Capital, Chokepoints and Cost: What this October Meant for Biopharma

A bill called the Biotech Investment National Security Act (BINSA), introduced in the House of Representatives on 2 June and in the Senate on 6 August, would bring biotech under US limits on American investment in China. It would do so by extending the COINS Act, a law passed in December 2025 that lets the US Treasury restrict American investment in Chinese companies working on sensitive technology. That law currently covers chips, artificial intelligence, quantum computing and supercomputing.

While Congress considers that bill, the US Treasury Department has started writing its own rules for investment in Chinese biotech, and they are narrower than many expected. In mid-September, Reuters reported that draft rules would let US companies keep licensing most drugs from Chinese firms, excluding pathogens and weaponisable biotech. The proposed rule published on 22 October largely bears this out. It adds biotech to the prohibited list, but standard in-licensing, meaning a company buying the rights to a drug a Chinese firm developed, is excluded. Law-firm analyses warn that licensing bundled with an equity stake, or structured as a joint venture or profit-sharing co-development, could be caught. Comments close on 7 December; the final rule is due by 13 March 2027.

How much those rules matter depends on how heavily the US drug pipeline already relies on Chinese-developed medicines. GlobalData, a data provider, counted almost half of US in-licensing deals in 2025 as Chinese-sourced. China holds about 50% of global out-licensing deals against roughly 28% for the US, and the average upfront payment has risen from $52 million in 2022 to $172 million in early 2026. John Crowley, chief executive of the Biotechnology Innovation Organization, the US industry body, said “we cannot restrict our way to success”. Beijing has answered with its own draft export controls, discussed below. Both governments are drawing one line: buying a drug is tolerable, owning the platform that makes it is not.

Separately from the China question, the USFDA changed its rules on 21 September to make it easier for drug developers to replace animal testing with other methods. The streamlined rule lets drug sponsors use non-animal methods, such as lab-grown human tissue models and computer simulations, to show a drug is safe before human trials. Acting Commissioner Kyle Diamantas said the aim was “not to replace one rigid approach with another.” The rule does not ban animal studies or change evidentiary standards. The burden shifts to developers to validate novel evidence.

Large drugmakers are spending heavily on AI to shorten early-stage drug development, and the industry’s forecasts are ambitious. One forecast cited in a Reuters factbox says AI could halve early-stage timelines and costs within three to five years.

The evidence that AI actually speeds up drug development is thinner than the spending suggests. Isomorphic Labs, an AI drug-design company spun out of Google DeepMind, raised $2.1 billion in May, yet as of 30 September had not announced an application to begin human trials, and its target for starting them has slipped from the end of 2025 to the end of 2026. Insilico Medicine, a Hong Kong-listed AI drug developer whose lung-disease candidate rentosertib is now in Phase 3, the final stage before approval, is the stronger proof point.

Europe’s drug supply has a different weakness: it depends on China for the basic ingredients of common antibiotics. Eurostat, the EU’s statistics office, shows China’s share of EU imports of penicillin and its derivatives rising from 44% in 2019 to 86% in 2025. Sandoz, the Swiss-based generic drugmaker, runs the plant at Kundl in Austria, the last fully integrated penicillin facility in the West. Its chief executive, Richard Saynor, says it is not commercially sustainable without higher payments, and that governments “see the price, but they don’t necessarily see the value.” The Council on Foreign Relations, a US think tank, found that for 20 of 56 active ingredients German regulators call essential, Europe has no ready alternative to China. Saynor warns that antibiotic supply “could feasibly become a tool of economic coercion.” Export controls on one side invite leverage on the other.

AI is becoming a core input to drug discovery, which links pharma’s research pipeline to technology markets that are now under stress. Eli Lilly’s LillyPod, an AI supercomputer switched on in February, runs 1,016 Nvidia Blackwell Ultra GPUs (the graphics chips that train AI models) to build models that design proteins and small molecules. Lilly says a wet-lab team tests about 2,000 molecular ideas per target per year; its computational “dry lab” can screen billions. Novo Nordisk signed with Anthropic, the AI company behind the Claude models, on 16 September to apply them across R&D. Thomas Fuchs, Lilly’s chief AI officer, concedes the limit: data remains a constraint, and “you cannot accelerate a cancer trial or a neuro trial.”

