Chai Discovery Just Convinced Pfizer. The AI Drug Discovery Race Has a Clear Frontrunner.

TL;DR

• Chai Discovery, a San Francisco-based AI drug discovery startup valued at $1.3 billion, has signed a second major pharma partnership — this time with Pfizer, following its January deal with Eli Lilly.
• The deals were unlocked by Chai-3, the company's latest antibody design model, which doubles the success rate of its predecessor and produces antibodies that bind 100 times more tightly to their therapeutic targets.
• Chai is now in talks to raise an additional $400 million at a $3.4 billion valuation, Forbes has learned.
• The company's platform-only business model — selling access to AI rather than developing its own drug pipeline — is a deliberate break from industry orthodoxy, and it appears to be working.

What Happened

On June 4, Forbes published an in-depth profile of Chai Discovery, revealing that the two-year-old startup had quietly signed Pfizer as its second major pharmaceutical partner. The deal follows Chai's January agreement with Eli Lilly to design multiple novel biologic therapeutics using its AI models.

The catalyst for both deals was Chai-3, the company's latest antibody design model, deployed earlier this year. According to the company, Chai-3 doubles the success rate of its predecessor (Chai-2) and produces antibodies that bind 100 times more tightly to their intended targets. In drug discovery, binding affinity is a proxy for effectiveness — a tighter bind means a more potent therapeutic with fewer off-target effects.

The numbers are striking. Chai-2 could design full-length monoclonal antibodies from scratch 16% of the time, compressing months of wet-lab work into roughly two weeks. Chai-3 improves on that significantly, and in roughly half of cases, the molecules it generates bind to their targets as tightly as approved drugs — meaning they could potentially skip the lengthy refinement process that typically follows initial discovery.

Chai is now in talks with more than 15 additional pharmaceutical companies and is negotiating a $400 million funding round at a $3.4 billion valuation, according to two investors familiar with the discussions.

What It Actually Means

The AI drug discovery sector has been awash in capital and short on clinical proof for years. Investors poured $11.4 billion into the space globally in 2025, more than double the $5.6 billion of 2024, according to PitchBook. Isomorphic Labs, Alphabet's AI drug discovery spinoff, raised $2.1 billion in May alone.

But capital is not the same as validation. The question that has hung over the sector is whether AI-designed molecules would actually work — whether the models were, as Chai cofounder Jack Dent put it, "smoke and mirrors."

The Pfizer deal changes the calculus. Pfizer is not a venture investor placing a speculative bet. It is a $63 billion (2025 revenue) pharmaceutical company with one of the most sophisticated internal R&D operations in the world. When Pfizer's scientists evaluated Chai-3 and decided to sign, they were making a build-versus-buy decision based on comparative performance data that the public has not seen.

"That got the Pfizer team really excited," Dent told Forbes, referring to Chai-3's performance improvements.

The Lilly deal, announced in January, was already a strong signal. Two top-five global pharma companies signing with the same startup within six months is not a coincidence. It is a market signal that Chai's models are producing results that internal teams at the world's largest drug companies cannot replicate — or cannot replicate as quickly.

The Business Model Bet

Chai's strategy is as notable as its technology. Most AI drug discovery companies develop their own pipeline of therapeutic candidates, because the potential revenue from a blockbuster drug is too large to ignore. The logic is straightforward: if your AI can design a drug that generates $10 billion in annual revenue, why would you sell the tool for a fraction of that?

Chai took the opposite approach. It sells platform access and refuses to develop its own drugs. The reasoning, as Dent explained to Forbes: "When we started, people told us the only way to make money is to make your own assets and become a drug company. That's the dogma we had to challenge."

The argument is that a software engine capable of spinning up scores of potential therapies is more valuable as a platform than as a single-pipeline company — and that pharmaceutical companies will not trust a partner that is also a competitor.

Mikael Dolsten, who retired as Pfizer's president of worldwide R&D and now sits on Chai's board, endorsed the strategy: "I think it was a brilliant insight that if you want to be the trusted one that traditional industries feel comfortable teaming up with, you cannot at the same time try to have your own little shop."

The platform model also means Chai's revenue is not binary — it does not live or die on a single drug's clinical trial outcome. If the platform works across multiple targets and modalities, the revenue compounds.

The Technical Leap: Why Chai-3 Matters

To understand why Chai-3 convinced Pfizer, you need to understand what it improved.

