eight weeks. The qualification is important: discovery speed is a platform measure, while years of toxicology, manufacturing and trials still lie ahead.

A discovery engine is a beginning, not a clinical result.

GLIMPSE-1, launched in 2025, is the computational part of this chain. Trained on native human antibody sequences, the model proposes changes to improve affinity, developability and other properties without first requiring a three-dimensional structural model. Infinimmune reported promising humanization and optimization benchmarks in a preprint. Those are useful research signals, though independent clinical evidence for its drug candidates is still to come.

Two bets on one itchy problem

The first real-world proving ground is atopic dermatitis, the chronic inflammatory skin disease better known as eczema. IFX-101 targets IL-22, a signaling protein for which there is no approved eczema biologic. IFX-201 targets IL-13, a familiar route in an already competitive field. The company calls both molecules half-life extended and hopes they can be given less often than current treatments.

That possibility matters to anyone who has organized life around recurring injections. It also sharpens the question for Infinimmune: can its antibody design produce enough practical improvement to persuade doctors and patients who already have options? Dupixent, Ebglyss, Nemluvio and Adbry are approved injectable biologics for atopic dermatitis. Other companies are also chasing longer-lasting treatments. A clever origin story will not secure a prescription.

In August 2026, Infinimmune raised $75 million in a Series A co-led by Regeneron Ventures and Playground Global. The financing is intended to move the two lead eczema programs into first-in-human studies in 2027. Add the $12 million seed round announced in 2022 and the two named equity rounds total $87 million. That is the disclosed cost of getting this far, not the cost of finishing the work.

“The human immune system makes extraordinary antibodies. We built a company to find them.”Wyatt McDonnell, co-founder and CEO

The partner business is already real

While the owned drugs wait for human studies, Infinimmune can offer its discovery engine to other drug developers. That is where the customer picture becomes clearer. A pharmaceutical company can bring targets and development capacity; Infinimmune can search its human repertoires and engineer candidate antibodies. The bargain divides scientific labor and future rights.

The largest announced example is Merck. In March 2026 the companies agreed to work on multiple undisclosed targets. Merck has exclusive rights to develop and commercialize resulting candidates. Infinimmune receives an upfront payment whose size was not disclosed and is eligible for up to approximately $838 million in aggregate potential payments. The number is a ceiling tied to progress, not money already collected. It is a measure of what successful programs might be worth to the partner.

The earlier agreements show how the platform can be used in different settings. Grid Therapeutics paired with Infinimmune in 2023 to study B-cell repertoires from people with non-small cell lung cancer, selected partly by their responses to treatment. Immunome signed a discovery and optimization collaboration in 2025. KBI Biopharma agreed to support antibody manufacturing. Each arrangement moves a different part of the drug-making problem onto the table: finding a signal, improving a molecule, then making it reproducibly.

The useful lesson is in the handoff

Infinimmune was founded in 2022 by Wyatt McDonnell, Mike Gibbons, Katie Pfeiffer, Lance Hepler and David Jaffe. The mix of immunology, sequencing, computation and development experience is not decoration. Each discipline inherits a problem from the last. If sequencing loses the natural chain pair, the model starts with a poorer record. If the model proposes a beautiful sequence that cannot be manufactured, the laboratory returns it. If the laboratory produces a potent antibody whose dosing or safety disappoints, the clinic sends the whole team back to its assumptions.

This is the part others can copy, whether they are building a biotech or a completely different enterprise: preserve the context around your raw material, keep the feedback loop short, and make the next handoff explicit. Infinimmune did not merely assemble a large database of letters. It connected those letters to cells, binding behavior, engineering choices and drug programs. The context is what makes a pile of sequences usable.

The conditions for success are demanding. A target must matter in disease; the rare antibody must be found; its attractive properties must survive engineering and manufacturing; and patients must do better, or have a meaningfully easier treatment experience, than they would with existing care. Human origin may improve the starting point. It cannot skip the last condition.

That leaves Infinimmune in a precise, interesting position. Its discovery engine has attracted serious partners and financed a clinical push. Its own treatments have yet to meet patients. The company’s wager is that somewhere in the body’s enormous memory lies a better sentence for medicine to finish. In 2027, it hopes to begin finding out whether the grammar holds up outside the library.