China has become the pipeline’s margin of safety. When multinational pharma companies needed early-stage assets with genuine novelty, they went there — and they went there in numbers that have never been seen before. “I think we have nine more deals already this year to date than we had last year,” said Lauren Martz, BioCentury’s executive director of biopharma intelligence, speaking on BioCentury This Week. The three deals announced in the week before the episode — by Merck, Novo Nordisk, and Novartis — tell the story in dollars and stage: all three carried upfronts above $300 million, total values above $2 billion, and two of the three were for preclinical assets.

That is the shape of the 2026 China licensing boom. It is bigger, earlier, and broader than what came before.

The China wave is no longer an antibody story

For years, pharma sourcing from China was dominated by antibody-based therapies and cancer assets. That pattern has broken. The three deals announced in the prior week span three different modalities and three different disease areas, according to Martz’s analysis of deals between the 21 multinational pharmas BioCentury tracks and China-based biotechs.

Only one of the three deals — Novartis’s agreement with Abogen for an RNA-encoded T cell engager — had any clinical data at all, and that was limited data from two ongoing investigator-initiated trials in China. The other two assets were preclinical. This is not discount shopping for de-risked molecules. This is multinational pharma paying top dollar for science that has not yet been proven in the clinic.

DealPartnerAssetStageUpfrontTotal valueNovartis–AbogenAbogenRNA-encoded T cell engager (LNP-delivered mRNA encoding a CD19xCD3 T cell engager)Limited data from two IITs in China>$300MUp to $7.8BMerck–CybrunchCybrunchKRAS G12D molecular gluePreclinical>$300M>$2BNovo Nordisk (third deal)Not specified in episodeGLP-1/GIP dual receptor agonistPreclinical>$300MUp to $2.6B

The Abogen asset stands out for its mechanism. An mRNA-LNP delivers instructions that make the patient’s own body produce a T cell engager for days or weeks after a single treatment — a “vectorized biologic,” as Martz put it, sitting alongside in vivo CAR-Ts and conventional T cell engagers in a still-unsettled competitive field. The Merck asset, a KRAS G12D molecular glue, is likewise a potential first-in-class approach; Martz noted she does not know how it differs from the one Estellas is developing.

Washington’s red line is not the check, it’s the co-development

Steve Usdin, BioCentury’s Washington editor, put the political dimension in blunt terms. The concern in Congress and at Treasury is not that a U.S. or multinational company might acquire an asset from China. The concern is co-development.

“They say, you know, that the concerns are not so much about American companies or multinational companies acquiring assets from China. It’s about co-development,” Usdin said. The reason is structural: a licensing deal hands over a defined asset. A co-development deal transfers know-how, process, and organizational learning — the things that let a Chinese partner build a company that can compete globally on its own account.

That distinction matters because Treasury is in the process of incorporating biotech into the COINS Act, the Comprehensive Outbound Investment National Security framework from Representative John Moolenaar. Usdin’s read is direct: if large co-development deals become more common, they are “likely to fuel more restrictive rules” when Treasury writes the biotech provisions.

Selina Koch, BioCentury’s executive editor, noted that co-development and even co-commercialization terms are appearing in some deals — arrangements that give the China-based company a window into the Western regulatory and commercial system that a simple handoff does not. But Martz confirmed the three deals from the prior week were all exclusive licensing deals, some granting global rights. The pressure point is still ahead, not behind.

Koch also flagged a recent guest commentary by Frank LeDeaux, the former long-time McKinsey leader in China and founder of Cruising Ventures, who has lived in China for two decades. His argument: Chinese assets are going global, but that is not the same as Chinese companies going global. Until a Chinese company owns its U.S. P&L in the largest market in the world — rather than feeding assets to Western partners through out-licensing — a truly global company on the scale of Novo Nordisk is unlikely to emerge from inside China. Not impossible. But the path may be slower than the optimists predict.

Beam v. YolTech: same mutation, same technology, and a theory no one wants to defend

The Beam Therapeutics lawsuit against YolTech, filed September 25 in the U.S. District Court for the District of Massachusetts, is the rare trade secret case where the overlap between the two programs is not in dispute. YolTech’s AATD product is an LNP-delivered base editor that corrects the same mutation as Beam’s product.

“Same mutation, same technology. Doesn’t mean trade secret theft. It could mean fast follow,” Usdin said. The overlap, in other words, proves nothing by itself. What will decide the case is whether Beam can prove something harder: that a former scientist took not just technical information, but knowledge of what did not work, and that this “negative know-how” let YolTech reach a product faster than it otherwise could have.

Beam alleges that former scientist Emma Wang secretly co-founded YolTech in July 2021 while still employed at Beam, where she remained until January 2022. The complaint lists 19 categories of allegedly misappropriated technology and names YolTech CEO Jensen Wu, YolTech itself, and Serafa Bio — a newer company that has licensed YolTech’s AATD therapy, with Wang as CTO — as defendants. Beam seeks an injunction, return or destruction of its information, a constructive trust over license proceeds, assignment of YolTech’s Chinese patent application, and damages.

Usdin reached out to the defense and got responses. Wu called the claims “entirely without merit” but would not discuss specifics. Wang made three points: she cited a STAT story noting Beam CEO John Evans publicly dismissed the theft accusations in April; she said Beam cannot connect her alleged data access to any specific YolTech invention; and she said YolTech has contemporaneous R&D records tracing its work. Usdin noted she did not address the allegation that she started or helped start YolTech while still at Beam.

