In May 2023, a man named Michael Thompson sat across from his oncologist in Iowa and was told he had stage 3 throat cancer. Two months later, in July, the same oncologist had to tell him something else.
The drug they’d planned to use for his treatment — cisplatin, the standard-of-care chemotherapy for his cancer for the last forty years — wasn’t available. He’d be getting carboplatin instead. A close cousin of cisplatin. Less effective for his specific cancer.
“You can’t prepare yourself for that kind of news,” Thompson told NBC News a few weeks later. “Dealing with cancer is hard enough. But you don’t want to be hit with the news that the treatments that you need aren’t available.”
His doctors told him that if the substitute didn’t work, they might have to surgically remove his larynx. He’d never speak again.
Now read the next sentence carefully, because this is the part that broke something in my brain when I came across it.
The reason Michael Thompson couldn’t get the right chemotherapy in Iowa in 2023 was that, six months earlier, three FDA inspectors had walked into a factory in a place called Matoda, on the outskirts of Ahmedabad in Gujarat, India.
What they found, according to their 36-page inspection report, was a pattern they later described in writing as a “cascade of failure.” Employees pouring acetic acid into a trash bin to destroy testing documents. Plastic bags filled with shredded quality-control records. A truck waiting at the loading dock, also full of shredded paper, dated from the days the inspectors were on site.
The factory was Intas Pharmaceuticals’ Ahmedabad plant. And it produced roughly 50% of the cisplatin sold in the United States.
After the inspection, Intas voluntarily shut the plant down. Within four months, 93% of US cancer centers were reporting carboplatin shortages. 70% were short on cisplatin. Hospitals started writing rationing protocols — internal documents that decided, for the first time since the 1970s, which American cancer patients would get the standard chemotherapy and which ones would get a substitute.
Cisplatin and carboplatin treat about 500,000 new cancer cases in the US every year. Bladder, brain, throat, cervical, lung, ovarian, testicular. They’re not exotic drugs. They’ve been on the market since the 1970s. They’re cheap. They’re mostly off-patent.
And the entire American supply ran through one factory in India, where employees were caught shredding documents in front of FDA inspectors.
That’s where my obsession started.
The number that nobody in finance is running
Here’s a small definition before we go any further, because we’ll use the term constantly: an API, or active pharmaceutical ingredient, is the molecule that actually does the work in a drug. The acetaminophen inside a Tylenol. The metformin inside a diabetes pill. The semaglutide inside Ozempic. Everything else — the binders, the coatings, the capsules — is essentially packaging.
No API, no medicine. Just sugar and dye.
Now, the chart that got me to drop everything I was working on:
As of August 2024, only 24% of API manufacturing facilities for drugs sold in the US were located on US soil. The rest — more than three-quarters — sit in China, India, and the EU.
But that number is the polite version of the story. Because the dirty secret of the global drug supply is that India, which the US treats as the friendly democratic alternative to China, buys roughly 70% of its own APIs and key starting materials from China. For critical antibiotics like penicillin and cephalosporins, the figure climbs to 90%.
When you multiply through, the actual exposure is brutal. A Senate testimony from the University of Minnesota’s PRIME Institute estimated that 46% of US daily generic drug doses contain source materials from China.
Almost half. Of every prescription generic taken every day in America. Routes back, somewhere upstream, to a Chinese chemical plant.
And generics, by the way, are not a niche category. They are 90% of all US prescriptions.
The line that finally made the politicians listen
There’s a story Joe Jackson — the CEO of Jackson Healthcare, which bought and reopened the last functional antibiotics factory in the US — has told publicly several times. I’ll paraphrase it to keep things tight.
During the 2017 US-China trade negotiations, the American delegation was making demands designed to shrink the trade deficit. The Chinese delegation listened, patiently, the way they always do. Then one of them said something to the effect of: you can do whatever you want, but you can’t win a war with us. We make 100 percent of your antibiotics.
At the time, that was factually true.
The last US plant making penicillin APIs had shut down in 2004 because it couldn’t compete on price with Chinese producers. The Bristol-Myers Squibb facility in East Syracuse, New York. After it closed, every gram of penicillin precursor consumed in the United States — every one — came from overseas. Mostly from one country.
