A computer-vision scientist who spent two decades inside university labs and a tech giant's research group took a different bet: skip the general-purpose chatbot race and aim straight at the factory floor.
His company just put its paperwork in front of Hong Kong's exchange, asking public investors to back a very specific kind of software-and-hardware helper.
The pitch is easy to say and hard to do — let machines inside factories inspect, sense, and decide without a human watching every step.
What the filing shows
SmartMore (思谋科技) submitted its prospectus to the Hong Kong Stock Exchange on March 16, 2026, intending to list on the Main Board. The joint sponsors are Morgan Stanley, China International Capital Corporation, and Deutsche Bank. The company calls its business "industrial AI agents" (工业AI智能体): a bundle of robots, edge sensors, and agent software that lets manufacturers move from fixed automation toward systems that can adapt on their own.
Founded in 2019 at Hong Kong Science Park by Jia Jiaya, a former Tencent Youtu Lab researcher and longtime academic, SmartMore deliberately stayed away from the crowded general large model (大模型) contest. Its own IndustryGPT underpins the agent stack, but the product is sold as finished industrial capability rather than as a model to be tinkered with.
The customer list is the headline
By the end of 2025 the company said it served more than 730 customers worldwide, a roster that includes Tesla, Carl Zeiss, Luxshare, Goertek, BOE, CRRC, CALB, and Kedali. It also said it had delivered about 140,000 industrial agents across consumer electronics, new energy, precision manufacturing, and rail transit. Those names are the strongest evidence that its approach can survive outside a demo room.
Revenue tells the same growth story in hard numbers. SmartMore reported 485 million yuan in 2023 (US$68.3 million; HK$534 million), 756 million yuan in 2024 (US$106.5 million; HK$832 million), and 1.086 billion yuan in 2025 (US$153 million; HK$1.19 billion), a year-on-year rise of about 43.7 percent. Gross margin improved from 30.5 percent to 37.3 percent across the same window.
The loss column is the caveat
The company is not yet profitable on a reported basis. Net loss widened to 991 million yuan in 2025 (US$139.5 million; HK$1.09 billion) from 735 million yuan in 2024 (US$103.5 million; HK$809 million) and 546 million yuan in 2023 (US$76.9 million; HK$601 million). On an adjusted basis, which strips out share-based payments and similar items, the loss narrowed to 272 million yuan in 2025 (US$38.3 million; HK$299 million) from 379 million yuan in 2024 (US$53.4 million; HK$417 million) and 394 million yuan in 2023 (US$55.5 million; HK$433 million). Management warns losses may continue while it invests in research and commercial expansion.
Who already owns a piece
Before the listing, SmartMore had completed seven financing rounds totaling about US$565 million, most recently at a post-money value of US$1.23 billion. IDG-affiliated entities hold roughly 11 percent and Lenovo-linked entities about 4.27 percent, with Sequoia China, Hidden Hill Capital, and ZhenFund also in the cap table. A notable backer is the Hong Kong Investment Corporation, the city's "patient capital" vehicle, which named SmartMore as its first strategic partner in June 2024.
Why Hong Kong, why now
Hong Kong has become the default venue for China's AI and robotics names that are too large or too unprofitable for the mainland's stricter boards. A Main Board listing gives SmartMore a currency to keep hiring researchers and to fund the commercial push into overseas factories. The prospectus says proceeds will go to technology and product development, commercialization, and strategic investment or acquisition.
SmartMore is not alone in the queue. A wave of China's AI companies has chosen Hong Kong over Shanghai's STAR Market or Shenzhen's ChiNext because the exchange tolerates pre-profit stories and gives foreign investors a cleaner on-ramp. That same tolerance means buyers must do their own diligence on the path to profit, since the listing gate is lighter than on the mainland boards.
Deployment takeaways
According to Li An, chief scientist at BrainNet (脑机网), the durable moat in industrial agents is proprietary field data from real production lines, not the model itself, and that is what a public market should price.
Honest limitations
Figures here come from the company's HKEX prospectus as reported by Chinese financial media; they are not independently audited by this publication. The customer and delivery counts are company-stated. SmartMore has filed but not yet priced or listed, so no offering price or market capitalization at listing exists yet. The adjusted-loss figures depend on the company's own non-IFRS definitions. Currency conversions use fixed rates of 7.1 yuan per US dollar and 1.1 Hong Kong dollars per yuan.
What readers can do now
- Founders should study how SmartMore packages agents as finished industrial capability rather than as a bare model, a lesson for any vertical-AI pitch.
- Investors should watch the adjusted-loss trajectory and gross-margin trend through to listing, since profitability timing drives the valuation more than revenue growth alone.
- Buy-side analysts should track the eventual offering price against the US$1.23 billion pre-listing value to gauge how public markets rate industrial-AI risk.