Ring the bell from Anthropic and see how China’s large models are priced?

Light cone intelligence

2026.11.20 10:47

· From Beijing

Full text 4808 words

Text | Light Cone Intelligence, author|Wei Linhua, editor|Liu Junhong

At 11:48 a.m. Eastern Time on November 19, under the aura of "the largest IPO in human history", Anthropic (stock code: ANT) officially opened for trading on Nasdaq.

Compared with Anthropic's finalized issue price of US$83, the opening price opened directly higher to US$111.50, a surge of 34.3%, and once reached US$115.10 during the session. On a fully diluted basis, Anthropic's valuation reaches a maximum of US$2.02 trillion, surpassing SpaceX, which went public in June this year.

However, the carnival after the opening lasted for less than half an hour. By the close, Anthropic's market value was like riding a roller coaster.

Starting in the afternoon, Anthropic's stock price plummeted, hitting an intraday low of $92.72 and finally closing at $97.17. Although it still closed up 17.07% from the issue price of US$83, Anthropic has lost approximately US$324.4 billion in market value since the day's high, and based on a fully diluted basis, the company's valuation has dropped from US$2 trillion to approximately US$1.76 trillion.

The bell is ringing on the other side of the ocean, and in the Chinese AI circle, almost all large model founders, investors and secondary market analysts are staying up late to keep an eye on the market. What everyone is looking for is not the rise and fall of Anthropic, but an answer: When the "anchor" of global AI valuation falls into the deep waters of the public market, how should China's AI models be priced next?

After all, on the eve of ringing the bell, Anthropic’s price has dropped again and again, surrounding Anthropic’s pricing, from its initial price of $2 trillion by investment institutions, to Anthropic’s valuation being slashed during its Q3 roadshow, to the U.S. midterm elections, Nvidia’s sudden “coaching change” and other impacts.

The impact on IPO continues at this node. For Anthropic, it is more like a stress test for the entire AI industry.

The listing of Anthropic has never been a story of "whether it is expensive or not", it is a story of "how to calculate".

The new ruler that Wall Street uses to examine Anthropic today will remain unchanged on every large Chinese model company.

Anthropic’s valuation has risen, but it has long been squeezed out

As a "demonstration" for global large model companies, the 17% increase in Anthropic's stock price is normal.

The buying enthusiasm in the market is still there. According to news revealed by Bloomberg, the scale of subscription orders has reached about 6 to 7 times of the available shares, and the subscriptions mainly come from large intention orders from institutional investors.

As a result, Anthropic not only has the confidence to raise prices, but also plans to make more money when the market is good, so as to save more funds for itself.

By November 18, Anthropic finally priced its stock price from $70-78 to $83. Anthropic forcibly expanded the number of issued shares by 19%, from the original planned 825 million shares to 980 million shares, which resulted in nearly $20 billion in additional cash.

When demand for stocks exceeds supply, Anthropic also has a number of computing power suppliers to protect it:

Just before the listing, a group of technology giants called "Global Computing All-Stars" - including Nvidia, Google, Microsoft, AMD, Broadcom, Micron, and TSMC partners, agreed to subscribe for Anthropic shares worth US$26 billion. Among them, Google and Nvidia were the most generous, spending US$10 billion each.

These computing power suppliers and cloud vendors have stepped forward, which is equivalent to being firmly bound to Anthropic and locking in a computing power order in advance.

The investment of US$26 billion is more like a "strategic lock-in fee" paid by the giants. As long as Anthropic continues to expand crazily and train next-generation models with this money, the money will go around and eventually turn into orders, flowing back to Nvidia's chips, Broadcom's servers, and the computer rooms of major cloud manufacturers along the procurement contract.

Among them, who is the largest holder behind Anthropic? The answer is Amazon and Google. According to data disclosed by Anthropic, Amazon’s total shareholding ratio is 14.7%; followed by Google, which holds 13.7%.

Although the company's market value finally fell to around US$1.76 trillion, if we look at the timeline, today's ups and downs in Anthropic's market value were actually foreshadowed three months before it went public.

In August this year, the market was still enthusiastically calculating an account of US$2 trillion for Anthropic. The pricing logic at that time was simple and crude, just like the enthusiasm for big models in China's primary market: as long as the model is still iterating and the revenue growth rate is still soaring, the imagination of the company's valuation can open the skylight.

But reality has poured cold water on it. Since October, Anthropic has experienced multiple "moments" when its valuation has dropped:

First, the revenue growth rate was lower than expected, causing investment institutions to begin to re-examine its value. Anthropic disclosed in its October roadshow that the company’s Q3 revenue growth rate dropped to 47% from 144% in Q2. Therefore, investment institutions gave it a squeeze and raised its valuation to 1.5 trillion.

The above-mentioned adjustments still question Anthropic itself. The subsequent decline in the company's valuation is more of a concern for the prospects of the AI ​​industry.