The plant in Bristol, Tennessee that Jackson Healthcare bought out of bankruptcy and restarted in 2022 makes finished antibiotic doses, but it still imports its key starting material from abroad. There is, at this moment, no fully integrated penicillin manufacturing chain on US soil. Not one.
Now layer the post-COVID lessons on top.
In March 2020, India banned the export of 26 essential APIs and finished drugs, including paracetamol. The reason given was domestic security. The real reason was that India was unsure whether China would keep shipping it the raw materials it needed. The ban lasted three weeks. Three weeks. Long enough for European hospitals to see paracetamol stocks evaporate. Long enough for US pharmacy chains to start rationing. Long enough for the price of certain lots to triple.
Three weeks of the world’s largest generic exporter pulling back, and the entire Western pharmaceutical inventory started to crack.
Nobody in Washington wanted to ask out loud what would happen if India did that for three months. Or what would happen during a Taiwan crisis. Or what happens when Chinese export controls — the same kind already used on gallium, germanium, and rare earths — get applied to APIs.
But that’s the math the market is going to be forced to run, sooner than most people realize.
Why the law of December 18, 2025 matters more than anyone is saying
On December 18, 2025, President Trump signed the Fiscal Year 2026 National Defense Authorization Act into law.
Buried inside it, as Section 851, was the BIOSECURE Act.
Quick translation, because the name doesn’t tell you what it does. BIOSECURE prohibits any company that receives US federal funding — which means essentially every major US pharmaceutical company, since they all touch Medicare, Medicaid, the VA, or NIH grants — from using equipment or services from designated “biotechnology companies of concern.”
Right now, the law’s automatic blacklist is the Department of Defense’s 1260H list. That list already includes BGI Group, MGI, and a handful of other Chinese genomics firms. Crucially, it does not yet include WuXi AppTec or WuXi Biologics — the two giants that, between them, account for a meaningful share of global biopharma contract manufacturing and research.
But here’s what the market is missing.
In December 2025, the chairs of multiple Senate and House committees — including the House Select Committee on China — sent a formal letter to the Department of Defense recommending that WuXi AppTec, WuXi Biologics, and WuXi XDC be added to the next 1260H update. The Pentagon updates that list every January. According to several legal advisories I’ve read, the next refresh is expected to designate WuXi.
The moment that happens, every existing US biopharma contract with WuXi enters a five-year wind-down. Every new contract is illegal.
This is not theoretical. The “de facto BIOSECURE effect” has already been running through the industry since 2024. Multiple US biopharma companies stopped signing new contracts with WuXi the moment the legislation looked likely to pass. Western contract manufacturers — Lonza, Samsung Biologics, FUJIFILM Diosynth, Thermo Fisher’s pharma services arm — have been quietly absorbing displaced capacity at premium pricing.
Total disclosed CDMO capacity investment in 2025 hit $24.86 billion, with $18.48 billion of that flowing into the United States. Western CDMOs with available US or allied-nation capacity now trade at 15–20x EBITDA, and the highest-quality names trade higher.
That’s not a re-rating that ends when AI capex slows or when Trump leaves office. That’s a structural reallocation of one of the largest supply chains in the world economy. And it’s barely begun.
The GLP-1 angle nobody’s pricing correctly
If you want a single case study of how brutally physical the bottleneck is, look at GLP-1s — Ozempic, Wegovy, Mounjaro, Zepbound. The drugs that have rewired the obesity and diabetes markets and that analysts now project will be a $150 to $200 billion category by 2030.
The molecule isn’t the problem. Novo Nordisk and Eli Lilly have plenty of semaglutide and tirzepatide in synthesis tanks. What they don’t have is fill-finish capacity — the sterile final step where the drug gets injected into pens, vials, or cartridges. That step is among the most regulated, capital-intensive, and slow-to-build operations in all of pharmaceuticals.
For more than two years, GLP-1 supply was capped not by demand, not by molecule, but by a few dozen sterile filling lines around the world.
So in February 2024, Novo Holdings — the parent of Novo Nordisk — bought Catalent, the largest pure-play CDMO on the planet, for $16.5 billion. As part of the deal, three Catalent fill-finish sites in Bloomington (Indiana), Brussels, and Anagni (Italy) were transferred directly to Novo Nordisk for an additional $11 billion.
