On January 8th, Beijing time, Zhipu (02513.HK) made history by officially listing on the Hong Kong Stock Exchange, becoming the world's first large-scale model company to go public.

Here are the key details of Zhipu's market debut: an issue price of HK$116.20, an opening price of HK$120.00, a low of HK$116.10, a high of HK$135.00, a closing price of HK$131.50, a peak market capitalization of HK$59.4 billion, a closing market value of HK$57.89 billion, a turnover rate of 8.12%, and a trading volume of HK$2.229 billion.

Zhipu's successful listing sets a precedent for MiniMax, which is slated to go public the following day. Futu's pre-market trading indicates that MiniMax has surged by over 20%.

Zhipu raised HK$4.348 billion through its IPO, while MiniMax secured HK$4.82 billion.

Under normal circumstances, such market valuations and fundraising figures would be deemed commendable.

However, apprehension arises when making comparisons, particularly in the realm of large-scale models.

The top three U.S. companies in this field are OpenAI, Anthropic, and xAI.

On January 6th, U.S. local time (the day before Zhipu's listing), xAI announced the completion of a staggering $20 billion funding round, surpassing its initial target of $15 billion, with a post-investment valuation soaring to $230 billion.

Meanwhile, Anthropic is reportedly in the midst of a new funding round, aiming for $10 billion, which would value the company at $350 billion. The day prior, Anthropic unveiled plans to acquire nearly 1 million TPU v7p chips for its self-built data center, in a bid to reduce reliance on NVIDIA's CUDA ecosystem.

Even more astounding is the news that OpenAI is planning a $100 billion funding round at an $830 billion valuation. Just recently, in October 2025, OpenAI completed a new funding round at a $500 billion valuation, including a $40 billion commitment from SoftBank. On the last day of 2025, SoftBank confirmed the receipt of the final installment of funds, with SoftBank now holding over 10% of OpenAI's equity.

(It's worth noting that companies typically resort to the secondary market only when they fail to secure funding in the primary market. Given their lofty valuations, OpenAI, Anthropic, and xAI are all poised to list on the U.S. stock market. After all, if SoftBank has already raised $40 billion for OpenAI, who else can step in?)

The majority of funds raised by these large-scale model companies are earmarked for purchasing computing power. A substantial portion of these funds flows to NVIDIA, bolstering its revenue and profits, and ultimately propelling NVIDIA to become the world's first company with a market value exceeding $5 trillion. In terms of GDP, this figure ranks only behind China and the U.S.

This underscores why only China and the U.S. possess the wherewithal to compete in this AI arms race.

Nevertheless, the enormous disparity in funding capabilities between Chinese and U.S. AI companies cannot be overlooked.

There was once talk of the "AI Six Little Dragons" among Chinese large-scale model startups. However, as of today, according to Doubao AI's search results, Baichuan Intelligence, 01.AI, and JieYueXingChen among the "AI Six Little Dragons" have not publicly announced new funding rounds in 2025.

Zhipu's closing market value today stands at nearly HK$60 billion. MiniMax will list tomorrow, with a current pre-market valuation similar to Zhipu's. Meanwhile, YueZhiAnMian (Kimi) announced the completion of its latest funding round on the last day of 2025, with a valuation of approximately $4.3 billion.

In essence, when you aggregate the valuations and funding amounts of all Chinese AI large-scale model startups, they fall short of any of the top three U.S. companies and even lag behind a single funding round of OpenAI.

The funding gap is even wider than the actual capability gap. Chinese companies, relying on limited computing power and funds, have developed models whose capabilities are not far behind their U.S. counterparts.

Zhu Xiaohu once remarked that large-scale model startups lack data, computing power, and scenarios, yet they come with a hefty price tag. This may explain why Chinese capital is hesitant to invest, and international capital is also constrained due to China-U.S. relations. Especially SoftBank, which once made significant investments in numerous Chinese companies, now largely refrains from doing so.

However, there remains a cohort of Chinese large-scale model companies that have not raised funds for various reasons, and these companies are striving to build China's foundational capabilities in large-scale models.

First and foremost is DeepSeek, which made waves globally at the beginning of 2025. It once caused the market value of the entire U.S. semiconductor industry to plummet by over 10% in a single day. DeepSeek is hailed as a national treasure in China, and its open-source models are competing head-to-head with the closed-source models dominated by U.S. companies.

The emergence of DeepSeek has given China hope for catching up! Because DeepSeek has blazed a trail to achieve, and even surpass, model performance without relying solely on stacking computing power. We eagerly anticipate DeepSeek's upcoming new release!

Secondly, Chinese tech giants such as ByteDance, Alibaba, Tencent, and Baidu have all made substantial investments in the AI field. Among them, Alibaba and Baidu have built capabilities in chips, models, and scenarios, coupled with their own data accumulation, giving them certain advantages.

Alibaba and Baidu are the closest equivalents to Google in China. Google has its own AI chips, and so do Alibaba and Baidu! With Google's recent impressive performance in large-scale models showcasing its AI prowess, its market value also surpassed Apple's at yesterday's close.

Since Wu Yongming took full charge of Alibaba in February 2025, he announced plans to invest over 380 billion yuan in the AI field over the next three years, with 65% allocated to hardware infrastructure. Research and development investments cover the QianWen large-scale model and AI application ecosystem.

ByteDance's investment in the AI field is even more noteworthy, as its capital expenditures in AI can better drive the prosperity of domestic computing power. In terms of revenue, ByteDance's revenue and profit scale this year are expected to approach those of Meta, one of the "Magnificent Seven" in the U.S. stock market. Meta just spent billions of dollars to acquire Manus, a Chinese AI agent startup, a deal that still requires approval from the Chinese government.

In the race towards Artificial General Intelligence (AGI), both China and the U.S. are forging ahead, fearful that their rival will reach the Holy Grail first.

In a recent interview at the beginning of 2026, Elon Musk suggested that AGI might be achieved this year! He also believes that China will overcome its chip challenges and leverage its massive energy infrastructure to brute-force its way through the computing power bottleneck!

China is acutely aware of its computing power shortage but boasts abundant energy and a wide range of application scenarios. Now, AI has permeated various industries to enhance production efficiency and create value, while China is also striving for breakthroughs in chips.

The current AI competition is humorously referred to as a battle between Chinese talent in China and Chinese talent in the U.S.