“The IPO is important for us to attract the most talented engineers to join us, to buy more GPUs, and to collect more data,” Victor Huang, Manycore’s chair and one of its cofounders, told Fortune ahead of the trading debut.

The Hangzhou-based startup is a bet on “spatial intelligence,” moving beyond the word- and language-based work of large language models like OpenAI’s GPT and DeepSeek’s V3 to instead create AI models that can autonomously work in the real world.

These programs, also called “world models,” are key to operations like robotics and autonomous driving, where machinery has to react to external stimuli, like how a robotaxi needs to slow down in response to changing traffic conditions. 

Huang described spatial intelligence as similar to a person or animal’s innate ability to understand the world around them. “When you enter a room, you can understand where you are and what’s in front of you. And if you want to take a seat, you can understand which seat is empty,” he explained.

“People are now trying to apply AI in the physical dimension,” Jixun Foo, senior managing partner at the Singapore-based venture capital firm Granite Asia, and an early backer of Manycore, said. He pointed out that viral videos of humanoid robots dancing, while impressive, are often carrying out pre-programmed routines. “If you want a different performance you have to program it again. You can’t just tell the robot to do this or that action.”

Graham Uden for Fortune

Some of AI’s biggest names are also working on “world models.” Both ImageNet creator Fei-Fei Li and former Meta chief scientist Yann LeCun see these models as the next step in AI development. 

LeCun has argued that video data can help train world models, but Manycore and Huang instead think the startup’s vast repository of 3D assets will be a more useful data set. “I don’t believe that if you have enough video, you can train the rules of the physical world,” Huang said.

Instead, “we’d accumulated a huge amount of 3D data, almost 500 million assets from the real world. We had the training data, so we believed we could make the best physical AI in the world,” he argued. 

China’s AI sector has released many of its models on an open-source basis, which has helped to boost the reputation of its AI startups and won converts in the global tech sector, including in Silicon Valley. “People try Chinese AI. It’s free. It’s open-source. And when they try it, it’s great,” Huang explained. Manycore has already released several open-source models, including SpatialLM, a spatial language model that can understand and generate 3D environments, and SpatialGen.

Still, in recent weeks, some tech companies, like Alibaba and Knowledge Atlas, better known as Z.ai, have started to release models on a proprietary basis, at least in the initial stages, as monetizing AI work has proved tricky for Chinese companies.

But Foo thinks a company like Manycore can preserve its edge even if it open-sources its models. “For Manycore, it’s not just about their model, but also the data set they have built. That dataset is unique to them, right? If you have a competitive edge that you can hold on to, then you can open-source something,” Foo said.

The first of the ‘Little Dragons’ to go public

Manycore, founded in 2011, is one of the “Six Little Dragons,” an informal group of six tech and AI startups based in Hangzhou, now one of China’s leading AI hubs. Manycore is the first of the “dragons” to tap public markets; Unitree, the buzzy robotics manufacturer, will list on Shanghai’s stock exchange later this year.

The company got its start as a design software business, building Kujiale, a platform that lets users create 3D renders of interior spaces, and its international counterpart Coohom, which now serves customers in more than 200 countries. IDG Capital and Hillhouse Investment are among its previous backers.

Huang was an engineer on Nvidia’s CUDA team before returning to China to build a business around rendering. “The economy in the U.S. wasn’t doing well at the time, but in China, real estate was booming,” he recalled.

That work helped convince Foo, who spent time at HP, to back the company. “I used to use a lot of 3D software when I was designing HP printers,” Foo explained. “And I thought this was pretty cool: I was using it for mechanical products, and now they are doing it for a physical world.”

According to its IPO prospectus, Manycore generated 820 million Chinese yuan ($120 million) in revenue last year, growing by 8.6% from the previous year. It also earned a slim operating profit of 18.6 million yuan ($2.7 million), even as it posted an overall net loss of 428 million yuan ($62.8 million). 

China’s real estate market is still in the throes of a prolonged slump, which Foo admits is a “headwind” for Manycore. Still, the company is expanding into international markets. “What gives me comfort is the resilience of the team. They stuck in there, and they figured things out.”

Design companies have been hit hard in recent months; shares in Adobe and Figma have plunged as AI providers like OpenAI and Anthropic work design tools into their models. Some design startups are reinventing themselves as AI companies: Canva on Thursday launched a new suite of agentic offerings on Thursday that allow users to automate much of the design process. 

A Hong Kong boom

This isn’t Manycore’s first attempt at public markets. The company was on track for a U.S. listing in 2021, before withdrawing its application after Beijing regulators scrutinized Didi Global’s U.S. IPO and forced the ride-hailing giant to delist, unnerving other Chinese companies considering U.S. listings. 

“Nowadays, the Hong Kong market is the best for a Chinese AI company,” Huang says. 

Manycore is just the latest AI IPO to hit Hong Kong’s market this year. AI and AI-related companies have led to a surge in debuts in the Chinese city, with listings raising almost $14 billion in the first quarter of the year. Some shares have surged by eye-popping amounts: MiniMax and Knowledge Atlas, two AI model developers, have risen by around 450% and 650% respectively since their early January IPOs.

More IPOs are on the way. Victory Giant, which makes printed circuit boards, hopes to raise $2.2 billion in its IPO; its shares will debut on April 21. Other startups reportedly considering IPOs are Kimi developer Moonshot AI and smart glasses manufacturer Rokid.

Bonnie Chan, CEO of Hong Kong Exchanges and Clearing, pushed back at this week’s HSBC Global Investment Summit against the characterisation of the market as simply a conduit for Chinese capital. “When people say that Hong Kong’s stock exchange is just hosting Chinese companies, it’s not doing those companies justice,” she said, noting that around 40% of companies that listed in Hong Kong last year generated more than half their revenue from non-Chinese sources. “I call them Chinese MNCs. They’re very international.”

China’s AI sector has been under fresh scrutiny from global investors since early last year, when DeepSeek—another of Hangzhou’s “Little Dragons”—released its powerful and surprisingly efficient models, changing the conversation about Chinese innovation.

Physical AI, in particular, is emerging as a Chinese strength. The country’s dense manufacturing ecosystem can produce robots, sensors, and advanced components at a cost below rivals in other countries. Heavy investment in the electrical grid also allows China to rapidly expand the data centres needed to train large models. “It’s not just a race on the foundational model,” Foo said. “It’s a race on the infrastructure. It’s a race on compute. It’s a race on energy.”

But Foo and Granite Asia, which was born from the Asia operations of former VC giant GGV Capital, aren’t entirely focused on the Chinese market. “We are more pan-Asian,” he explained. Earlier this year, Granite Asia partnered with DBS to launch a new $110 million fund to give the Southeast Asian bank’s customers “early access” to IPO-stage companies in the region.

“Our strategy is to be able to invest in companies early. We want to help them grow and scale outside of their home market,” Foo says. “We have a good pipeline of companies going IPO.”