廣州・深圳創業資訊日報｜2026年10月23日（五）Guangzhou & Shenzhen Startup Briefing | Friday, 23 October 2026

[Zheng Ge's Take] Column: Zheng Ge's blog commentary | Date: 2026-10-23

Five items of startup news from Guangzhou and Shenzhen today — coffee, cross-border e-commerce, AI product sourcing, youth-run shops and cross-border services. They are all the same thing: somebody has already made the road work, and the road is there.

1. AI surges through Guangzhou's coffee industry: more than 8,000 stores, and a 'Coffee Ten' subsidy of up to RMB 5 million

By the end of 2025 Guangzhou had over 8,000 coffee stores — second in the country and first in South China — with an annual output value of RMB 67 billion across the whole industry chain and more than 150,000 jobs; during the coffee culture season from May to August this year the city added over 500 stores. The new 'Several Measures to Promote the High-Quality Development of the Coffee Industry' cover faster customs clearance for green beans, a digital trading platform, up to RMB 30 million per project for technical upgrading and a 'coffee loan' on the finance side, plus a subsidy of up to RMB 5 million for a brand opening an Asia-first or China-first store in Guangzhou. Intelligent equipment is the new selling point: an embodied-intelligence six-axis robot arm plus an AI store-manager system lets one person run ten machines, pushing labour from 30–50% of store cost down to under 8%.

Source: https://gd.chinadaily.com.cn/a/202610/19/WS6ad5764ce4b09a165c78fe7b.html

2. Shenzhen cross-border e-commerce enters 'version 2.0': the big sellers of Huanancheng are receding, small sellers still turn over RMB 30–40 million a year

From backpackers in Huaqiangbei to a hundred-billion industry in Longgang: the 2021 Amazon account ban hit Shenzhen cross-border e-commerce hard — 600 brands and more than 3,000 stores were shut, with industry losses above RMB 100 billion; the Huanancheng trade plaza in Longgang, which once produced the 'Five Tigers of Bantian' and the 'Four Young Masters of Huanancheng', has gone quiet. Yet people in the trade say a small seller can still grind out RMB 30–40 million a year: Zhou Heng, 38, started in 2012 dropshipping on AliExpress with no stock of his own, and now sells small power tools in Huanancheng, turning over RMB 30–40 million a year in a single store and taking a new product from zero to RMB 300,000 in two months. Cross-border, he says plainly, is 'seven parts product, two parts operations, one part luck'; the dropshipping model has no competitive edge left, and building a brand is where things are going.

Source: https://www.ygx2.com/7/274.html

3. Three failed startups, then AI turns the game around: Yan Jingsen of Shenzhen, RMB 5.5 million a year

Born in Longgang, Shenzhen in 2001, Yan Jingsen was already in Huaqiangbei in high school, buying and reselling second-hand phones for the margin. He started with RMB 7,000 from his mother and failed three times. From January 2025 he began selling Guangdong goods — smart accessories, small appliances — into Russia, using a self-built AI sourcing tool to lift screening from 50–100 products a day by hand to 500 products screened properly, close to ten times the throughput; a single storefront holds a steady monthly turnover of RMB 450,000–500,000, and the automated sourcing plus ERP system, live since October 2025, has accumulated 8,000 registered users and brings in over RMB 200,000 a month. In May last year he raised a HK$2 million angel round; the company turned over close to RMB 5.5 million last year, over RMB 600,000 a month on average, and got through the hardest days on a school subsidy and a street-office introduction. 'The one thing Shenzhen never runs short of,' he says, 'is chances to try and fail.'

Source: https://www.163.com/dy/article/L14UKBLL05129QAF.html

4. Liwan's 'youth shops' in Guangzhou: nine young-run stores open a Xiguan stamp trail

The Liwan 'youth shop stamp map', run by the municipal youth league with Wanglaoji and others, is in full swing, covering nine youth shops across the '1+5' commercial zones — railway models, reptile keeping, neighbourhood coffee, handmade crafts, Cantonese-language theatre and more. Collect all nine stamps and you can redeem a design gift and a ticket to an ice-and-snow experience. Behind it is Guangzhou's three-tier 'youth shop' model — collective store, independent store, pop-up — with rent waivers plus a mentor on the ground for those who move in. The Cantonese-culture IP plush toys of Hong Kong young entrepreneur Ip Kin-fai have been picked two years running for Guangzhou's 'must-buy city gift' list. For small and medium bricks-and-mortar founders, this kind of rent-free plus footfall support channel is worth watching.

Source: https://bang.ifeng.com/c/8xHIQN8wbQj

5. A New Zealander called Andre: from 'parcels from Shenzhen' to a company in Shenzhen

Andre Schellenberg, a 32-year-old New Zealander, was breaking up old computers and selling the parts online with his mother's account at the age of 11, and noticed every parcel was stamped 'Shenzhen'. At 19 he came to Huaqiangbei for the first time; at the end of 2025 he registered a company in Shenzhen doing e-commerce, international trade and supplier sourcing, helping overseas firms connect with Chinese suppliers. This year he set up the NZGBA, the New Zealand Guangdong–Hong Kong–Macao Greater Bay Area Network, to bridge importers, online sellers, small businesses and young entrepreneurs in New Zealand: 'Shenzhen's components, engineering, manufacturing, logistics and people turn ideas into reality.'

Source: https://www.10100.com/article/154913165

Zheng Ge's Take

These five look scattered, but they run on one line: the platform dividend has been scattered to the wind; what counts is the ability to land things and the product itself.

First, the number worth watching in the coffee item is not the 8,000 stores — it is labour going from 30–50% of cost to 8%. One person running ten machines plus an AI store manager is not a gimmick; it is the cost structure being re-cut. Tea, dried tangerine peel, the same: the picking, the sorting, the judging of vintage, the parts that lean on the human eye and experience — AI and robots can take over a great deal of it. A tea factory or a peel warehouse ends up on this road too: not cutting people, but using tools so that the same people do more.

Second, cross-border '2.0' is a real turning point. After the Amazon ban wave 600 brands and more than 3,000 stores died, yet the sellers who keep a steady flow still do RMB 30–40 million a year. The difference is exactly Zhou Heng's line — 'seven parts product, two parts operations, one part luck': dropshipping has no moat; a brand does. And that post-2000 founder in Shenzhen, Yan Jingsen, started with HK$7,000 and failed three times, then built his own AI sourcing tool to nearly ten times the throughput and added an ERP that holds 8,000 users and earns RMB 200,000 a month. The tool comes second, but he used the tool to amplify his own judgement — that is the biggest difference between him and the people who think AI will change their fate for them.

Third, the 'youth shop' item is the most down-to-earth, and of these five it is the one we businesspeople should read most closely. Rent-free space, a mentor alongside, a footfall slot — the three things young founders lack most, and the government has topped up all three. Small and medium operators should watch these support channels; do not force yourself to swallow rent at the start, find the policy window and save.

Fourth, the NZGBA and Andre's story shows the part Hong Kong people play. He was taking apart old computers and selling the parts at 11, came to Huaqiangbei for the first time at 19, and ended up running a cross-border sourcing company in Shenzhen — we have walked this road for 26 years. What importers, online sellers and small businesses in New Zealand need is precisely someone who knows the Shenzhen supply chain. What we do is, in essence, exactly this kind of 'landing service'.

One line to finish: new shops open and new brands rise in Guangzhou and Shenzhen every day, but the ones that actually make money are none of them leaning on luck — they grind a threshold out of product and tools. Shenzhen is never short of chances to try and fail; what is short is the skill that stays after the failing.