San Francisco, 12–15 January. The 44th JP Morgan Healthcare Conference — biotech’s annual January migration to Union Square, the Westin St. Francis and every rentable meeting suite within four blocks — produced its usual weather of hallway huddles, non-consecutive lunches, and 300-slide corporate deck reveals. What it did not produce was the megamerger everybody had booked their week around. The single biggest storyline out of JPM26, once the noise settled, was that the industry had decided this cycle’s deal-making would go through licensing agreements, precision partnerships, and — increasingly — Chinese biotech, rather than through the ten-figure biopharma acquisition of prior years.
The mood was more optimistic than in 2024 and 2025, but disciplined. The IPO window, closed for most of the recent cycle, cracked open. Nine Western-pharma-to-China licensing agreements had already been announced by the time the ballrooms filled on Monday morning. And two announcements — Novartis on the licence-in side, and NVIDIA-Eli Lilly on the AI-for-drug-discovery side — set the tempo for a year in which the pharmaceutical industry has decided to buy scientific capability rather than build it.
All the meetings, none of the megadeals
Every year at JPM, the industry rumour mill fires an unreasonable number of ten-billion-dollar mega-M&A speculations. In most years, one or two land. In 2026, none did — at least not during the week itself. The absence was noted, and it was strategic rather than accidental. Multiple large-pharma CEOs used their fireside chats to argue that the current cycle favours internal pipeline prioritisation and billion-dollar licensing over acquisition. Takeda’s R&D head said as much on the record; several of the large-cap CFOs said it more diplomatically in the same direction.
The strategic logic is coherent. Late-stage biotech assets are being priced at levels that make outright acquisition uncomfortable relative to the risk of the underlying data. Licensing agreements, with milestone-based upside for the seller and staged commitment for the buyer, distribute risk more sensibly. And in a cycle where patent-cliff exposure at the largest pharmas is well-flagged and well-financed, the acquisition option remains available for the specific late-stage assets that clear an internal hurdle. The mega-M&A is not off the table for 2026. It is on a shorter list.
Novartis picks up the licence-in playbook, twice, both in China
The two most-talked-about deals of the week both had Novartis’s name on them. Novartis signed a worldwide licensing and collaboration agreement with China-based SciNeuro Pharmaceuticals worth more than $1.6 billion for a set of de-novo amyloid-beta antibody candidates targeting Alzheimer’s disease. It was, on any reasonable read of the market, the most consequential Alzheimer’s-adjacent commercial move of the week — a large pharma writing a nine-figure upfront and a nine-figure milestone envelope for an area where the science remains contested and the commercial upside is enormous if the trials clear.
Novartis also entered a worldwide licence agreement for approximately $50 million with China-based biotech Zonsen PepLib Biotech for global rights to an undisclosed peptide-based radioligand therapy for oncology indications. The Zonsen deal is smaller and more targeted, but the pattern is the same: a Western pharma paying a Chinese biotech to acquire commercial rights to a scientific asset the Chinese biotech generated. That is the story of the week for the biotech industry.
Nvidia and Eli Lilly stand up a $1B AI-for-drug-discovery lab
On the AI-for-life-sciences side of the show, the announcement that landed with the most weight was Nvidia and Eli Lilly’s five-year, $1 billion commitment to a “co-innovation AI lab” targeting the hardest problems in drug discovery. The scope, on the joint press-conference framing, includes foundation models for molecular representation and design, multimodal protein-and-ligand interaction modelling, computational chemistry pipelines, and — importantly — the compute infrastructure and MLOps that make those pipelines commercially reproducible at pharma scale.
The deal is consequential for two reasons. First, it is a specific dollar commitment to a specific capability, on a specific timeline, at a scale that requires board-level sign-off at both companies. That is a different signal than the usual “AI transformation partnership” press release. Second, it formalises the relationship between one of the world’s largest AI-infrastructure vendors and one of the world’s most successful drug-discovery companies at a moment when the drug-discovery-AI category is repricing rapidly. Where Lilly and Nvidia go, others will need a coherent answer to.
The China chapter is now the industry’s most consequential
Perhaps the most under-appreciated statistic of the week was that nine deals between Western pharmaceutical companies and Chinese biotechs had already been announced by the opening of JPM 2026. That is not a Chinese biotech story or a Western pharma story. It is a pharmaceutical-industry story: roughly one-third of the innovative-medicine pipeline is being sourced from China. Western pharma is now systematically licensing-in from that pipeline. Chinese biotechs are increasingly comfortable with global commercial partnerships that keep them on the science side and put Western pharma on the commercialisation side.
The geopolitical overlay is real. Export controls, technology-transfer scrutiny, and the political posture toward China-anchored capital have not softened. What has changed is that the industry has decided the science is too important to leave off the table for policy reasons, and is structuring deals — licensing rather than acquisition, milestone-based rather than upfront-heavy, with clear IP and territory splits — to work inside the political constraints.
Novo, obesity, and the direct-to-patient learning curve
Novo Nordisk used its fireside session to acknowledge, publicly, what most of the room already knew: it has ceded the obesity-market leadership position to Eli Lilly, at least for the current generation of GLP-1 receptor-agonist products, and is now — in the CEO’s phrasing — focused on “mastering” direct-to-patient sales. The candour was appreciated; the strategic challenge is real. Novo remains one of the world’s most successful drug-development companies and has the follow-on pipeline (oral, longer-acting, combination assets) to compete. The question is whether the commercial infrastructure catches up in time to matter.
IPO window cracks open
The biotech IPO window, closed for most of 2024 and 2025, cracked open. A high-profile biotech IPO priced successfully in the run-up to the conference, and two more high-profile biotech IPO filings were made during the week itself. The window is not the 2020 or 2021 window; it is a more disciplined one. But it is open, and the crossover-investor community and biotech CFOs are actively lining up books for Q1 and Q2 pricings.
What JPM 2026 actually moved
- Licensing is the shape of the 2026 pharma deal book. Mega-M&A is on a shorter list.
- China is the industry’s licence-in pipeline. Nine deals before the conference opened. More during the week.
- AI-for-drug-discovery got its first billion-dollar commitment. Nvidia-Lilly sets the market bar.
- The IPO window is open. Disciplined, but open.
- Novo Nordisk conceded the obesity moment. The commercial rebuild starts now.
Open questions
Do the SciNeuro / Zonsen and equivalent Chinese-sourced assets actually clear their pivotal-data hurdles at Western regulators, or does the pipeline sourcing prove more geopolitically constrained than the deal flow implies? Does the Nvidia-Lilly AI lab produce reproducible programme wins by 2027, or does drug-discovery AI take another cycle to graduate? And does the IPO window widen through H1 2026 to absorb the biotech candidates queueing behind the current filers?
Bottom line: JPM 2026 was the year the biotech industry stopped waiting for the next mega-merger and started running the licensing playbook. The deals are smaller, more numerous, and better structured than the acquisitions of the prior cycle. The China chapter is not an anomaly; it is the industry’s new sourcing model. And the AI-for-drug-discovery category has, with Nvidia-Lilly, its first industrial-scale reference deployment. From here, the follow-through is data — pivotal readouts, IPO absorption, licence-in productivity — rather than announcements.