After the U.S. midterm elections, pessimism about the tightening of data center regulations has caused another bloodbath in U.S. AI stocks. On November 6, the Nasdaq fell 5.52%, and the Philadelphia Semiconductor Index fell about 12%.

In the process, Anthropic's market value fell, and the imagination premium evaporated.

In addition, the alignment of major manufacturers and the binding of computing power will also make investors cautious. After all, how much of the glorious financing and soaring revenue on the books comes from the natural demand of the real market, and how much is a cyclical game created by the giants who "invest money with the left hand and sell computing power with the right hand"?

On the basis of a total commitment of US$900.9 billion in purchase commitments and lease agreements, the sudden resignation of Jen-Hsun Huang, who has been at the helm of Nvidia for 33 years, and the new signal released by CFO Colette Kress after taking over as CEO, have taken away the most critical load-bearing pillar in the AI ​​computing power financing game.

During NVIDIA's earnings call, Kress publicly stated the new rules: NVIDIA will scrutinize companies that provide financing support extremely strictly in the future. The data center guarantee will only be activated after the facility is actually completed and ready. No guarantee will be provided during the construction period; once the data center construction is delayed, NVIDIA's guarantee will no longer exist.

In the past two years, Nvidia has not only sold shovels, it has also been the "super guarantor" of the entire AI infrastructure construction. As of the end of October this year, NVIDIA's external AI infrastructure financing guarantee exposure has snowballed to US$11.8 billion.

In the past, the market dared to give high valuations to large-scale model giants like Anthropic, which are highly bound to computing power. To a large extent, it was acquiesced that NVIDIA was behind it and its big brother would take care of it if anything happened. Now, the eldest brother has suddenly changed. The first thing the new guy does when he comes on board is to take back the unlimited blank check. It is hard for the market not to worry.

When the credit endorsement is no longer effective, everyone has to re-examine whether Anthropic's future growth expectations can be fulfilled, and how much it can be fulfilled.

The same is true in the Chinese market.

In the past two years, many large domestic model companies have relied on binding cloud resources from major manufacturers and making strategic investments to support high valuations. However, Anthropic's experience has revealed the outcome in advance. Once it enters the open market, sooner or later investors will have to separate the subsidies from large manufacturers from their commercial performance and settle accounts separately.

Anthropic's listing has sent a signal to China's large model companies - the era of relying on a single-dimensional high-growth story to dominate the secondary market is over.

Looking at the reality that Anthropic has dropped from US$2 trillion to US$1.5 trillion in the cloud, we can't help but worry. It also relies on revenue growth to raise its valuation. What will be the next fate of domestic large model companies?

Valuations are squeezed and demand is frantic. How should China’s large model companies be priced?

How to view the valuation of China's large models starts with the changes in Anthropic, because what has changed is not Anthropic, but the pricing ruler in the hands of the capital market.

In fact, compared to Wall Street, which has only begun to "squeeze water" from Anthropic today, the market value of China's large model companies has actually been in a more rational position.

Take the currently listed Zhipu and MiniMax as examples. Long before Anthropic rang the bell, the two companies had already demonstrated the process of "valuation squeeze" for the entire industry in advance.

When it landed on the Hong Kong Stock Exchange in January this year, Zhipu and MiniMax were equally successful, and Zhipu even set a record of a market capitalization of one trillion yuan. Why were they so ridiculously expensive back then? The core lies in two words: scarcity of supply and surging ARR narrative.

When MiniMax was listed, there were only about 17.07 million shares outstanding in the Hong Kong stock market, accounting for less than 6% of the total share capital. However, it attracted 420,000 crazy subscriptions, with oversubscription as high as 1837.17 times; Zhipu also achieved an oversubscription record of 1159.46 times.

But just a few months later, looking at the current valuation of the company, compared with the highest point, the market value of both companies has dropped by more than 80%. Business income is still soaring, but when the lifting of the restricted stock ban is approaching, the scarcity premium of the circulating stock is instantly wiped out.

What is even more worrying is the large model companies that are still on the poker table and have not yet been listed - the chill in the secondary market is inevitably pouring back into the primary market.

Overseas big brother OpenAI, in order to replenish ammunition before Anthropic goes public, OpenAI has also been intensively promoting a new round of financing recently. OpenAI's original ambition was to raise US$30 billion at a valuation of US$1.4 trillion, but investors' appetite was obviously not that big.

In the end, the valuation of this round of financing dropped from US$1.4 trillion to US$1.25 trillion, directly evaporating US$150 billion, a drop of 10.7%, and the initial payment also shrank to about US$15 billion. Under the scrutiny of capital, OpenAI deserves a discount.

Even OpenAI has to take the initiative to lower its head and offer discounts. So if we turn our attention back to China, how much room will there be for the large Chinese model companies that have not yet been listed to increase their valuations in the future?

After the water is squeezed out, the value of China's large model companies no longer depends on the pie, but must be tied to the real commercial monetization capabilities.

However, if we zoom out to the entire global AI industry, although the capital market is frantically squeezing out AI valuations, there is still huge room for exploration of the real demand for AI in other industries.

