Doing Business in China as a Western Tech Founder: What the Guides Don't Tell You

Here is the answer up front: in 2026, the legal barriers to doing business in China are the lowest they have been in decades – and they are not what will stop you. The Foreign Investment Negative List is down to 29 items. Manufacturing is fully open to foreign ownership. A wholly foreign-owned company takes 4–8 weeks and roughly $13k–25k to set up. What will stop you is something no regulator publishes: you cannot enter this market randomly. You have to know people. China runs on relationships and intermediaries in a way that no amount of Western go-to-market playbook prepares you for, and the founders who fail here are almost never blocked by law – they are blocked by the relationship ceiling they never saw coming.

This essay is written from the ground. I am getting an MBA at Peking University's Guanghua School of Management. I founded a student club that grew to 4,000+ members across Peking University and Tsinghua – the two universities at the center of China's tech elite. Before that, I ran recruiting for San Francisco companies hiring Chinese engineers, sitting directly on the seam between the two tech cultures. Now I am entering robotics, which in practice means entering China's hardware supply chain. What follows is what I have observed on the ground, combined with the regulatory and structural facts every Western founder should have in hand – drawn from a research corpus we maintain at Game Changer Labs precisely because clients keep asking us how China actually works.

Why is China so different from the US market?

Start with the observation that frames everything else: China is simply a different market than the US in terms of culture. Business does not work the same way, and the mistake Western founders make is assuming the difference is cosmetic – that under the different etiquette sits the same transactional machine they know from home. It does not. The machine itself is different. In the US, the default unit of business is the transaction: two strangers meet through a cold email, evaluate each other through diligence documents, and close through contracts. In China, the default unit of business is the relationship: the deal is downstream of the trust, and the trust is built through people, time, and presence.

One difference surprises Westerners in the opposite direction: people here are more straightforward than the stereotype suggests. The popular image of Chinese business as endlessly indirect does not match what I see day to day – once you are inside a relationship, communication is often blunter and faster than the hedged, liability-conscious style of American corporate dealings. The indirection foreigners experience is mostly what the outside of a relationship-gated market feels like. Inside the gate, people tell you what they think. The hard part is getting inside the gate.

And that is the honest headline: it is genuinely difficult to enter this market "randomly." You cannot simply show up with a product, run outbound, and expect the market to respond the way Austin or Berlin would. You have to know people. Everything else in this essay – the entity structures, the tax rates, the payment rails – is downstream of that single fact.

What is the relationship ceiling – and why is guanxi structural?

Western founders hit a ceiling in China without relationships. I have watched it happen: a company enters with a strong product, gets polite first meetings, maybe closes a pilot – and then progress simply stops at a level far below what the product deserves. Nothing is wrong on paper. The ceiling is invisible because it is made of everything that does not happen: the introduction that is never made, the procurement process you never hear about, the supplier terms that are somehow always better for someone else.

The Western literature files this under guanxi and usually gets it wrong in one of two directions – either romanticizing it as ancient cultural mystery or dismissing it as a euphemism for corruption. Both readings miss that relationship capital in China is structural, and rationally so. In a market of this scale, with regional variation this wide and information asymmetry this deep, the network is the diligence mechanism. A counterparty vouched for by someone you trust has posted a bond: their sponsor's reputation. A stranger has posted nothing. Trusting the network over the cold inbound is not sentimentality – it is the economically correct move, and everyone here makes it.

The research corpus bears this out in concrete terms: government procurement contracts, deals with large state-owned enterprises, and certain regulatory permits are effectively gated behind relationships that a foreign company operating alone cannot build quickly. And relationship capital is priced like the asset it is – a Chinese partner who contributes connections, licenses, or land rather than cash will expect an equity premium or annual management fees in return. When relationships appear on the cap table with a price tag, you should stop thinking of them as culture and start modeling them as infrastructure.

Why is China a middle-man culture?

Here is the single most useful thing I can tell a Western founder, and it is the observation I find myself repeating most often: China is a middle-man culture – middlemen make everything easier. Much easier. Introductions, deals, hiring, apartments, suppliers, government paperwork: the default path through almost any process runs through a person whose role is to connect the two sides and stand behind the connection.

The Western instinct is to treat intermediaries as friction – a toll booth to be disrupted, a margin to be disintermediated. Apply that instinct in China and you will burn months discovering why the toll booth exists. In a relationship-gated market, the intermediary is not overhead on the transaction; the intermediary is the interface to the market. If you want a software metaphor: guanxi is the protocol, and middlemen are the API. You can try to speak raw protocol as an outsider, but every call will fail authentication.

