The US Treasury is drafting rules that will let American drugmakers keep licensing Chinese medicines, restricting only deals touching pathogens or biotechnology that could be weaponised. Reuters broke that on September 18, and it has been written up almost everywhere as a national-security story about where Washington draws a line with Beijing. That misses the commercial fact sitting underneath it. A broad restriction would have hit American pipelines before it hit Chinese ones, because almost half of the drugs US companies now license from abroad originate in China. Treasury is not going easy on Beijing. It is declining to cut off its own industry’s supply of molecules.
The dependency is recent and it is steep. GlobalData found that close to half of all US in-licensing deals in 2025 involved Chinese partners, and the pattern has held through 2026. Two deals show the shape of it. Bristol Myers Squibb signed with Jiangsu Hengrui Pharma on a package worth up to $15.2 billion across thirteen early-stage programmes in oncology, haematology and immunology. Pfizer signed with Innovent Biologics on a collaboration worth up to $10.5 billion spanning twelve oncology programmes.
Read the Deal Structure, Not the Headline Number
Those two figures get added together and reported as $25.7 billion flowing into Chinese biotech. That is not what is happening, and the structure of the Bristol Myers deal makes the point cleanly. Of its $15.2 billion, $600 million is cash upfront, two anniversary payments of $175 million follow in 2027 and 2028, and the remaining $14.25 billion is tied to development, regulatory and commercial milestones and to option exercises. Ninety-four percent of the headline number depends on programmes succeeding, and most early-stage programmes do not.
What American pharma is actually buying is option value, cheaply. The Pfizer arrangement is similarly graduated: a global licence on four programmes where Pfizer carries all development costs, an ex-Greater China licence on four more where it carries most of them, and four co-developed globally with shared costs and split US and European profits while Innovent keeps China rights. These are not acquisitions. They are structured bets that pay out only if the science works, which is precisely why they have become the default way to refill a pipeline.
That distinction matters for the policy question. A rule restricting outbound pharmaceutical investment would not have stopped $25.7 billion leaving the country, because $25.7 billion was never leaving. It would have removed a cheap source of clinical-stage assets from American companies while leaving those assets available to European and Japanese buyers.
Who Actually Wanted the Restrictions
The lobbying tells you more than the rulemaking does.
The push for tighter limits has come from midsize and smaller US biotech, not from national-security hawks alone. Ginkgo Bioworks’ chief executive met Treasury officials to argue that continued outbound investment to China will damage the American and European industry over time. Read that argument in its commercial context. When Pfizer can option four Chinese oncology programmes for a fraction of the cost of buying a US biotech outright, the US biotech does not get bought. Restricting Chinese licensing would not create American science. It would reduce the competition for American science, and raise what big pharma has to pay for it.
A rule sold as protecting American innovation would mostly have protected American valuations.
That is a legitimate thing for a domestic industry to want. It is not a security argument, and it should not be allowed to travel as one.
Where We Come Out
Treasury has the narrow question right, and the pathogen carve-out is the correct place to put the line. Work on dangerous pathogens and weaponisable platforms is a genuine national-security category with a defensible boundary. Oncology assets are not, and pretending otherwise would have been industrial policy wearing a security badge.
Where we would push back is on the process. These rules are being written in the shadow of a summit: Treasury is unlikely to publish anything before Trump and Xi meet on September 24, the drafts are explicitly subject to change, and officials acknowledge the President could intervene and reset them. That means a rule with multi-year consequences for drug development is currently a negotiating chip. Pharmaceutical R&D runs on ten-year horizons and cannot price a policy that moves with the diplomatic calendar. If the pathogen line is the right line, Treasury should publish it and defend it independently of what happens in Washington on Thursday.
The deeper problem is one no rule addresses. American companies are in-licensing from China at this rate because Chinese labs are now producing clinical-stage oncology assets faster and cheaper than the US biotech sector is. That is a competitiveness outcome, not a loophole. Restricting the licences would hide it rather than fix it, and would leave US patients waiting longer for drugs that exist.
Three Tests for Whether This Holds
Whether the published rule keeps the pathogen boundary narrow or expands it into vague categories like sensitive technologies, which is where scope creep always begins. Whether the Thursday summit produces any pharmaceutical language at all, which would confirm the rules were leverage. And whether the in-licensing rate holds: if US companies keep signing Chinese deals into 2027 at close to half of all in-licensing volume, the domestic biotech sector has a structural problem that no outbound-investment rule was ever going to solve.
The rule as drafted is the right call. The reason it is the right call is not the one being reported.