In drug discovery, the challenge is often described as a lock-and-key problem. The therapeutic target — a protein, enzyme, or receptor involved in a disease — is the lock. The drug molecule is the key. A good key fits the lock precisely; a great key fits so tightly that it stays bound, producing a sustained therapeutic effect with minimal side effects.

Chai-2, released in mid-2025, could design antibodies from scratch — a significant achievement. But the antibodies it produced did not always bind tightly to their targets. They still required the same lengthy refinement process as conventionally discovered molecules.

Chai-3 changes that. The 100x improvement in binding affinity means the model is producing molecules that are closer to clinical-grade from the start. In roughly half of cases, the binding affinity matches that of already-approved drugs. That is the difference between a tool that accelerates discovery and a tool that potentially skips entire phases of it.

The founders also see Chai-3 as a step toward designing antibodies that can bind to multiple targets simultaneously — creating more complex, precise therapies than current approaches allow. "Most antibodies are just jamming up some targets. State-of-the-art is jamming two targets," CEO Josh Meier told Forbes. "In the future, we will modulate targets in more powerful ways."

Hype Deconstruction

Let's be clear about what this is not.

This is not an AI-discovered drug in clinical trials. Chai's molecules have not entered human testing. Lilly's chief information and digital officer Diogo Rau told Forbes in March that AI-developed medicines from the collaboration would likely not reach the market until the "mid-2030s, if not late-2030s." The drug development timeline — even with AI acceleration — remains measured in years, not months.

This is not validated clinical efficacy. Binding affinity in silico and in vitro is not the same as safety and efficacy in humans. The molecules Chai-3 produces still need to clear the same regulatory hurdles as any other drug candidate.

The $3.4 billion valuation is a negotiation, not a fact. Chai is in talks to raise at that figure. It has not closed. In the current market, AI drug discovery valuations are volatile — Isomorphic's $2.1 billion raise set a ceiling, but it also raised the bar for what investors expect in return.

What this is: the strongest signal yet that AI-designed molecules are crossing the threshold from "interesting research" to "something the world's largest pharmaceutical companies are willing to pay for." That is a genuine milestone. It is not the finish line.

Stakeholder Landscape

Who wins:

• Chai Discovery — two top-five pharma partnerships in six months, a platform model that scales, and a fundraising narrative that writes itself.
• Pfizer and Lilly — early access to a platform that could compress drug discovery timelines and unlock targets traditional methods cannot reach. The risk is that the molecules don't work in humans; the reward is a structural advantage in biologic drug development.
• Patients with diseases currently considered "undruggable" — the long-term promise of AI drug discovery is reaching targets that have resisted conventional approaches. Chai-3's multi-target ambitions point in this direction.
• The platform-model thesis — if Chai succeeds, it validates the argument that AI drug discovery companies are more valuable as tool providers than as drug developers, reshaping how the sector is funded and structured.

Who should be watching carefully:

• Isomorphic Labs — Alphabet's AI drug discovery unit raised $2.1 billion but has a different model (it develops its own pipeline alongside partnerships). Chai's platform-only approach is a direct competitive challenge.
• Other AI drug discovery startups — the bar for "convincing a top-five pharma to sign" has now been set. Chai-3's performance metrics (doubled success rate, 100x binding improvement) are the new benchmark.
• Regulators — the FDA and EMA have not yet established clear frameworks for AI-designed biologics. As molecules designed by models like Chai-3 enter clinical trials, the regulatory pathway will need to adapt.

Cross-Layer Implications

Commercial: The platform model, if it works, changes the economics of drug discovery. Instead of binary bets on individual molecules, pharma companies could pay for access to a design engine that produces candidates continuously. This shifts risk from the startup (which no longer lives or dies on a single trial outcome) to the pharma company (which still bears the clinical development cost).

Talent: Chai's founding team — Meier (Harvard, ex-OpenAI, ex-Meta), Dent (Harvard, ex-Stripe), McPartlon (ex-Absci), Boitreaud (ex-Aqemia) — represents a new archetype: the AI-native drug discovery founder. These are not biologists who learned to code. They are engineers and computer scientists who applied themselves to biology. The talent market for this profile is about to get very competitive.

Geopolitics: Both Chai and Isomorphic are US-headquartered (though Isomorphic is a UK-founded Alphabet subsidiary). The AI drug discovery race is, for now, an American story. China has significant AI capabilities but its pharmaceutical sector is structured differently, with less emphasis on the kind of platform-model partnerships Chai is pursuing. This could become a strategic asymmetry.