What a good intermediary actually does for a foreign founder:

• Transfers trust. Their introduction lends you their reputation. The meeting you get through them is a different meeting than the one you get through a cold email – not a warmer version of it, a categorically different one.
• Filters counterparties. They know which factory actually has the capacity it claims, which distributor pays on time, which "well-connected" person is actually connected. That knowledge is not available at any price on the open market.
• Translates intent, not just language. They tell you what the other side meant, what a "maybe" was, and when a deal died three meetings before you noticed. They also keep both sides from losing face during hard negotiation – which, per the research on China dispute practice, matters enough that confrontational Western tactics measurably backfire here.
• Compresses time. The clearest economic argument: what a well-connected intermediary does in weeks, a direct approach does in months or not at all. Their fee is almost always cheaper than the burn rate of the delay they remove.

One caution from the research corpus, because the failure mode is as structural as the benefit: do not let a single local hire become your only intermediary. Companies that route every supplier, distributor, and government relationship through one employee wake up to find that the network belongs to the person, not the company. Spread relationship ownership across multiple people from day one – the same way you would never let one engineer hold the only keys to production.

Do you actually need a Chinese partner in 2026?

Legally? For most tech founders, no – and this is where the facts have outrun the folklore. Foreign investment access in China is governed by the Foreign Investment Negative List: if your sector is on it, ownership is restricted or prohibited; if it is not, you can own 100% of your Chinese entity. According to the Beijing municipal government's coverage of the NDRC/MOFCOM decree, 2024, the 2024 revision cut the list from 31 to 29 items effective November 1, 2024, and removed the last remaining restrictions on foreign investment in manufacturing. For a robotics or hardware founder, that sentence is the whole ballgame: the supply-chain capital of the world is, on paper, fully open to you.

But notice the two different questions hiding inside "do I need a partner?" The legal question – can I own my company outright? – is now usually yes. The practical question – can I operate here without Chinese relationships? – is still emphatically no. The resolution is the distinction this essay keeps returning to: you need an intermediary, not a partner. A partner takes equity and shares control forever. An intermediary takes a fee and opens doors. Confusing the two is how founders end up in joint ventures they did not need, giving up ownership to acquire relationship capital they could have rented.

WFOE vs JV vs Rep Office: which entity should you set up?

When the relationships are in place and it is time to formalize, China gives you three main vehicles. The market has already voted: the WFOE – wholly foreign-owned enterprise – is the vehicle for the overwhelming majority of foreign-invested companies, and for good reason.

Three notes on the table. First, the WFOE's registered capital: there has been no statutory minimum since 2014, but under Article 47 of the 2024 Company Law, whatever capital you commit must be fully paid in within five years – so the old habit of impressing regulators with a large committed number is now a real liability. Typical commitments for a tech WFOE run $50k–150k; commit what you will actually fund. Second, the JV premium is not just the setup cost – budget two to three times the legal and advisory spend of a WFOE, plus ongoing governance overhead, plus compensation for whatever the partner contributed. And third, the Rep Office is the trap that keeps catching first-timers: because it cannot invoice, it earns nothing, yet under the 2010 rules in State Council Order 584 it is taxed on a deemed profit calculated as a markup on its own expenses. You pay tax on revenue you are structurally forbidden from earning – the cost profile of a small WFOE with none of the capability.

How do payments and banking actually work for a foreigner?

This is the section where reality has improved fastest and reputation has lagged furthest. The horror stories you have read about foreigners locked out of China's cashless economy are largely stories from before 2023. On the ground today: banking as a foreigner is fairly simple now. Alipay isn't bad for foreigners anymore; WeChat Pay is a bit worse. That ranking – Alipay first, WeChat Pay second – is consistent with my daily experience and worth taking literally when you set up your own phone.

The structural reason is verifiable public fact. In July 2023, ahead of the Hangzhou Asian Games, Ant Group overhauled Alipay's international onboarding: according to the South China Morning Post, 2023, overseas users can link Visa, Mastercard, Diners Club, Discover, and JCB cards directly to Alipay and pay at tens of millions of Chinese merchants – no Chinese bank account required to get started. For a founder making exploratory trips, that single change removed the most annoying friction of pre-2023 China travel: you land, you bind your home credit card, and the QR-code economy opens.

Two practical layers sit above that. First, once you have a resident visa and a local bank account, bind the local account too – some smaller merchants, peer transfers, and deposits behave better against domestic rails, and WeChat Pay in particular becomes more usable once it is fed by a Chinese account. Second, keep personal payments and company money strictly separate from day one: company revenue belongs in the WFOE's account, invoiced properly through the fapiao system, because the repatriation route described above – the one that gets profits home – only works for money that entered the company's books correctly. The capital controls that make China's currency system famous mostly bind at the company and remittance level, not at the coffee-and-taxi level, and conflating the two levels is how founders create their own horror stories.

