DNYUZ
No Result
View All Result
DNYUZ
No Result
View All Result
DNYUZ
Home News

China’s A.I. Ambitions Are Running Into a Money Problem

November 20, 2026
in News
China’s A.I. Ambitions Are Running Into a Money Problem

As the United States and China vie for artificial intelligence supremacy, Chinese tech companies are facing a mounting challenge: finding the money to keep up.

Like their American rivals, China’s A.I. start-ups face soaring costs to train increasingly powerful models. But while Anthropic raised more than $80 billion in a record-breaking initial public offering this week, its Chinese competitors face a far more unforgiving financial landscape.

Venture capital investment in China lags far behind that of the United States. Alibaba, Baidu and Tencent, three tech giants backing some of China’s leading A.I. companies, are grappling with negative cash flows. And investor enthusiasm for A.I. has soured. Since the end of July, China’s STAR 50 index, its version of the Nasdaq, has plunged 43 percent in a deeper sell-off than the recent downturn in American tech stocks.

China’s biggest tech companies are expected to double capital spending this year to $139 billion, even though the revenue from their A.I. labs covers less than 8 percent of that amount, according to Rhodium Group, a research firm. It forecasts that spending could top $193 billion in 2027.

“When you look at their numbers, it looks like they’re running on a treadmill that’s speeding up and getting harder to run,” said Kyle Chan, a fellow at the Brookings Institution, who specializes in China’s technology and industrial policies.

Much of the debate over China’s ability to compete in A.I. has focused on American export controls to restrict Chinese access to advanced semiconductors. Chinese A.I. labs have found costly workarounds, including renting data centers in Southeast Asia equipped with Nvidia chips. The Trump administration said this week it was moving forward with new rules to close that loophole.

But money might prove an equally formidable obstacle. Even if Chinese companies can find the chips they need, they must raise enough capital to buy them and finance the rising cost of developing frontier models.

Logan Wright, a partner at Rhodium Group who co-wrote a report on China’s A.I. financing, said those financial constraints “will be probably more decisive” than the chip shortage in the long term.

“The big problem is that the revenue mismatch with the Chinese models relative to capex is arguably more extreme than the U.S. case,” Mr. Wright said.

American A.I. companies spend roughly five to six times what their Chinese counterparts do, but bring in substantially more money. China’s six leading A.I. firms collectively generate about one-tenth the combined revenue of OpenAI and Anthropic, according to Epoch AI, a nonprofit research organization.

China’s embrace of open-source A.I. has made that imbalance even harder to overcome.

By allowing developers to use and modify their models freely, Chinese companies have accelerated the global spread of their technology. But that strategy has also undermined their ability to profit from it. Companies can charge for access, only to see rivals offer the same technology at lower prices. Fierce price wars have squeezed profits further.

Z.ai, a Chinese A.I. start-up, learned that lesson in August when it released its latest open-source model, GLM-5.3 Flash. The company halved prices for two weeks to entice users to run the model on Z.ai’s servers rather than their own. But competitors quickly moved in.

By November, roughly 20 rival providers were charging less than Z.ai for access to its own technology, according to Open Router, a platform connecting users with A.I. models.

“Your own pricing power is lost during this process,” said Cheryl Wu, a researcher at GovAI, a think tank.

“No one has figured out how to make money off consumers,” said Rui Ma, a technology investor, referring to Chinese A.I. start-ups.

The Cost of Compute

Another obstacle worsens China’s financing problems: access to advanced Nvidia chips, the gold standard for training the most powerful A.I. models.

From 2022 to June 2026, American companies possessed computing capacity equivalent to about 20 million Nvidia H100 chips, compared with roughly 2.6 million in China, some of it acquired through smuggling, according to EpochAI.

On Goofish, a Chinese secondhand online marketplace, sellers have advertised Nvidia’s DGX B300, an A.I. supercomputing server, for nearly $3 million — about eight times the U.S. price. Asking prices have reportedly quintupled since the start of the year, underscoring the premium Chinese companies must pay for advanced computing power.

