Japan Is Buying AI as Insurance, Not Efficiency

In March, METI stated that Japan intends to build a domestic physical AI sector and capture roughly 30% of the global market by 2040. The commitment behind it is substantial — on the order of $6.3 billion toward AI and robotics under the Takaichi government.

Targets like this are easy to dismiss as industrial-policy theatre. This one deserves more attention, not because the number is credible in itself, but because of what sits underneath it. The driver is not ambition. It is arithmetic.

What is actually moving

Japan's population declined for a fourteenth consecutive year in 2024. Working-age people now make up around 59.6% of the total, a share projected to fall by nearly fifteen million over the next twenty years. The commonly cited figure is a shortfall of about eleven million workers by 2040, with 29% of the population already over 65.

There is no policy lever that reverses this inside the relevant time horizon. Immigration at politically feasible volumes does not close a gap of that size. Birth-rate intervention, even if it worked immediately, produces workers in the 2050s.

So Japan is left with one remaining option, and it has chosen it deliberately.

The starting position is not bad. Japanese manufacturers accounted for roughly 70% of the global industrial robotics market as of 2022. There is deep existing strength in mechatronics and hardware supply chains, and — unusually — broad cultural acceptance of robots in workplaces.

The technology standpoint

What has changed technically is the movement from robots that repeat programmed motions in structured environments to systems that perceive and act in unstructured ones. That is the physical AI transition, and it is the reason Japan's existing industrial robotics dominance does not automatically carry forward. The new category is harder to own than the old one.

Japan's approach is notably pragmatic. Rather than waiting for the technology to mature, systems are being pushed out of laboratories and into warehouses, factories and service roles now — accepting imperfect performance because the alternative is roles going entirely unfilled. That tolerance for imperfection is itself a competitive advantage relative to more regulation-heavy markets.

It is also attracting serious capital. Salesforce Ventures, Toyota's Woven Capital and Global Brain are all backing Japanese physical AI companies, and these are not moonshot allocations — they are bets on companies putting robots into real operations.

The business standpoint — and this is the part that matters commercially

Here is the sentence I would put on the first slide of any physical AI pitch in Japan, paraphrasing a framing I have seen from investors in this space: physical AI is being bought as a continuity tool. The question customers are asking is how to keep factories, warehouses, infrastructure and service operations running with fewer people.

Not how to run them more cheaply. How to run them at all.

A 2024 Reuters/Nikkei survey found labour shortages to be the primary force pushing Japanese firms toward AI adoption. As one investor put it, the driver has shifted from efficiency to industrial survival.

This is a different purchase. An efficiency purchase competes against every other efficiency purchase for a slice of a discretionary budget, and it gets deferred when times are uncertain. A continuity purchase competes against the risk of the operation failing, and it does not get deferred, because the thing it prevents is worse than the economic cycle.

The market standpoint — what this means if you're selling here

Your ROI model is probably the wrong instrument. A payback calculation based on replacing N workers at ¥X per hour assumes those workers exist and could be employed. Frequently they do not and could not. The honest model compares your system against the cost of the line running below capacity, the store closing early, or the maintenance schedule slipping — and those numbers are usually far larger than a labour-cost saving.

Do not lead with headcount reduction. Beyond being commercially weaker, it misreads the relationship. Japanese employers under labour pressure are generally trying to supplement their workforce, not shrink it, and the roles being automated are ones nobody applied for. A pitch framed around cutting staff signals you have not understood the market and, in a consensus-driven organisation, gives every internal sceptic an easy objection.

Reframe the risk conversation. The instinctive sales response to a conservative buyer is to minimise perceived risk in adopting. In a continuity purchase you should also be raising the perceived risk of not adopting — because for once that framing is accurate rather than manipulative. The operation genuinely is on a trajectory that does not resolve itself.

Expect the budget to sit somewhere unexpected. Continuity spending does not always come from the IT or innovation line. It may come from operations, from facilities, from a business-continuity budget, or from a capital programme. Finding the right budget is often harder than finding the right champion, and it is the step most foreign vendors skip.

The larger point

Japan is going to be the most instructive market in the world for physical AI over the next decade, because it is the one where the demand is structural rather than speculative. Other countries with ageing populations — Korea, Germany, Italy — will follow, and they will follow the patterns established here.

For a foreign company, that makes Japan valuable well beyond its own revenue. It is where the category's commercial norms are being set: what customers expect, how deployments are structured, what proof is required. Being present while that happens is worth more than the pipeline alone would justify.

But only if you arrive selling what the market is actually buying. Efficiency is a story about a better quarter. Continuity is a story about whether the business still functions in 2035. Japanese executives are thinking about the second one.