An embodied AI startup outside China's largest technology hubs has spent a year recruiting for a core technical role. More than 30 interviews have produced no appointment. The founder begins to price an office in Shenzhen or Shanghai: perhaps moving R&D, or establishing a small recruiting base there, would solve the problem.
There is a plausible geographical argument. Data attributed to the Liepin Big Data Research Institute puts Shanghai at 18.57% of embodied AI vacancies, Shenzhen at 17.56% and Beijing at 10.95%—almost half between them. Across AI more broadly, Beijing, Shanghai, Shenzhen and Hangzhou account for nearly 74% of demand. Maimai's figures put AI at 30.17% of newly advertised roles in Beijing, or roughly three in ten.
The talent pool is concentrated. That does not establish that geography caused this particular search to fail.
Separate a location problem from a hiring problem
Review the previous year's interviews before looking at office rents. Record why candidates did not join, rather than simply why the company rejected them.
If 20 of 30 candidates lacked the required technical direction or capability, the brief may be the problem. A job description demanding an implausibly broad combination of skills will be difficult to fill in any city.
If 15 of those 30 were technically suitable but declined because they would not relocate, geography is a more credible explanation. A China.com Henan report citing Liepin identifies precisely this challenge in Robotics: a small local supply, the need to attract people from leading cities, and limited willingness to move without a stronger employer proposition and compensation incentive.
Pay structure, company stage and interview quality deserve the same scrutiny. Moving the search to Shenzhen will not repair an unclear mandate.
Four options, with different costs
Recruit elsewhere and ask the person to relocate
This involves the least organisational change, but it can be the least effective option. A senior candidate is being asked to change both employer and city. Family commitments, housing and children's schooling make that a larger decision.
A Zhaopin survey reports rising concern about job stability and more than half of respondents willing to accept a modest pay cut for a less demanding job. A relocation proposition therefore needs a convincing combination of compensation, equity and technical opportunity, not just an interesting title.
Allow remote work
Code, models and data can be shared remotely. Embodied AI, however, ultimately has to work on a physical robot. A role involving frequent hardware discussions, laboratory work and live debugging is harder to distribute than a largely software-based assignment.
A 2026 Wenjuan.com business survey identifies communication, team cohesion, project management and progress tracking as challenges for dispersed teams. A small startup needs to decide which work can genuinely be remote and which requires people beside the machine.
Establish a satellite R&D centre
This is the most expensive option, although it may create a more durable route to talent. The Liepin figures cited report year-on-year embodied AI vacancy growth above 100% in Wuhan, Hefei and Suzhou. These cities may offer expanding talent pools with a different competitive environment from the largest hubs.
Local support schemes include RMB 300,000 in construction and operating support for recognised out-of-town coRPOrate R&D centres under a Quanzhou policy, and Shishi schemes with RMB 150,000 and RMB 100,000 support amounts. Eligibility depends on the location and terms of the relevant programme.
Two sites also mean two teams, more coordination and a greater risk of divergent working cultures. A Series A or Series B company may not yet have the management capacity to make that arrangement productive.
Develop local talent
If relocation, remote work and a second centre are impractical, reconsider how much immediately usable experience the role really needs. Liepin data shows 87.80% growth in robotics vacancies requiring a master's degree, the fastest growth among the qualification groups discussed.
The proposed alternative is a local candidate with a strong technical specialism and the ability to learn, supported by deliberate internal development. For some startups, that is more realistic than continuing to pursue an all-round senior expert who does not want to move.
Budget beyond the office lease
Cross-city interviewing can take at least two days per trip. Repeated dozens of times over a year, that removes substantial founder or technical-leader time from the product.
Distributed work also requires decisions about assessment, communication and common working practices. A 2026 Feishu Project study associates AI-assisted R&D project management with around 40% less resource waste, but established collaboration processes are a prerequisite. Startups may not have them yet.
Compensation is another cost. Maimai's cited average monthly pay for embodied AI roles is RMB 61,625. A company seeking candidates from a high-paying hub must account for their alternatives, whether through relocation incentives, remote working or a local office.
Make the diagnosis before choosing the city
Ask three questions: why did the suitable candidates decline; is the compensation and technical proposition strong enough to support a move; and can the company manage a distributed R&D team at its present stage?
UB hunter begins with the work the position must deliver—an algorithmic advance, engineering implementation, Team Building or supply-chain integration. That makes it possible to compare local development, a wider search, remote work and a satellite centre against the actual requirement.