In October, the China Marine Economy Expo opened in Shenzhen with hull-cleaning robots, river-cleaning craft, seabed cable machines and deep-sea survey vehicles on the floor. Outside the hall, the money had already gone in.
Seawit closed an A round above RMB 1 billion in June. Deep Sea Intelligence raised more than RMB 500 million in August. Ocar took a C round in September, bringing its total for the year to about RMB 300 million. Weilan Deep Sea disclosed five rounds in nine months in October.
But funding is the least interesting number in this sector. The orders are what changed.
Contracts, not demonstrations
Seawit reports more than RMB 1 billion of orders this year. Deep Sea Intelligence signed eight work-class ROVs in South America and South East Asia with a combined contract value of RMB 300 million, for delivery across 2026 and 2027. Ocar and its subsidiaries won two water-cleaning projects worth about RMB 27.17 million.
Three numbers, three different end markets: hull servicing, offshore engineering and municipal waterway maintenance. Demand is not appearing in one niche by accident.
Why the maths works under water
Above the surface, a robot’s rival is a worker’s wage. Below it, the rival is an entire system. Past 120 metres, a saturation diving quote can exceed RMB 1 million per day, and the customer is also paying for vessel time, weather windows, downtime and the cost of an accident.
The value unit is therefore not headcount saved but ship days avoided. Manual cleaning of a large cargo ship takes more than ten days. A robot compresses that to roughly ten hours. In July, Lujian Technology, part of Shandong Port Logistics, used a robot for the first underwater hull clean of the Huizhong 5.
That is why Zhu Xiaohu, one of China’s most quoted early-stage investors, argued in September that hull cleaning is a better first market than the home. It is invisible to people on shore, it produces a measurable saving, and the capability generalises to offshore solar cleaning and port dredging far more easily than it generalises into a living room.
Three companies, three commercial strategies
Deep Sea Intelligence sits closest to conventional offshore equipment. Its task is not to prove a model is general but to get a multi-tonne machine into the procurement systems of oil, gas, subsea cable and marine engineering customers. In October it delivered an all-domestic commercial deep-sea robot to a Chinese state-owned enterprise and completed a 1,200-metre sea trial. Its August round drew market institutions, central state capital and energy industrial capital, including a fund anchored by TotalEnergies.
Weilan Deep Sea inverted the order. It sells demonstration projects, then standardised services, then machines. Large oil, gas and wind customers will not hand over subsea assets on the strength of a demo, but they will accept an inspection result. The company enters through the acceptance criterion rather than the capital purchase.
Ocar took the scale route on the surface, where recovery, communications and recharging are easier. Its surface robots and unmanned boats cover more than ten countries and 75 cities, with over 1,000 units delivered and more than 2 million kilometres of autonomous navigation. Its commercial boundaries have since widened from cleaning boats to cargo vessels, official craft and unmanned workboats, and its incubated electric yacht brand MIRAVA raised close to RMB 50 million in angel funding in October.
The three pages still blank
The first is data. Language models had the internet. Humanoids have video, simulation and teleoperation. Underwater has almost no public pool. Weilan’s founder Zhang Dayu has said plainly that underwater data is scarce and that accumulation is the whole game. In October, Dalian Maritime University and others released the UPED dataset combining RGB imagery, sonar, lidar point clouds and magnetic measurements, which tells you how early the base layer still is.
The second is profit. A RMB 300 million contract can mean high revenue or a long delivery cycle and heavy working capital. A recurring service contract can mean predictable income or a permanent crew and a fleet of boats.
The third is proof. No Chinese company with marine robotics at its core is listed, so there is no multi-year record of revenue, gross margin and cash conversion to compare against. Deep Blue is in STAR Market review and not yet profitable. Nauticus Robotics reported USD 900,000 of revenue, a USD 11.1 million net loss and about USD 2 million of cash in its second-quarter 2026 results.
That is the whole argument in one line. Marine robotics has found a customer who will pay a real price for a real outcome, at a price point that dwarfs anything a humanoid can charge today. It has not yet shown that the operator keeps any of it.
Editor’s note: This is an adapted translation of the original OFweek report. It has been trimmed and restructured for readability for an international business audience. The full original (in Chinese) is at https://robot.ofweek.com/2026-10/ART-8321202-8420-30708799.html.