Objectively speaking, the real demand for AI applications is not only firmly established, but will also usher in a real full-scale outbreak period.

In the past two years, China's AI circle has been quite anxious about the commercialization of large models: B-side MaaS revenue is limited, and C-side has yet to find a real hot product.

By 2026, with the explosion of Coding and the demand for Agents, Token prices began to increase.

According to the report "Large Model Word Element Service Performance Monitoring Report (2026)" released by the China Academy of Information and Communications Technology, the domestic normalized average price in the second half of 2025 will be 6.8 yuan, while the average price so far this year is 13.46 yuan, an increase of nearly 100%. However, foreign prices are still higher than domestic ones. The average normalized price of typical models released since 2026 is about 117.8 yuan.

Why are people still rushing to use it even if the price has increased? Because when large models truly penetrate complex production links such as code development and enterprise-level agents, customers no longer want cheap chat toys, but productivity that can directly work for others. As long as the model is smart enough and can run the workflow stably, companies are willing to pay real money.

Nowadays, when technology still has the final say, large model companies roll out models, and computing power is still a scarce item. Therefore, AI hardware and cloud vendors are still the biggest beneficiaries:

During NVIDIA's earnings call, the interim CEO recently disclosed a set of data. The rental price of the new generation Blackwell architecture B300 chip has soared by more than 60% compared with when it was first launched; even the multi-year renewal price of the previous generation Hopper architecture old card is up to twice the original price. Whether it is a very large-scale cloud vendor or an emerging computing power cloud, all the computing power in hand is operating at full capacity.

Domestic computing power is still in short supply. On November 18, Alibaba CEO Wu Yongming said in an earnings conference call that as production capacity ramps up, the proportion of domestic computing power in the overall computing power will continue to increase, becoming the main engine supporting cloud revenue growth.

Next, with supply chain constraints and the fierce competition among major manufacturers for cards, the shortage of computing power cards on the market will only become more severe in 2027. When the continuously growing application demand hits a tighter supply of computing power, the price of Token will most likely continue to rise.

Seeing this, you can understand that the current AI industry is just capital debugging large model companies, but the industry is still delivering business to AI companies.

Whether it is the circulation game of Hong Kong stocks or the valuation jump in the primary market, as long as the valuation is ahead of commercial realization, sooner or later the secondary market will use the most brutal way to squeeze out the water for you.

Players who stay at the poker table need to use Token’s “price increase right” to prove its irreplaceability. The fact that the Token price will increase in 2026 and the number of calls will skyrocket has pointed out a way for China's large model companies to survive - whoever can make the company still inseparable from your API after the price increase will have the real pricing power.

In the final analysis, the AI ​​valuation correction sweeping across China and the United States often has nothing to do with the efforts of a single company, but the process of the entire AI industry moving from adolescence to adulthood.

What Anthropic is ringing is the end of the old pricing era. When the tide goes out, the big model companies that are swimming naked in the water need to re-examine how much they are worth.

Respect for the original work, if you are qualified, you should be appreciated.

According to the real-name requirement of the Cybersecurity Law, please bind your mobile phone number before posting a comment.

Express

11:04

National Energy Group established a hydrogen energy technology company in Ordos with a registered capital of approximately 350 million

11:00

Domestic gold jewelry price falls below 1,100 yuan per gram

10:55

Korean exchange launches SIDECAR mechanism and suspends KOSPI programmed selling

10:54

Pork concept rebounded in intraday shock, Jin Xinnong went out of 3 boards in 6 days

10:51

22 private equity firms appeared in the first batch of 15 active ETFs, holding a total of 524 million shares

10:48

Alibaba Q2 of fiscal year 2027: Xianyu daily activity increased by 23% year-on-year, and order volume increased by 29%

10:47

Thai Prime Minister Anutin arrived at the meeting

10:46

Five departments issued the 2026 version of the "Guidance Catalog" for marine industry investment and financing

FTSE China A50 Index futures decline widens to 2%

10:43

The development of my country's marine medicine and functional products industry is accelerating, with a scale of nearly 100 billion

10:40

The robot concept made some changes in the market, and Clai Electromechanical hit the daily limit.

10:33

Spot silver's intraday decline expanded to 3%, at $41.56 per ounce

10:28

The Shanghai Composite Index fell more than 2%, and nearly 5,300 stocks fell in Shanghai, Shenzhen and Beijing.

10:22

The concept of physical AI is partially recovering, and Fantuoshu has increased by more than 10%.

The agricultural sector rebounded in shock, and Wanxiang Denong hit the daily limit.

10:14

Xi Jinping meets with Australian Prime Minister Albanese

10:09

PCB concept rose again during the session, Ultrasonic Electronics and Aohong Electronics hit their daily limit

10:08

Hang Seng Technology Index's decline widened to 4%

SK hynix considers converting its Kioxia-related convertible bonds into stocks

10:07

The concept of ice and snow economy is partially active, and the hot and cold icebergs hit the daily limit

Scan to download App