Beam’s complaint points to a patent with multiple claims Wang received, arguing it was implausible she could have invented those things on her own in the stated timeframe. But Usdin also pointed to a 2020 poster presentation — before Wang was hired — in which Beam put much of the needed information into the public domain. The poster laid out the steps but said more work was needed on mRNA and LNP formulation. Those refinements, Usdin expects, are what Beam will accuse Wang and YolTech of taking.

The negative-know-how theory is the case’s most contested element. Simone Fishburne, BioCentury’s editor-in-chief, pressed on whether it is demonstrable. Usdin said nobody he spoke to addressed it, called it a common feature of such cases, and drew a parallel to Cold War Soviet technology acquisition — where “having intimate understanding of what worked and what didn’t work” saved large amounts of time. He agreed it may be difficult to prove.

On Serafa, Usdin expects the defense to point to due diligence by RTW and RA Capital, the lead investors. And he noted this is not Beam’s first misappropriation claim: in 2025 Beam sued Prime Medicine over its AATD program, the dispute went to arbitration, and Prime prevailed. Prime and Beam share at least one co-founder and the same investors — a relationship that does not exist with YolTech.

“I think that the outcome of a trial, again, if it does go to trial, is going to hinge a lot on the two sides’ ability to make very complex scientific information accessible to jurors and, honestly, on which side the jurors choose to believe,” Usdin said.

GLOBE shrank from $11.9 billion to $440 million. The precedent is the point.

The GLOBE model — a mandatory most-favored-nation pricing scheme for Medicare Part B drugs, mostly injectables and biologics — was announced with a projected $11.9 billion in savings over seven years, all of it taken from the pharmaceutical industry. That figure has since collapsed to approximately $440 million.

MetricValueOriginally projected savings (7 years)$11.9 billionRevised projected savings~$440 millionCompanies CMS says will be hit4 (Usdin hears it may be 2)Scheduled start dateJanuary 1

The shrinkage comes from exemptions: companies that agreed to extend MFN pricing to Medicaid, plus gene and cell therapies, orphan-only drugs, plasma-derived products, and drugs facing biosimilar competition. But Usdin’s argument is that the shrinking dollar figure clarifies something important: the administration’s claim that the precedent — not the money — is the principal thing is actually true.

“The precedent really does matter more than the money here. And that precedent is using CMS’s demonstration authority to impose international reference pricing that Congress never enacted,” he said.

There is a second precedent, and Usdin thinks it will outlast the GLOBE model itself. “The other precedent that has been set is a precedent for pharma CEOs to go in and negotiate prices of particular drugs or portfolios of drugs with the President of the United States and with the administration.”

The problem for industry is one of its own making. The large pharmas that signed MFN deals and made public statements accepting foreign prices as a fair benchmark for U.S. prices are now in the position of saying they did not really mean it. Democrats already cite that precedent — and, Usdin predicts, when they hold power, they will try to apply it through the IRA and the Medicare drug price negotiation program.

GARD, the MFN model for Medicare Part D retail prescription drugs, is set to take effect January 1, the same date the administration pushed GLOBE to. Usdin expects litigation — probably from the trade associations — seeking an injunction to block both from taking effect.

Optogenetics won the Nobel, but the clinic is still mostly in the eye

The Nobel Prize in Physiology or Medicine went to Karl Deisseroth, Peter Hegemann, and Georg Nagel for discoveries related to light-gated ion channels and optogenetics. Koch explained the technology with welcome clarity: light-sensitive proteins called opsins — the same class of proteins that make retinal photoreceptors work — are genetically encoded and expressed under cell-type-specific promoters. That lets researchers activate or inhibit specific cell types and circuits in a way that conventional electrical stimulation, which activates multiple neuron populations at once, cannot match.

The research payoff has been enormous. Optogenetics lets neuroscientists move from observing correlations to establishing causation — activating a defined neural population and watching a specific behavior appear or disappear. Koch said this has been most valuable for understanding what goes wrong at the circuit level in dementias and psychiatric diseases, giving finer-grained views of disease biology than were previously possible.

The therapeutic footprint is much smaller, and mostly in the eye. The eye has a natural window for light, which is why optogenetic approaches are most advanced in diseases where photoreceptors degenerate: express an opsin in the next, still-intact cell in the synaptic chain, make that cell light-sensitive, and restore some lost visual activity.

Usdin interviewed Deisseroth in 2015, when he received the Lurie Prize — an award that often predicts Nobel winners. He described Deisseroth as a researcher driven by tools and methods rather than a single biological hypothesis. Asked then about the path to solving hard neuroscience problems like Alzheimer’s, Deisseroth argued against top-down moonshot programs and said progress would require basic research across disparate areas — and that the answer would come as a surprise, from research that did not set out to answer those questions. Koch agreed, citing CRISPR as another example of a tool-driven revolution that outran its initial therapeutic promise.

What to watch

Three threads converge on the same question: how much capability transfer the U.S. will tolerate. The China deal data shows earlier-stage, more novel assets moving to multinational pharmas at premium prices, and Usdin’s read is that co-development terms — not asset purchases — are what will drive Treasury’s biotech rules under COINS. The Beam case tests a harder version of the same question in court, where the outcome may turn on whether a jury accepts a negative-know-how theory that no one Usdin spoke to would address directly. And GLOBE’s collapse from $11.9 billion to roughly $440 million does not retire the precedent it set: international reference pricing imposed through CMS demonstration authority, and pharma CEOs negotiating prices with the White House. Both GLOBE and GARD are scheduled to start January 1, with injunctions likely sought before then. For investors, the throughline is that the China licensing boom is real, it is accelerating, and its political and legal boundaries are being drawn right now — in Treasury rulemakings, in a Massachusetts courtroom, and in the year-end drug pricing fights.