Investment: The $11.4 billion that flowed into AI drug discovery in 2025 was already a record. Chai's $400 million raise at $3.4 billion — if it closes — will be another data point suggesting the sector is not cooling. But the bar for new entrants is rising. "We're past acceptance and into the excitement part of the adoption curve," Dent said. The implication: the window for "we have an AI platform for drug discovery" without pharma partnerships is closing.

What This Means for You

If you are in the pharmaceutical industry: Chai-3's performance metrics are now the benchmark against which internal AI efforts and external partnerships will be measured. If your organisation is not evaluating AI-native drug discovery platforms, you are now behind Pfizer and Lilly. The question is not whether to engage — it is which platform and on what terms.

If you are an investor in biotech or AI: The platform model versus pipeline model debate is no longer theoretical. Chai's traction with two top-five pharma companies is real-world evidence that the platform approach can work. But the revenue from platform deals — typically tens of millions upfront with billion-dollar potential downstream — is a different risk profile than owning a piece of a blockbuster drug. Know which you are betting on.

If you are a patient or patient advocate: No AI-designed drug has reached the market yet, and the timelines remain long (mid-to-late 2030s). The molecules Chai-3 produces still need to survive clinical trials. But the fact that the world's largest drug companies are betting on this approach — and that the models are improving at the rate Chai claims — means the probability of AI-designed therapies reaching patients is rising. It is no longer a question of "if." It is a question of "when" and "for which diseases first."

If you are a technologist or engineer: The Chai story is a case study in applied AI. The founders did not invent a new model architecture from scratch. They applied existing techniques to a specific, high-value problem — protein and antibody design — and iterated relentlessly on performance. "We knew we had to make things literally 100 times better for it to be valuable for real drug discovery programs," Dent said. That is the engineering mindset that separates research from product.

Uncertainty Ledger

• No clinical data yet. Chai-3's performance metrics are computational and in vitro. The molecules it designs have not been tested in humans. The 100x binding improvement is real in the lab; whether it translates to clinical efficacy is unknown.
• The Pfizer deal terms are undisclosed. We do not know the upfront payment, milestone structure, or royalty arrangement. Comparable deals in the space (Genesis Molecular-Incyte, Isomorphic-Lilly) have involved tens of millions upfront with total potential value above $1 billion. Chai's deals are likely in this range, but unconfirmed.
• The $3.4 billion valuation is in negotiation. It has not closed. In a market where AI valuations are under increasing scrutiny, the final number could be lower — or higher, if the Pfizer deal attracts competing term sheets.
• Competitive dynamics are fluid. Isomorphic Labs has $2.1 billion in fresh capital and DeepMind's research depth. Other well-funded competitors (Recursion, Insilico Medicine, Generate Biomedicines) are pursuing different models. Chai's current lead is real but not unassailable.
• Regulatory pathways for AI-designed biologics are undefined. The FDA has not issued specific guidance on how it will evaluate molecules designed by AI models. This is a known unknown for every company in the space.

Bottom Line

Chai Discovery has done something no other AI drug discovery startup has: convinced both Eli Lilly and Pfizer to pay for access to its platform within six months. The technical leap from Chai-2 to Chai-3 — doubling success rates and improving binding affinity by 100x — is the kind of performance curve that turns sceptics into customers. The platform-only business model is a deliberate bet against industry orthodoxy, and it is working. But the molecules Chai-3 produces have not been tested in humans, and the timelines for AI-designed drugs remain measured in years, not quarters. The AI drug discovery race has a frontrunner. The finish line is still a long way off.

Sources:

• Forbes, "Why Pfizer And Eli Lilly Are Betting On This $1.3 Billion AI Drug Discovery Startup," Amy Feldman, June 4, 2026 [Tier 2 — reputable business publication, primary reporting]
• Forbes InnovationRx, "SpaceX's Healthcare Plays," Amy Feldman and Alex Knapp, June 10, 2026 [Tier 2 — newsletter format, references Chai Discovery]
• BioSpace, "Chai Discovery Announces License Agreement with Pfizer," June 5, 2026 [Tier 2 — press release via industry publication]
• PitchBook, AI drug discovery investment data 2024–2025 [Tier 2 — industry data provider]
• Forbes, "How Lilly Used AI To Crank Up Production Of Its Popular GLP-1s," Amy Feldman, March 7, 2026 [Tier 2 — primary reporting, includes Diogo Rau timeline estimate]
• Forbes, "Isomorphic Labs' $2.1 Billion Fundraise Is The Biggest Bet Yet On AI Drug Discovery," Amy Feldman, May 13, 2026 [Tier 2 — primary reporting]