engineering drawings on table

Building a China team

The Mandate Changed. The Hiring Plan Did Not.

Research and design services drew 127.8 per cent more foreign investment in the first quarter of 2026, while overall inbound capital still fell. The centres being opened now are asked to serve global programmes, but most hiring plans still describe a local support function. I have sat in enough of those meetings to recognise the slide.

By SunTzu Recruit | Data as of November 18, 2026

Sources: Ministry of Commerce FDI statistics 2025 and January–February 2026 / Shanghai municipal and Pudong government announcements / China Economic Net and Economic Daily reporting on foreign R&D centres / German Chamber of Commerce China Business Confidence Survey 2026 / publicly announced investment projects by BMW, Covestro and Mercedes-Benz

The building is leased. The equipment list is signed. Someone from headquarters has put one box on the organisation chart marked “R&D director, China” and looked across the table at me as if the rest of the chart were my problem. It usually is, by that point.

I have lost count of the number of these meetings I have sat in, in Shanghai, Suzhou and Shenzhen. The conversation always starts with laboratory space and ends with a question nobody prepared for: who actually goes into the building. What surprises me, even now, is how often the brief has changed while the hiring plan stayed still. The mandate on the slide says the centre will support global programmes. The headcount sheet underneath it describes a local adaptation team, with one manager and a handful of engineers who report to someone eight time zones away. Those are two different organisations wearing the same name.

The numbers explain why this is happening now. Research and design services attracted 127.8 per cent more foreign investment in the first quarter of 2026 even as total inbound capital remained under pressure. In 2025, the research and technical services sector took close to one fifth of all foreign investment in China, a share that has risen for seven consecutive years and now stands at 3.8 times its 2018 level. Just over 14,000 foreign enterprises were registered in that sector, up 27.2 per cent.

By February 2026 Shanghai had certified 1,084 regional headquarters and 647 foreign R&D centres; Beijing had 332 as of March, having added 55 in a single year. A survey of multinational executives put the share planning to invest in “China R&D serving the world” at 42 per cent over the next two years. That phrase would have sounded odd a decade ago. It is now the brief I hear most often.

The brief on the slide

Mercedes-Benz used to describe its China R&D as “in China, for China”. The public wording has shifted to “in China, for the world”. Two characters changed, and with them the entire logic of what a local team is for. Under the old version, the centre translated a global product for a local market. Under the new one, the centre produces something the global network consumes.

BMW opened an IT research centre in Nanjing that sits among its six largest information technology research sites worldwide. Covestro started a thermoplastic polyurethane application development centre in Guangzhou in May 2026, built around a production base in Zhuhai and a regional centre in Taiwan, China. These are not token offices. They are nodes in a global delivery network, and the people inside them are judged by teams in Germany or the United States.

The policy environment has moved in the same direction. A revised catalogue of encouraged industries for foreign investment took effect in February 2026, adding categories in new drug research and digital creative technology. Import duty exemptions for research equipment used by qualifying foreign R&D centres were extended. These measures do not create engineers, and they do not shorten a search. They do, however, make the decision to open a centre easier than the decision to staff it.

That changes the hiring question in a way most requisitions miss. A local adaptation team needs people who know the Chinese market and can work within a defined framework. A global contributor needs people who can argue with that framework, write specifications that travel, and defend a technical decision in a meeting where nobody shares their first language. The first group is easier to find, cheaper to employ and far more common in the local candidate pool.

Local adaptation is a different job

I keep a simple distinction in my own notes. If the deliverable stops at the China border, hire for market knowledge and execution speed. If the deliverable leaves the country, hire for the ability to work inside somebody else’s quality system and still make a case of your own. Very few candidates are strong at both, and a requisition that asks for both without saying which one matters will attract the wrong person.

What headquarters writes

“R&D engineer, China — 5 to 8 years experience, fluent English, able to work independently.”

What the role needs

“Engineer who has delivered a validation package into a global platform, can name the review board that accepted it, and writes change requests that a team in another country can act on.”

One box, four jobs

The single box on the chart is the root of most of the trouble. What gets called an R&D centre is usually four functions sharing a floor: research, engineering and validation, regulatory and quality, and the interface with global programme teams. Collapsing them into one reporting line works while the centre is small. It stops working the moment headquarters expects the centre to own a deliverable.

I watched one Shanghai team lose nine months this way. The centre had been asked to take over a component of a global device programme. Everyone assumed the R&D director would cover the regulatory side because the title sounded broad enough. He had spent his career in process engineering. The first design history file came back with questions the local team could not answer, and each round of clarification travelled via the regional office in Singapore. By the time a regulatory specialist was hired, the programme timeline had already been re-baselined.

What belongs in the local team

  • A research lead who can set a technical direction and defend it to a global review board.
  • Validation and engineering staff who have worked inside the parent company’s quality system, or can learn it quickly.
  • A regulatory and quality specialist whose knowledge transfers across markets — a medical device background does not substitute for a drug registration background.
  • A programme interface who can read headquarters’ process language and translate it into a local plan without losing the requirements.

What should stay with the parent company

  • Final authority over portfolio decisions and the allocation of global development budget.
  • Ownership of the intellectual property strategy, even when the experimental work happens locally.
  • The relationship with global regulators that the centre supports but does not lead.

Getting that split wrong is expensive in a specific way. A local hire asked to make a decision they have no authority to make will escalate everything. An expatriate placed into a role that needs local registration knowledge will spend the first year learning a system their team already knows.