The financial risk from this link falls mainly on smaller AI-focused biotech firms, which rent computing power and sell their services to large drugmakers. Iambic Therapeutics, an AI drug developer backed by Nvidia, signed AbbVie, filed for a Nasdaq listing on 21 September and priced a $150 million initial public offering (IPO) on 14 October. That was the day CoreWeave, a heavily indebted company that rents out AI computing capacity, fell 15% and the Nasdaq had its worst session since July. AVANT, a research group, counts third-quarter biotech venture funding at $2.3 billion across 64 deals, against $4.1 billion across 132 in the second quarter, with later-stage rounds taking about 95% of dollars.

Funding for AI companies is tightening in both China and the US, and the strain is visible in stock indexes and credit markets. China’s STAR50 index has fallen 40% in under four months, erasing about $480 billion, Reuters reports. The STAR Market is a board of the Shanghai Stock Exchange, launched in 2019 as China’s answer to the Nasdaq, while the STAR50 tracks its 50 largest companies. In the US, Justin Biemann of JPMorgan Private Bank wrote that “the AI label alone is no longer enough to reassure lenders.” One AI infrastructure borrower that depends on financing trades at 1,100 basis points on credit-default swaps, contracts that pay out if a borrower defaults; that price implies a near-50% chance of default within five years. Healthcare has been the beneficiary. The S&P 500 Healthcare Index is up about 23% since April, healthcare funds drew roughly $4 billion over 65 days while technology funds lost $6 billion, and Matt Orton, chief market strategist at Raymond James Investment Management, calls the sector “an anti-AI trade.” But Lilly, AbbVie and Johnson & Johnson lead that rally. Incumbents gain; venture-stage AI-bio firms do not obviously share it.

Recent research on making AI text generation faster shows how efficiency gains could help drug developers while hurting the lenders that finance AI infrastructure. A Stanford preprint on next-concept modelling, a method in which an AI model first predicts the broad concepts of a text and then fills in the words, reports generation 10 to 31 times faster than diffusion baselines, the older approach it is compared with. Diffusion is also the model class behind Lilly’s protein work. The paper tests text only and reports neither lower memory use nor heavier use of the computer’s central processor, so that premise is unsupported. If such gains transfer to biology, a developer pays less per candidate, while the arithmetic worsens for leveraged data-centre borrowers priced for rising demand.

The US-China contest over chips also reaches pharma, because both governments are now restricting AI and biotech technology together. The MATCH Act, a bill in Congress that is not yet law, would give the Netherlands, Japan and others 150 days to match US controls on lithography, the machines that print circuits onto chips. China accounted for 29% of 2025 sales at ASML, the Dutch maker of those machines. Beijing’s reply is revealing: its 16 October draft export catalogue treats AI and biopharma in one instrument. Of 40 technologies, 14 would be prohibited from export and 26 restricted, including AI-chip design and computing-cluster know-how alongside cell and gene therapy and siRNA platforms. White & Case, a law firm, advises companies to check model weights, the trained parameters of an AI model, and training tools developed in China. Chinese AI-drug firms such as Insilico and METiS TechBio raised capital through Hong Kong listings. Hong Hao, chief investment officer of Hong Kong-based Lotus Asset Management, says the STAR sell-off “is now going to put a stop for many of the companies trying to get themselves listed.” If that caution spread to Hong Kong, it would narrow their exit. That is an inference, not a reported fact.

