ZHUHAI, Guangdong — The gray-green water off Zhuhai in South China's Guangdong province churns as teams of small underwater robots compete to make a big revolution at China's marine ranches.

This is not a laboratory test. It is the final round of the inaugural Guangdong-Hong Kong-Macao Greater Bay Area marine underwater robot application challenge, held in an active marine ranch in the South China Sea.

By shifting the testing ground from laboratory tanks to the open sea, the competition offered a compelling glimpse into how underwater robotics can turn marine ranching's most persistent "headaches" into a catalyst for a smarter, more resilient ocean economy.

Over two days in late March, 16 teams from across China pitted their machines against real-world trials far from shore, where unpredictable currents, waves and turbidity put their machines to a rigorous test.

Unlike previous competitions staged in calm pools or shallow coastal zones, this contest required participants to tackle tasks that marine ranch operators face daily: retrieving submerged mooring anchors, harvesting shellfish from the seabed and cleaning biofouling from netting.

Judges stressed that the real-sea setting "forces robots to be optimized not as mere prototypes, but for cost-effectiveness, reliability and user-friendliness, with the potential for rapid commercialization".

The pain points are anything but hypothetical. In 2025, Typhoon Ragasa dealt a heavy blow to the marine ranches of a local fishery called Yuehe, causing more than 700 iron anchors to be lost on the seabed. Since human divers are very costly, most of the anchors remain lost.

"That's nearly 1 million yuan ($145,000) in damages," said Lin Jincheng, the fishery's general manager. Several teams in the competition, he noted, showed impressive autonomy and anti-interference capability — exactly what the industry needs.

For Shenzhen Hanhai Huafan Cleaning Robotics, an underwater cleaning robot producer based in Shenzhen, also a winning team at the event, the market opportunity is clear.

Cai Qianxia, the company's marketing manager, said that its cleaning robots can operate 24 hours a day, achieving efficiency more than 10 times greater than traditional methods.

"Before, if you wanted to clean the hull of a small sailboat, you had to either hire divers to go underwater or wait for your turn at a dry dock. That meant your boat was out of service, you lost money, and honestly, there were just a lot of limitations," said Cai.

In the inspection and monitoring category, Westlake University's team leveraged underwater embodied AI with large language models and multimodal perception, ultimately securing first prize.

Team member Wang Zhangyuan said that the competition has brought algorithms out of the lab and connected them with industrial needs in real-world scenarios.

The challenge in Zhuhai was conceived as a direct response to these industrial needs.

Before the competition, organizers released a 150-million-yuan "opportunity list" covering real operational needs — net inspection, debris retrieval, and ecological monitoring.

That was followed by over 100 million yuan in potential orders from 17 marine ranch developers across Guangdong. Four award-winning teams have already signed preliminary agreements to set up operations in Zhuhai's Xiangzhou district.

The event fits into a broader policy landscape. China's underwater robotics market surpassed 10 billion yuan in 2024 and is forecast to reach 40 billion yuan by 2027.

For the first time, China identified "deep-sea technology" as a strategic emerging industry in its Government Work Report in 2025.

As part of this push, Zhuhai, a major marine economy city, has built 10 truss-type platforms and 452 gravity net cages by the end of 2025.