The first three people

When a client asks me who to hire first, my answer tends to disappoint them. It is usually not the director. The director matters, but the director inherits whatever structure exists on day one, and by the time the search closes the team is already six weeks behind. The first hires should be the people who make the centre legible to both sides.

  1. Someone who has worked inside the parent company’s systems, even if only for two years. They know what a change request looks like, which committee signs off, and how long an approval actually takes.
  2. Someone who knows the local engineering pool and can judge a candidate’s real contribution in a team, rather than the title on their business card.
  3. A regulatory or quality specialist with experience in the specific product class. This is the hire that most often gets postponed, and the one whose absence costs the most.

Once those three are in place, the director search gets easier for a reason that has nothing to do with salary. Candidates can see what they would be joining. A director deciding between two offers will take the one where the team already functions, and a centre with a credible technical core will attract people who would never have replied to a cold approach.

The order matters more than the speed. I would rather see a centre open with four well-chosen people and take eight months to reach full headcount than fill fifteen seats in the first quarter and spend the next year replacing half of them. The second pattern is common, and it is visible in the way the centre talks about itself: they describe the org chart rather than the work.

There is a practical reason for this beyond morale. A centre with a technical core can absorb a late arrival; a centre made entirely of late arrivals has nobody to orient them. I have seen a team of twelve spend its first quarter asking headquarters a series of questions that a single experienced systems engineer would have answered in an afternoon. The salary bill for that quarter was substantial, and none of it bought capability.

What the investment numbers hide

There is a temptation to read the foreign investment figures as a story about confidence. The more useful reading is about shape. In 2025, 70,392 foreign enterprises were newly established in China, up 19.1 per cent, while actual utilised investment fell 9.5 per cent. More entities, each smaller. More than 53,000 foreign enterprises are now operating in the country, and 8,000 of them increased their investment last year.

A smaller entity cannot carry the old staffing model. When a regional headquarters funded a generous expatriate package and a local support team, the overhead could be absorbed elsewhere. When a newly registered R&D centre has a budget for twelve people, every seat is contested between the technical work and the coordination work. The centre still has to talk to headquarters, still has to satisfy local registration requirements, and still has to produce something the global network will accept.

The German Chamber’s 2026 survey found 61 per cent of respondents planning to increase investment in China over two years, the highest share since 2023. That is encouraging, and it also means the competition for the same narrow group of engineers will intensify. A centre that cannot explain what makes its work distinctive will lose candidates to a better-funded neighbour in the same park.

The budget that arrives first

Most budget conversations happen before the technical brief is settled. Money is allocated against a headcount template borrowed from another country, then the centre is told what it will deliver. Reversing that order — agreeing the deliverable, then deriving the team — is the single most useful change I can suggest to a client at this stage. It rarely requires more money. It requires someone to say plainly which of the four functions the centre will own.

MandateWhat the team is measured onFirst hire
Local adaptationSpeed of market response, cost, local complianceApplication engineer with market knowledge
Global contributionAcceptance by a global review board, specification qualitySystems engineer with parent-company experience
Mixed mandateBoth, with a stated priorityProgramme interface plus a regulatory specialist

How I check a candidate for this role

The candidates who look best on paper are often the ones who worked in a large multinational with a well-known name. That experience is useful, and it can also be misleading. What matters is whether the person owned an output or supported someone who did. I ask for the artefact, not the title.

  • A technical document they wrote that went through an external review — a validation report, a design history file entry, a change request that was accepted.
  • The name of the review board or committee that accepted it, and what objection had to be answered before acceptance.
  • A clear account of which parts of the work they did personally and which were done by a team they coordinated.
  • For regulatory roles, the specific product class and market they registered in. Registration knowledge does not transfer freely between devices and drugs.

The last point catches people out more often than any other. Shanghai’s foreign R&D base includes a large medical device cluster, and drug registration experience is regularly treated as interchangeable with device registration. It is not. The documentation, the clinical evidence expectations and the review pathways are different enough that a specialist needs a year to cross.

One more thing I look for, though it is hard to put in a shortlist: whether the candidate can describe a technical disagreement they lost. People who have only worked inside a well-oiled system often have no such story. People who have contributed to a global programme almost always do.

What I cannot decide

A recruiter can establish that a candidate wrote a document, held a position and worked on a named programme. We can check employment dates, ask for redacted material and speak to references who saw the work. We can also tell you when a job description is describing two different jobs at once.

We cannot tell you whether the centre should own the deliverable in the first place. That is a portfolio decision, and it belongs to the parent company. We cannot judge the legal force of an IP arrangement or a registration strategy; that belongs to your counsel. What we can do is make sure the person you interview is the person the brief actually calls for, and that the sequence of hires matches the date the centre is meant to start producing.

I have come to believe that the first three months of a China R&D centre decide more than the following three years. Not because the early hires are more talented, but because they set the boundaries of what the local team believes it is allowed to do. A team that starts by owning a small deliverable will ask for the next one. A team that starts by forwarding questions to Singapore will keep forwarding them.

If you are opening a centre in China and the chart still has one box on it, that is a workable starting point. It is worth asking, before the first interview, what the centre will hand to the global network in its first year. The answer tells you who to hire, and in what order.

SunTzu Recruit works with multinational companies on China market entry and local team build-out, from the first entity decision through to the technical hires that make a centre credible to headquarters. We write about what we see in those searches, and we keep the boundary between what we can verify and what you must decide.

More on hiring in China: https://www.suntzurecruit.com/news/

Categories:

Tags:

Comments are closed