Eli Lilly, which had also been using Catalent, called the deal “an oddity” and lobbied antitrust regulators against it. They lost. The deal closed at the end of 2024.
In response, Lilly has now committed over $50 billion to its US manufacturing buildout since 2020 — Lebanon, Indiana for tirzepatide API, plus new sites in Wisconsin, Pennsylvania, North Carolina, Virginia, and Texas. Through 2027, the company will have invested about $23 billion in capital expenditures specifically to increase fill-finish capacity. Novo has matched with $4.1 billion in Clayton, North Carolina, plus expansions in Denmark.
Combined GLP-1 manufacturing capex announced or under construction since 2024 is now north of $100 billion, with new facilities not coming online until 2027–2029.
The reason that capex matters for investors goes well beyond Novo and Lilly. Every fill-finish line built for a GLP-1 is one that isn’t available for the next blockbuster. Every peptide synthesis tank booked by Lilly is one that an emerging biotech can’t access. The CDMOs that have surplus sterile capacity right now — and a small handful of equipment suppliers and aseptic specialists that feed them — are sitting on what is functionally the most valuable real estate in pharmaceuticals.
And almost none of them are priced like it.
What the consensus is still getting wrong
The current take on pharma supply chains, from most generalist investors I talk to, goes something like this: “Sure, there’s a story, but reshoring takes years, generic margins are terrible, and the real money in pharma is still in IP and biotech.”
Half of that is right. Reshoring does take years. That’s exactly the point.
What the consensus misses is that the buildout has already started, the capital is already committed, and the political tailwind has already passed the threshold of being reversible. BIOSECURE is law. CHIPS-style subsidies for pharmaceutical manufacturing are working their way through Congress. The Defense Production Act has already been used to allocate funding for domestic API production, and a Pentagon pilot program with the lab Valisure is now testing 40 prescription drugs deemed critical to military readiness.
Even if a Trump-Xi grand bargain materialized tomorrow morning at 9 AM, none of this stops. Because pharmaceutical reshoring isn’t only about China. It’s about four converging structural drivers, none of which depend on a single political relationship:
Drug shortages have become structural, not cyclical. As of late 2024, the FDA was tracking around 100 active shortages, with more than half lasting over two years. The 2023 cisplatin crisis was the most severe in modern oncology history, and it was caused by a single foreign factory. Generic margins are too thin to support redundant supply, so any disruption now cascades.
GLP-1 demand is rebuilding the entire fill-finish industry. $100 billion+ of capex, multi-year lead times, and a market that’s still under-supplied at the global level. This wave will run through 2030 minimum.
BIOSECURE and tariff overlay are forcing CDMO re-shoring. $24.86 billion of CDMO capex in 2025 alone, 74% of it US-bound. Premium valuations for Western capacity. A bifurcated valuation landscape that favors specific names.
AI-driven drug discovery is increasing the pipeline of complex modalities — peptides, antibody-drug conjugates, cell therapies, oligonucleotides — that cannot be made in conventional small-molecule plants. Each of these requires specialized CDMO capacity that takes 4–7 years to build.
The institutional money to fund this catch-up is in motion. The IRA already allocates billions to onshoring. The Defense Production Act has been activated. The EU is rolling out its Critical Medicines Act. Japan, Korea, and India are running parallel sovereignty programs.
These aren’t think-tank reports. They are signed budgets, awarded contracts, broken ground.
And in the public markets, the leaders haven’t yet been re-rated to reflect any of this. Most CDMOs and specialty pharma capacity names trade at multiples that imply normal generic-cycle economics. Several of them are sitting on order books that already cover three-plus years of revenue.
Why I built this playbook
For the last three months, I’ve been doing on this sector exactly what I did on energy infrastructure earlier this year.
I read FDA inspection reports going back to 2019, looking at which specific facilities account for what share of US supply for the 200 drugs the FDA classifies as essential. I went through the financials of every Western CDMO with more than $500 million in revenue. I built a model of GLP-1 fill-finish capacity by site through 2030. I tracked 1260H list updates and BIOSECURE legislative milestones month by month. I had calls with industry contacts in Basel, Hyderabad, Boston, and Suzhou.
And I built what I think is the most comprehensive investment playbook available right now on the pharmaceutical sovereignty trade.
Not a hot-take article. Not a list of seven tickers. A real playbook.