Japan's Robot Makers Are Paying a China Discount to Skip It

China builds industrial robots roughly 20 percent cheaper than anywhere else on earth, by Morgan Stanley's own estimate. Japan's two largest robot makers are betting on the United States anyway.

Fanuc and Yaskawa Electric, the industrial automation giants whose controllers and robot arms already run a meaningful share of the world's factory floors, are each expanding their American manufacturing and partnership footprint even as China's cost advantage in robotics supply chains remains, by most independent measures, larger than ever. Fanuc America is building a new 840,000-square-foot plant and logistics center in Michigan for roughly US$90 million, expected to open in 2027 and add about 225 jobs. Yaskawa's most recent quarterly results show revenue climbing on the back of semiconductor- and data-center-related demand tied directly to the artificial intelligence buildout, even as its core robot segment's profitability fell sharply. Kawasaki Heavy Industries has taken a parallel path, partnering with Nvidia to open a physical-AI robotics hub in the United States. The pattern across all three is the same: proximity to AI-driven capital spending is starting to outweigh the raw cost advantage of building in China.

The Cost Math That Should Point the Other Way

Morgan Stanley's research is specific about how large that China cost gap actually is. The bank estimates China's localized supply chain lowers manufacturing costs by roughly 20 percent compared with building the same equipment elsewhere, with individual robotics components running about 16 percent cheaper domestically. The starkest illustration involves humanoid robotics rather than industrial arms: Morgan Stanley calculated that replicating the bill of materials for a Tesla Optimus Gen 2-class humanoid outside China would nearly triple the total cost, with actuator costs rising from around US$22,000 to US$58,000, chip and software costs climbing from roughly US$3,000 to US$7,000, and the full bill of materials increasing from about US$46,000 to US$131,000. Those are not marginal differences a manufacturer absorbs quietly. They are the kind of gap that, on paper, should keep production anchored in China for as long as competitive pricing matters.

Fanuc and Yaskawa are choosing to pay that premium anyway, at least for a growing share of their US-bound capacity, and the reason has less to do with the cost of building a robot than with where the customers placing the biggest orders are actually located.

Data Centers Are Becoming Yaskawa's Best Customer, Not Its Robots

Yaskawa's most recent quarterly disclosures make the shift unusually explicit. Revenue rose 10.6 percent year on year to approximately ¥138,982 million, or roughly US$927 million at current exchange rates, driven largely by demand tied to semiconductor and data-center construction rather than the company's traditional robot segment. Operating profit actually fell 19.2 percent to roughly ¥8,486 million, or about US$57 million, and the robot segment specifically saw operating profit collapse 82.3 percent even as demand for general industrial robots held up globally. For full-year fiscal 2026, Yaskawa is guiding to revenue of roughly ¥580.0 billion, about US$3.9 billion, and operating profit of ¥60.0 billion, close to US$400 million, banking on continued strength in AI-adjacent infrastructure spending to offset softness in the robot arms business that built the company's reputation.

That divergence matters for how to read Fanuc and Yaskawa's American expansion. This is not primarily a story about robot manufacturers chasing robot customers across the Pacific. It is a story about industrial equipment makers repositioning their entire businesses around the physical infrastructure boom that AI investment is creating, of which robotics is only one product line among several, including the cooling systems, substrate-mounting equipment and precision components that data centers consume at scale. Being close to where hyperscalers and chipmakers are breaking ground now carries more strategic weight for a company like Yaskawa than being close to the cheapest possible robot component supply chain.

Kawasaki and ABB Show Two Different Ways to Read the Same Signal

Kawasaki Heavy Industries reached a similar conclusion through a different structure, partnering with Nvidia, alongside Microsoft, Analog Devices and Fujitsu, to build a US physical-AI robotics hub focused initially on medical and mobility applications. The collaboration uses Nvidia's simulation tools to develop Kawasaki's four-legged personal mobility robot Corleo, a deliberately different bet from Fanuc's factory-floor expansion but one built on the same underlying premise: that being embedded inside America's AI infrastructure ecosystem, close to the compute providers and platform companies setting the technical standards, matters more right now than defending a cost position built around Chinese manufacturing.