Pharma’s growing reliance on digital systems and large datasets also makes it a bigger target for cyberattacks. Quest Diagnostics, a large US laboratory-testing company, disclosed on 22 October a breach of its Quanum Lab Services Manager, a system clients use to order tests and read results. Novo Nordisk said in June that attackers copied pseudonymised clinical-trial data, and a May ransomware attack disrupted manufacturing and shipping at West Pharmaceutical Services, a supplier of injectable-drug components.

San Diego’s biotechnology ecosystem is booming. At its heart, it consists of a mosaic of startup and mature companies working on the honorable task of discovering new therapies for patients, but it’s also peppered with organizations helping turn promising new drug candidates into clinical-stage products, and beyond.

In our inaugural San Diego spotlight feature, we’re taking a closer look at Catalent’s San Diego operations. From early development through clinical supply, Catalent’s San Diego facilities provide a connected pathway designed to help biotech companies move candidates into the clinic faster and with less complexity.

A 68,000-square-foot development facility supports early-phase small-molecule and peptide drug candidates through preformulation testing, formulation and analytical development, GMP manufacturing, stability testing, clinical packaging, with strategies for compounds with poor solubility.

What makes Catalent San Diego unique is the seamless transition from development to clinical supply. As programs progress, sponsors can leverage two closely connected facilities to support development, packaging, distribution, and clinical trial execution through a single integrated network.

Catalent’s purpose-built clinical supply facility extends that support through clinical supply management, packaging, complex labeling, cold-chain services, storage and global distribution. Together, the two facilities can help biotechnology companies move candidates from early development through clinical supply, helping sponsors reduce complexity and maintain momentum as programs advance towards clinical trials, all within San Diego’s growing life sciences community.

Rising inflation in the US and Europe, driven largely by an energy shock from the Iran war, is raising drugmakers’ costs and pushing central banks to raise interest rates. US consumer prices rose 0.6% in September and 3.7% over twelve months, from 2.4% in February, driven by an energy shock from the Iran war; core inflation, which excludes food and energy, was 2.6%. The Federal Reserve raised its benchmark interest rate by 25 basis points (a quarter of a percentage point) in September to 3.75–4.00%, its first increase in three years; markets expected a hold at this week’s meeting and a hike in December. Euro-area inflation reached 3.8%, raising costs for energy-intensive plants such as Kundl.

Higher interest rates hit biotech shares harder than most because the sector’s profits lie far in the future, and long-term borrowing costs have risen sharply. The ten-year Treasury yield, the interest rate on US government borrowing that anchors long-term borrowing costs, climbed from 4.16% in January to 5.31% on 5 October and sat at 4.99% on 23 October. Biotech equities are sensitive to it: one analysis estimates about 1.85% off the XBI, a widely followed fund of US biotech shares, for every 10 basis points of yield. The XBI is up 22.6% this year but 11.8% below its August high.

Drugmakers face rising costs but have limited freedom to raise prices in response. Medicare inflation rebates require drugmakers to repay money when prices on covered drugs rise faster than inflation, so higher general inflation widens the room but does not remove the cap. Most-favoured-nation (MFN) deals tie US prices to the lowest prices paid abroad, and the midterm elections on 3 November make drug costs a campaign subject. Section 232 tariffs, imposed on national-security grounds, entered their broad phase on 29 September, with a 100% default rate on patented drugs, a 15% cap for the EU, and a 0% rate for firms with MFN and US-manufacturing commitments.

From our partners

DIT AgTech turns one free hardware installation into years of recurring supplement revenue. Ranchers commit to a three-year nutrition plan, while methane-reducing nutrition creates carbon credits that generate an additional revenue stream for both ranchers and DIT AgTech. Proven in Australia. Now scaling into the massive U.S. livestock market.

