Author: Peng Zhao
In recent intensive exchanges with multiple enterprises, the anxieties and reflections from all parties ultimately point to the same proposition: Where exactly should the value realization of Physical AI be completed?
Piecing together these frontline insights, I have reached three judgments that are completely different from the inertia of hardware going global over the past decade or so:
First, spatial mismatch: The domestic market is the training ground, while the overseas market is the real business arena;
Second, anchor shift: Bidding farewell to cost-effectiveness based on material costs, and shifting towards an intelligence premium based on overseas labor;
Third, interest concession: Shifting from merely exporting products overseas to deeply binding with local entities and conceding interests and equity.
This article will analyze these three judgments.
Training Ground and Business Arena
Let us first look at a somewhat fragmented industrial phenomenon.
As a highly anticipated branch of Physical AI, Embodied AI is experiencing unprecedented capital enthusiasm in China, with intensive financing following one after another and dynamic demonstrations emerging endlessly at various exhibitions. However, if we turn our attention to the real frontline of manufacturing, the procurement attitude of factory decision-makers is exceptionally calm.
An entrepreneur calculated a set of figures for me: the hourly wage of production line workers in second-tier cities is about CNY 18 to 22; factories purchasing automated equipment generally require an 8-month payback period, and almost no one accepts more than 24 months. Roughly calculating based on his figures, for a CNY 300,000 robot to pay for itself within 24 months, it needs to replace CNY 12,500 worth of labor per month. At an hourly wage of CNY 20, this translates to 625 working hours, equivalent to the single-shift workload of three and a half workers, not including operation, maintenance, and electricity costs.
His conclusion is straightforward: if Embodied AI only targets the domestic market, it will be difficult to achieve positive gross profit within five years, unless the revenue comes mainly from the government.
Placing the same robot in Japan, Europe, or the United States changes the ledger entirely. Local labor costs are several times higher than those in China, and recruitment difficulties are the norm. For a robot that can work stably, the payback period will be significantly shortened.
As a training ground, the domestic market is almost irreplaceable. It features dense manufacturing scenarios, a complete supply chain, and a large number of engineers capable of rapid iteration. A problem at a workstation discovered today can be modified and retested by next week.
For robots to transition from being demonstrable to deliverable, they need to be repeatedly tested and refined in real-world scenarios, and such opportunities are most easily obtained in China.
In a batch of early-stage projects I have recently observed, there is a telling phenomenon: the companies that generate revenue first are often those that help robots "go to school," focusing on pilot validation, data collection, testing and evaluation, and core sensors, while the companies that actually build the robots fall behind.
The logic is simple: robot companies have funds and are eager to implement their solutions, so they are willing to pay to make the robots learn how to work. I once referred to this type of work in an article as apprenticeship hours: robots work for manufacturers and laboratories, producing data and capabilities, and the party paying is the one that wants them to learn how to work. What the domestic market buys the most of today is precisely apprenticeship hours.
Thus, a new division of labor is taking shape: training at home and earning money abroad.
The path for the previous generation of hardware going global was validation in the domestic market followed by replication and expansion overseas; for Embodied AI, the overseas market might be the "only" profitable one, and the concept of a "second market outside China" needs to be updated.
This means that for robot body manufacturers, going global has shifted from an option after growing up to a prerequisite written into the company's design from day one. I have already seen entrepreneurs who barely do any domestic business.
Cost-Effectiveness and High Premium
With the overseas market established as the real business arena, the core proposition that follows is pricing. After stepping out of the country, what exactly should be the price of our products?
Over the past two decades, the cognitive label of domestic hardware in overseas markets has been highly solidified: focusing on "cost-effectiveness." To put it bluntly, it means using similar product quality to exchange market share by lowering prices.
This label is loosening. A founder who has been fighting in overseas markets for years has an observation: When talking about the solid quality of domestic hardware overseas, few people believe it; but when saying it is smarter, almost no one objects.
Electric vehicles, robot vacuums, drones... have made intelligence a prominent label for domestic hardware for the first time, and AI will continue to amplify this label.
The label has already changed, yet the pricing habits of many founders remain stuck in the previous era.
He gave an example of a lawn mower. This is a mature category where mainstream products from international brands sell for about USD 900. The typical practice of the previous generation of domestic enterprises was to produce products of similar quality and sell them for USD 400.
They did the opposite by introducing LiDAR, which is already mature in automobiles, into lawn mowers. This solved the boundary-less problem that has plagued the industry for years, meaning there is no need to pre-burrow boundary lines under the lawn, and then they priced it at more than twice that of international brands. According to him, the product sells very well.