ABB offers the clearest counterexample, and it complicates any simple story about Japanese robotics makers uniformly abandoning China-based cost advantages for American proximity. Rather than expanding its own robotics manufacturing footprint, ABB agreed in October 2025 to sell its entire robotics division to SoftBank Group for an enterprise value of US$5.375 billion, a deal expected to close in the middle to late part of 2026 pending regulatory clearance across the European Union, China and the United States. ABB's exit is not evidence that robotics manufacturing has stopped being a viable business. It is evidence that a European industrial conglomerate concluded it could not compete effectively in the current environment without either the AI-infrastructure proximity Fanuc and Yaskawa are now chasing or the capital depth to build it, and chose to sell rather than make that bet itself. SoftBank, which has spent 2026 assembling a broad portfolio of AI and robotics assets, is betting it can supply both.

What This Means for the Buyer Signing the Purchase Order

For a procurement team evaluating industrial robot vendors today, this repositioning is not an abstract strategic story. It changes where a buyer's fleet actually gets built, how quickly a warranty claim gets resolved and how exposed a supply contract is to the next round of export-control tightening between Washington and Beijing. A manufacturer sourcing robot arms for a new US facility now has a genuine choice between Chinese-built units carrying a real cost advantage and Japanese-built units increasingly assembled closer to home, each carrying a different risk profile. The Chinese option remains cheaper on a unit-cost basis by Morgan Stanley's own numbers, but it also carries more exposure to tariff shifts, export licensing delays and the kind of geopolitical friction that has already disrupted component shipments for robotics buyers once this year. The domestically assembled Fanuc or Yaskawa option costs more upfront but reduces that exposure meaningfully, a trade a growing number of American manufacturers appear willing to make even at a premium.

That calculation looks different again for a buyer building out AI infrastructure rather than a traditional factory floor. A data center operator or chipmaker evaluating automation vendors for a new US facility increasingly wants a supplier who is physically present, has local service technicians and understands the specific reliability requirements of cooling and substrate-handling equipment running continuously at scale. Yaskawa's own revenue mix, now leaning more heavily on exactly that kind of AI-infrastructure equipment than on traditional robot arms, is itself evidence that the company sees this buyer segment as large enough to reorganize around. A robotics vendor with a plant and service network already inside the US market has a structural advantage in winning that business that a purely export-based Chinese supplier, however competitive on price, currently cannot match.

Kawasaki's Bet Looks Different From Fanuc's, and Buyers Should Notice

Kawasaki's move deserves to be separated from Fanuc and Yaskawa's, because the underlying logic is not identical even though the geography is. Fanuc's Michigan plant and Yaskawa's data-center-driven revenue shift are both about serving existing product categories, industrial arms and automation components, closer to where American demand is concentrated. Kawasaki's Nvidia partnership is a bet on an entirely new product category: physical AI systems built around simulation-trained control, starting with the four-legged personal mobility robot Corleo rather than a traditional industrial arm. For a buyer, that distinction matters because it signals which of these companies is defending existing market share close to home and which is trying to establish a foothold in a market segment that does not fully exist yet. Fanuc and Yaskawa's moves are closer to a defensive repositioning around proven revenue. Kawasaki's is closer to a venture bet, run inside a century-old industrial conglomerate, on where physical AI product categories go next.

Why This Is a Repositioning, Not a Retreat From China

None of this amounts to Japan's robot makers abandoning China. Fanuc, Yaskawa and their peers continue to operate substantial manufacturing and sales operations across the Chinese market, and the cost advantages Morgan Stanley documented are not disappearing. What is changing is the marginal dollar of new capital expenditure. When a Japanese industrial robot maker decides where to put its next US$90 million plant or its next strategic partnership, the calculation increasingly weighs proximity to AI-driven demand, and the political and export-control stability of serving American customers from American soil, more heavily than it weighs shaving another few percentage points off unit production cost. That is a meaningfully different strategic posture than the pure cost-arbitrage logic that shaped industrial robotics supply chains for the previous two decades, and it is one other equipment makers watching from Europe and South Korea will have to answer for their own capital plans regardless of what conclusion they reach.

The Morgan Stanley numbers on Chinese manufacturing costs are not wrong. They are simply no longer the only number that determines where an industrial robotics company builds its next factory, and 2026 is the year that stopped being a hypothetical.

This analysis synthesizes company statements and public market activity as of the publication date and should not be read as investment, financial, or professional advice; it is provided for general information purposes only.

Hero image credit: FANUC America.