Is your IP actually safe in China?

Safer than the folklore says – if, and only if, you respect how the system works. China is a first-to-file jurisdiction: rights generally belong to whoever registers first, not whoever invented first. The operational consequence is blunt. Register your trademarks and patents at CNIPA – the China National Intellectual Property Administration – before you start taking meetings, showing decks, or sampling suppliers, because every unregistered mark you expose is available to a faster filer.

Structure matters as much as timing. The WFOE's single strongest structural advantage is that it registers IP in its own name and controls enforcement unilaterally – no partner's consent needed to file, to license, or to sue. The high-risk structure is the joint venture, and the research corpus is unambiguous about the rule: never assign core IP to a JV entity. License it in under a recorded agreement with automatic termination rights, so that if the partnership dies, the IP walks out with you instead of remaining co-owned by your former partner. Most of the famous China IP disasters are, on inspection, JV-structure disasters.

When disputes do happen, the venue options are Chinese courts – which have grown considerably more credible for IP plaintiffs, including foreign ones – or arbitration through CIETAC or HKIAC. And here the cultural thread of this essay reappears in legal costume: Chinese legal culture emphasizes relationship preservation and face-saving, and the aggressive, confrontational posture that wins in an American courtroom measurably backfires here. Counsel who know how to negotiate within that culture routinely get better outcomes than counsel who litigate at it. Even your lawsuits go better with a good intermediary.

Do you need to be there in person?

Yes. Face-to-face meetings still matter a lot here, and people prefer them – that preference is not nostalgia, it is how trust gets built in a relationship-gated market. The deal velocity difference is stark: relationships advance over dinners and site visits at a pace that video calls simply do not replicate, because showing up is itself the signal. Presence demonstrates commitment; repetition demonstrates reliability; and both are prerequisites for the trust that everything else in this market runs on.

For a Western founder, this converts directly into planning arithmetic: a China entry run entirely from abroad is, in practice, an entry that has chosen the relationship ceiling as its altitude. If the market matters to you, someone senior – ideally a founder – spends real, recurring time on the ground, especially in the first year. This is also, quietly, the best argument for the student and alumni networks around institutions like Peking University and Tsinghua: they are dense, high-trust relationship networks that compound precisely because everyone in them keeps showing up in person.

How would we sequence a China entry?

Pull the threads together and a sequence falls out. Most Western founders run it backwards – entity first, relationships last, because that is the order a US launch would use. Here is the order that matches how this market actually works:

1. Relationships first. Before any paperwork, spend time on the ground. Attend the industry events, work the alumni and chamber-of-commerce networks, take the dinners. You are not selling yet – you are becoming someone who can be vouched for. This phase feels unproductive to a Western operating instinct. It is the highest-leverage work of the entire entry.
2. Engage an intermediary. Once the network can produce one, retain a trusted middleman for your first real transactions – supplier negotiations, distribution conversations, government touchpoints. Budget for them the way you budget for legal counsel, and resist the disintermediation instinct. Middlemen make everything easier here. Much easier.
3. Register IP before you need it. File at CNIPA – first-to-file means the safe time to register is before your brand or technology is exposed to the market, not after the first term sheet.
4. Form the entity – almost always a WFOE. With your sector off the 29-item Negative List, take the 100% ownership the law now gives you: $13k–25k, 4–8 weeks, full IP control, full repatriation rights. Commit only the registered capital you will actually pay in within Article 47's five-year window. Reach for a JV only if the Negative List forces it or a partner holds something genuinely irreplaceable – and even then, license IP in, never assign it.
5. Operationalize payments and tax early. Alipay with an international card on day one; local bank account and properly configured fapiao invoicing as soon as the WFOE exists; HNTE certification on the roadmap for the 15% CIT rate; the SAFE-registered dividend route understood before the first profit needs to travel.
6. Keep showing up. The entry is not an event, it is a cadence. Face-to-face preference does not expire after the first deal – relationship capital, like the technical kind, depreciates without maintenance.

The meta-lesson is the one we apply to every complex system we work on at Game Changer Labs, whether it is implementing technology inside an enterprise or structuring an IP deal in gaming: find out how the system actually works before optimizing for how you wish it worked. China's system runs on relationships, intermediaries, and presence, with a regulatory layer that has quietly become one of the more open in the developing world. The founders who thrive here are not the ones who fight that design – they are the ones who learn to use it. The market is open. The gate is human. Go meet the people who hold it.