Behind this is a flywheel: only when the product is innovative enough can it be priced high; only with a high price can there be sufficient gross profit to roll out global channels; and scale and profit then feed back into the next round of R&D.
The starting point of the flywheel is precisely the step founders are least accustomed to: daring to tag the highest price globally.
We are too accustomed to exchanging diligence for market share. First, we relied on the supply chain dividend, and later on the engineer dividend. But now engineers are also becoming more expensive, and market share gained solely through cheapness cannot sustain continuous innovation.
Whether it is the smart lawn mower on the lawn or Embodied AI on the production line, essentially, they are all Physical AI that embeds AI into hardware to work in the real world.
The pricing reference for a robot overseas should be the local wage it replaces. Pricing the overseas market using domestic cost logic is equivalent to giving away the future of sustainable development.
I once analyzed in a previous article that the valuation unit for Embodied AI should shift from the number of units shipped to working hours on duty, and the same logic applies to pricing: when pricing by unit, the anchor is materials; when pricing by working hours, the anchor is wages.
In my view, switching the anchor and logic of pricing is the kind of innovation domestic enterprises dare to do the least.
High Efficiency and Safety Cushion
After switching the pricing anchor, another deeper challenge emerges: how to make the overseas market accept the commercial success of the enterprise in the long term, so that this high gross profit can be continuously realized?
There is a story that left a deep impression on me. An investment promotion official in Southeast Asia told visiting entrepreneurs: "Your engineering team built the longest bridge in our local area for us. The quality is good and the speed is fast, but from design to construction, down to every single nail, everything used is your own. Even the lunch boxes for the workers' meals were transported from China. The bridge is built, but nothing is left for our local area." It is precisely for this reason that countries are now tightening their investment promotion policies, no longer welcoming this kind of "all-inclusive" Chinese projects, and instead mandating binding with local interests.
The heartbreaking part of this story is that these enterprises did not do anything wrong. Efficiency first is precisely the force that has pushed Chinese manufacturing to where it is today, but overseas, it may become the biggest risk.
The reason is not complicated. In 2025, China's trade surplus in goods has exceeded USD 1 trillion. Many countries want to buy Chinese goods, but they are increasingly finding nothing to sell back.
When an outsider brings only products and leaves no employment or industry, the more economically successful it is, the more politically fragile it becomes. Elected governments have to answer to voters, and voters care about jobs, with price and efficiency ranking lower.
A founder deploying in Europe offered another solution. He found a local century-old factory to cooperate with for production, used their sub-brand, helped them hire new workers, and borrowed their ready-made service network.
His summary is very incisive: In the face of increasingly high localization thresholds, proving one's innocence is always passive; a better position is to have local partners willing to stand up and speak on your behalf.
Another entrepreneur thought even more thoroughly. He has mentally prepared himself that when entering Europe and the United States for scaled business in the future, he may need to establish a joint venture with a well-connected local enterprise, with the local party holding the controlling stake. He said this is very similar to the arrangement when foreign car companies entered China back in the day, where they could only form joint ventures with local enterprises, and the foreign equity ratio could not exceed 50%. If one day the other party's shareholding ratio can gradually decrease, it would be a win for him.
This makes me realize that the endgame for the new generation of enterprises going global may be shifting from capturing markets to designing profit distribution: which profits are left for local partners, which jobs are left for local employees, which links are handed over to local suppliers... Turning a portion of profits into wages and shares for locals is a more enduring protection than any tariff negotiation.
The core of the previous generation going global was the one-way export of products; this generation going global must learn to concede equity and form a community of interests with locals.
Conclusion
To summarize, the three judgments above are actually three progressive stages of the same commercialization path: using the domestic market as a training ground to solve the engineering problem of "whether the product can be delivered"; using the overseas market as a business arena to solve the profitability problem of "how to monetize the technology"; and using equity and profits to build a safety cushion to solve the survival problem of "whether the business model can be sustained".
Having done industrial research for over a decade, I have seen too many companies with far-leading technologies ultimately fall due to mismatched commercial structures.
In this wave of Physical AI, Chinese enterprises hold excellent cards: dense manufacturing scenarios, a globally recognized "intelligence" label, and ultimate engineering capabilities. But having good cards does not equal victory in the endgame.
The winning move for the next stage may lie in how entrepreneurs answer three more essential business questions: where to find the real payers, whether they dare to tag the highest price globally, and whether they are willing to concede interests to coexist with locals.