Invest in DIT AgTech Today

^{𝘐𝘯 𝘮𝘢𝘬𝘪𝘯𝘨 𝘢𝘯 𝘪𝘯𝘷𝘦𝘴𝘵𝘮𝘦𝘯𝘵 𝘥𝘦𝘤𝘪𝘴𝘪𝘰𝘯, 𝘪𝘯𝘷𝘦𝘴𝘵𝘰𝘳𝘴 𝘮𝘶𝘴𝘵 𝘳𝘦𝘭𝘺 𝘰𝘯 𝘵𝘩𝘦𝘪𝘳 𝘰𝘸𝘯 𝘦𝘹𝘢𝘮𝘪𝘯𝘢𝘵𝘪𝘰𝘯 𝘰𝘧 𝘵𝘩𝘦 𝘪𝘴𝘴𝘶𝘦𝘳 𝘢𝘯𝘥 𝘵𝘩𝘦 𝘵𝘦𝘳𝘮𝘴 𝘰𝘧 𝘵𝘩𝘦 𝘰𝘧𝘧𝘦𝘳𝘪𝘯𝘨, 𝘪𝘯𝘤𝘭𝘶𝘥𝘪𝘯𝘨 𝘵𝘩𝘦 𝘮𝘦𝘳𝘪𝘵𝘴 𝘢𝘯𝘥 𝘳𝘪𝘴𝘬𝘴 𝘪𝘯𝘷𝘰𝘭𝘷𝘦𝘥. 𝘋𝘐𝘛 𝘈𝘨𝘛𝘦𝘤𝘩 𝘩𝘢𝘴 𝘧𝘪𝘭𝘦𝘥 𝘢 𝘍𝘰𝘳𝘮 𝘊 𝘸𝘪𝘵𝘩 𝘵𝘩𝘦 𝘚𝘦𝘤𝘶𝘳𝘪𝘵𝘪𝘦𝘴 𝘢𝘯𝘥 𝘌𝘹𝘤𝘩𝘢𝘯𝘨𝘦 𝘊𝘰𝘮𝘮𝘪𝘴𝘴𝘪𝘰𝘯 𝘪𝘯 𝘤𝘰𝘯𝘯𝘦𝘤𝘵𝘪𝘰𝘯 𝘸𝘪𝘵𝘩 𝘪𝘵𝘴 𝘰𝘧𝘧𝘦𝘳𝘪𝘯𝘨, 𝘢 𝘤𝘰𝘱𝘺 𝘰𝘧 𝘸𝘩𝘪𝘤𝘩 𝘮𝘢𝘺 𝘣𝘦 𝘰𝘣𝘵𝘢𝘪𝘯𝘦𝘥 𝘩𝘦𝘳𝘦: https://bit.ly/4bzuWCi​}

Over the short term, through March 2027, financing conditions are likely to matter more to biopharma than regulation does. Because the Treasury rule is narrower than feared, licensing volumes should hold through the 7 December comment deadline; the likelier shift is toward plain licences and away from equity-linked or co-development structures. Higher yields and a possible December hike weigh on smaller biotechs; large companies with MFN and US-manufacturing agreements are shielded from tariffs.

The slump in AI financing could help or hurt biopharma, depending on where the money goes. Capital leaving AI has gone mostly into large-cap healthcare and Treasuries, not into venture-stage biotech, and a flight to bonds pushed yields below 5% in the week of 19 October. Smaller biotechs gain only if yields fall, which requires inflation to ease, and the Fed sees AI demand as a short-term source of price pressure. Tighter data-centre credit also shrinks the budgets behind AI-pharma partnerships, so the 2026 cohort of collaborations probably carries more counterparty risk than its announcements imply.

A change in one market will spread to others, particularly between the US, Europe and China. US tariffs push companies to invest at home; European prices must rise to keep them there, yet European energy costs leave governments less able to pay. European generics makers meanwhile face the widest cost gap with China just as Beijing gains leverage from American restrictions. Over the long term, that tilts Europe toward state-funded procurement.

The longer-term danger is that biopharma’s main dependencies fail together. Biopharma now depends on Chinese science for its pipeline, Chinese factories for its ingredients, and AI-linked credit for its tools. If a single shock, a credit event in AI or a retaliatory restriction on ingredient exports, hits two of these at once, how many companies hold the balance sheet to bridge both?