In April, a Taipei company called AItewan won American clearance for software that maps brain tumors on an MRI to guide radiation planning. It already held Taiwanese clearance, a double almost no rival carries. No molecule. No decade-long trial. No nine-figure deal. A device, cleared in months.
When people say "Taiwan biotech," they mean drugs: the out-licensing deals, the clinical assets, the next molecule. I have spent two essays in that world myself, on the bio shield and why Taiwan is the fastest first market in Asia. But if the first wave of US-Taiwan wins lands anywhere, my bet is it lands in devices, and almost no one is writing about it.
I am in Taipei this week on a life-sciences trade mission hosted by TAITRA and Mosaic Venture Lab, representing the United States, and devices are an explicit part of the brief. Separately, I am also having meetings with BioPalace's investors and clients. Medical Taiwan, the island's main medtech show, runs in parallel. So this is the natural week to make the case the drug-centric coverage keeps missing.
Why drugs get the ink
Drugs get the headlines because drugs are the romance of the industry. A new molecule is a ten-year story with a binary ending, and the deals are big enough to print. Devices are unglamorous by comparison: incremental, engineering-led, rarely worth a press release on their own. So the coverage follows the molecules and skips the machines.
The numbers run the other way. Taiwan's medical-device sector produced about NT$210 billion, roughly $6.5 billion, in 2024, a larger slice of the country's biomedical output than pharmaceuticals. Device exports came to $6.33 billion in 2023 and an estimated $6.73 billion in 2024, and the single biggest buyer is the United States. The sector exports more, and it moves faster, because a device clears regulators in months where a drug takes years. If you are looking for where a US-Taiwan deal closes first, the structural odds favor the machines.
The manufacturing fit
The reason is the same industrial base that built the chips. Taiwan's integrated-circuit, printed-circuit-board and sensor supply chains feed straight into device manufacturing, and ITRI, the state institute that incubated TSMC and UMC, runs a biomedical-device arm doing the same kind of work for medical hardware. ICT firms are moving into smart devices on purpose, and the government subsidizes up to half the cost of cross-industry projects that pull electronics into medicine. When Canada wanted a partner for an AI-in-medtech delegation this April, it picked ITRI's startup platform, which now spans more than 120 biomedical and medtech firms.
The edge is real but narrow, and worth pinning down before anyone inflates it. Taiwan is strong in mid-range hardware, homecare products, diagnostics, and contract manufacturing, the OEM and ODM work that its electronics firms already know how to do. It is not, today, a maker of high-end surgical robots or advanced imaging systems. More than 70 percent of the advanced devices used in Taiwan's own hospitals are imported, the largest share from the United States. So the picture is not Taiwan-beats-Germany. It is something more useful.
Trade that runs both ways
That import dependence is what makes the trade real rather than rhetorical. It runs both ways, and both ways are moving.
Going out: Taiwanese diagnostics, point-of-care testing, and a young crop of AI-software-as-a-medical-device firms, built on the country's long ICT strength and its decades of national health-insurance data. Going in: US high-end equipment, into a market that needs it and is slowly paying for it. Intuitive Surgical has sold its da Vinci robots in Taiwan since 2004, and the national insurer has been widening coverage: 17 procedure types from March 2023, then 46 more in September 2024, reaching about 8,400 patients a year. Each expansion enlarges the market for an American device maker. Taiwan is the buyer there, and a growing one.
A pure-export story would be fragile. Trade that runs both ways, where Taiwan sells what it makes well and buys what it does not, is the kind that lasts.
What clears, and what does not
Devices also clear the regulatory gates faster than drugs, in both directions, which is the practical reason the first deals may land here.
Into Taiwan: a mid-risk Class II device averages about 140 days at the TFDA, a Class III or genuinely new device about 220, and US manufacturers get an abbreviated quality-system route. The TFDA joined the international medtech regulators' forum in 2023 and stood up a dedicated office for AI and software devices, revising its machine-learning guidance again last August. Into the US: most Taiwanese devices enter through the FDA's 510(k) notification, where clearance typically runs four to nine months against a 90-day target. AItewan, the Taipei AI firm, won that 510(k) clearance for its brain-tumor mapping software in April, on top of its Taiwan approval. HippoScreen cleared the FDA in 2020 and the TFDA in 2021. These are real wins, but not yet a trend: a Taiwanese device clearing the FDA is still an event, not a routine, and there is no Korea-scale wave of submissions behind it.
Two cautions belong in any honest version of this. First, the February trade agreement names devices directly, committing Taiwan to accept FDA marketing authorizations for US-made devices without extra requirements. That would be a real change. It is also not yet law: it waits on ratification by Taiwan's legislature, so today it is a commitment, not a clearance. Second, reimbursement is the choke point. Taiwan's insurance budget is tight, foreign device makers have flagged coverage limits for years, and the robotic-surgery expansions stand out precisely because coverage usually moves so slowly. The door is opening. It is not open.
Where the near-term deals sit
Put the pieces together and you can see where a first wave actually closes. Four lanes look live.
Diagnostics and point-of-care, where Taiwan already exports at scale and the home-grown health data sharpens AI-driven products. AI imaging and software devices, the fastest-moving segment, where a single product can clear both regulators in one cycle, as that April clearance showed. Contract manufacturing, where US firms can use Taiwanese electronics-grade production without building it themselves. And US high-end equipment moving the other way, as the insurer keeps widening what it will pay for. The orthopedics tie-up between California's THINK Surgical and Taiwan's United Orthopedic, running since 2019, is the older template: US robotics, Taiwanese manufacturing and distribution, one market designated a center of excellence.
What to do now
So if you build or buy devices, three moves are worth making this quarter. If you are a US device maker, price in the trade agreement now while it is still a commitment, so you are positioned the day it ratifies rather than scrambling after. If you are a Taiwanese founder, treat an FDA 510(k) as the asset that doubles your company's value, the way AItewan did, because dual clearance is what makes you a partner rather than a vendor. And if you are an investor, this is the medtech story hiding behind the drug story, cheaper to underwrite because the regulatory clock is measured in months.
I came to test whether the device story is more than a slide in a trade brochure. On the evidence so far, it is the part of the US-Taiwan relationship doing the most work with the least notice. The molecules get the headlines. The machines may get there first.