Unsated by Hong Kong listing, Knowledge Atlas seeks more funds in Shanghai

Just five months after its Hong Kong IPO, the capital-hungry AI large model company has launched a plan to list on Shanghai’s Nasdaq-style STAR market

Key Takeaways:

• Knowledge Atlas plans to raise up to 15 billion yuan through a listing on Shanghai’s STAR Market, more than double what it raised in its Hong Kong IPO earlier this year
• The AI company’s Hong Kong-traded shares rose more than 10-fold at one point after their debut at the start of this year

By Lee Shih Ta

Talk about double dipping.

Just five months after debuting its shares in internationally focused Hong Kong as “China’s first large model AI stock,” Knowledge Atlas Technology Joint Stock Co. Ltd.（2513.HK) has announced plans for a second listing on the Shanghai Stock Exchange’s Nasdaq-style STAR Market targeting domestic investors. Emboldened by the meteoric rise of its Hong Kong shares since their January debut, the company is aiming to raise up to 15 billion yuan ($2.22 billion) in the second listing, more than triple what it raised in the first one.

As a newly listed company, Knowledge Atlas’ rapid return to the capital markets in such a short span underscores just how China’s AI sector has become a frenzied dash for cash, as companies rapidly burn through their capital.

According to its Hong Kong filing last week announcing the new IPO plan, Knowledge Atlas plans to issue up to 38.8 million Shanghai-listed shares, representing about 8% of its enlarged share capital. The new funds will be used in three ways: for AI general-purpose foundational large language models (LLM); for a large model-as-a-service (MaaS) one-stop service platform; and for the replenishment of working capital. The AI general-purpose foundational large model project is the most capital-intensive, requiring 12 billion yuan alone.

The company, previously known as Zhipu, raised net proceeds of about HK$4.9 billion ($625 million) from its Hong Kong IPO in January. As of May 29, it still had about HK$2.84 billion in unutilized funds from the listing, indicating the Shanghai IPO is being driven more by industry competition and long-term planning than short-term operational needs.

Knowledge Atlas has developed at a striking pace these last few years. Its revenue last year reached 724 million yuan, up 131.9% from 2024, though its adjusted net loss was even higher at a hefty 3.18 billion yuan, widening from a 2.21 billion yuan loss in 2024. Put differently, the company’s loss last year was more than quadruple its revenue, showing large model developers are still huge cash burners. The company’s focus isn’t short-term profitability, but rather developing model capabilities, computing power scale and gaining market share.

Knowledge Atlas’ proposed Shanghai fundraising is equivalent to more than 20 times its full-year revenue for 2025. For an unprofitable large model enterprise, such massive financing exceeds conventional business expansion needs and looks more like preemptive resource stockpiling as competition is likely to remain intense in the years ahead.

Hong Kong first, Shanghai later

Knowledge Atlas’ decision to tap the Shanghai market isn’t due to a lack of enthusiasm for its Hong Kong shares. After going public by selling shares for HK$116.20 apiece in January, the company’s stock skyrocketed as high as HK$1,291, representing a 10-fold rise, pushing its market capitalization past HK$570 billion. Its current price-to-sales (P/S) ratio has reached an astronomical 769 times, positioning it at the very high end among even the priciest global AI concept stocks.

That shows Hong Kong investors have not only bought into Knowledge Atlas’ AI narrative, but have embraced the entire package by awarding the company such an aggressive valuation. That reality further supports the premise that the new Shanghai listing is aimed at raising more funds, rather than simply chasing a higher valuation.

In the past, many companies listed in Shanghai and Hong Kong typically listed in the former city first, tapping domestic investors, before moving to the more internationally focused Hong Kong later. But AI enterprises are breaking with that tradition. As Hong Kong investors become more receptive to unprofitable tech startups, a growing number of Chinese AI startups are beginning to opt for Hong Kong first, before returning to domestic markets in Shanghai and Shenzhen later for additional financing.

In terms of its capital depth and policy support, the STAR Market still boasts advantages that Hong Kong can’t easily match. In recent years, domestic markets in Shanghai and Shenzhen have routinely assigned higher valuations to hard tech companies than Hong Kong. Take semiconductors, for instance, where players like Cambricon Technologies (688256.SH) and Hygon Information (688041.SH) have both achieved market capitalizations in the hundreds of billions of yuan. Investors prize not merely profitability, but long-term narratives such as independence, domestic substitution and being part of China’s national tech roadmap.

Knowledge Atlas is not an isolated case. MiniMax (0100.HK), another AI company that listed in Hong Kong early this year, has also initiated the process for an eventual listing in Shanghai or Shenzhen. Although it remains premature to conclude whether the “Hong Kong first, Mainland later” route will become mainstream for Chinese AI companies, the nearly simultaneous return to the Mainland by two leading large model companies demonstrates that capital markets are increasingly vying for hot AI companies.

For Knowledge Atlas, the return to Shanghai after debuting in Hong Kong has some strategic significance beyond simply more fundraising. The Hong Kong market has already taken the lead in establishing a market-driven valuation for the company, proving that investors are willing to pay a premium for its large model narrative. But valuation is merely the first step for a company with such big cash needs, and more crucial is its access to sustained financing.

What Knowledge Atlas needs is not simply general corporate growth capital, but rather massive funds to prop up its model R&D, computing power construction, and commercial expansion for the next few years. Listing on the STAR Market after going public in Hong Kong seems, to some extent, like a strategy of proving its worth in Hong Kong first, before returning to the Mainland market where funding and policy support are more abundant.

Both Knowledge Atlas and MiniMax are currently in a phase of intense investment, signaling that China’s large model sector still has a long way to go before offering a more mature business model for investors. Over the past two years, Chinese AI firms have competed on model capabilities. In the years ahead, the battlefield may well shift to financing prowess. As technological gaps steadily narrow, whoever can keep securing computing power, talent, and capital support will have the best shot at staying in the game.

To subscribe to Bamboo Works weekly free newsletter, click here