The squeeze could worsen. Jensen Huang, who resigned as Nvidia’s chief last week amid an investigation into a romantic relationship with a Chinese businesswoman, was a powerful advocate of doing business with China. On Tuesday, Nvidia said it would stop shipping its chips to China, a deal that President Trump had previously permitted at Mr. Huang’s urging. The company also said it had established an export control oversight committee.

To reduce their dependence on American technology, Chinese developers are adapting their systems to run on chips made by Huawei, the Chinese telecommunications and electronics company leading the country’s chip charge.

Liang Wenfeng, chief executive of DeepSeek, the Chinese start-up that stunned Silicon Valley by developing a powerful A.I. model at a fraction of the expected cost, said Huawei’s technology is only about two years behind Nvidia’s.

But the gap remains consequential. DeepSeek would need four times as many Huawei 950 chips to match the performance of a single advanced Nvidia GB300 chip, Mr. Liang said, according to a leaked transcript of his call with investors, excerpts from which Chinese media outlets published. Huawei also lacks sufficient production capacity, he added.

“We simply can’t afford to train the largest models available right now,” he said. DeepSeek did not respond to a request for comment.

Yet China’s A.I. industry has repeatedly demonstrated an ability to do more with less. Chinese researchers have pioneered techniques to train and run powerful models more efficiently. Silicon Valley has accused Chinese firms of copying leading American technologies through a process called distillation, reducing the time needed to build foundational models with expensive computing infrastructure.

Worries of a Bubble

For China’s A.I. companies, finding investors willing to bankroll the competition is increasingly difficult.

DeepSeek, which has raised more than $20 billion, is preparing for an initial public offering in Shanghai early next year, according to Chinese media reports. Its investors include Tencent, the battery maker CATL and state-backed investment funds.

Few competitors have access to such deep pockets.

China’s venture capital industry has been shrinking since 2021 and though it has recovered slightly in the third quarter, it is still a fraction of its American counterpart. Banks are lending to China’s tech giants, but they are more cautious with smaller A.I. labs. Credit growth has fallen to historic lows, and the central bank governor has called for slower, “higher-quality” lending.

Foreign investment has also dried up. U.S. dollar-denominated investment funds in China have plunged partly because Washington has barred American venture capital from certain advanced Chinese technologies.

In 2025, private A.I. investment in China totaled $12.4 billion, compared with $280 billion in the United States, according to the Stanford A.I. Index.

Han Lin, the China country director for the Asia Group, an advisory firm, said China had exceptional engineering talent and a government capable of mobilizing enormous resources for strategic industries.

But he added, “What frontier technology also needs is harder to mobilize — risk capital that can absorb failure.”

Even the public markets, a lifeline for cash-hungry start-ups, are becoming less welcoming. Amid the sell-off, the Chinese authorities have instructed some companies to delay or suspend planned public listings, according to Cailian Press, a Chinese business publication, suggesting Beijing is concerned about an A.I. investment bubble.

China has announced investment funds to support A.I., but much of the money has gone toward semiconductors rather than companies developing large language models.

Government investors face many of the same constraints as private ones. Wang Renxuan, an assistant finance professor at the China Europe International Business School in Shanghai, said government funds also need a credible path to returns.

Erik Walenza, a Shanghai-based managing partner at Asia Growth Partners, a consultancy advising on A.I., said it was difficult to see how Chinese A.I. companies could make their finances work “without a lot of sustained government financing.”

The challenge, he added, is not only whether Chinese engineers can build models to rival America’s but also how much of the financial burden the government is prepared to shoulder.

“It becomes more of a political will question, of how much sacrifice are we willing to make to compete in this area,” Mr. Walenza said.

The post China’s A.I. Ambitions Are Running Into a Money Problem appeared first on New York Times.

GTA 6 was made by humans, but its publisher is all for using AI on admin
News

GTA 6 was made by humans, but its publisher is all for using AI on admin

by Business Insider
November 20, 2026

Strauss Zelnick is onboard with using AI for tasks like making pie charts. Steven Ferdman/Getty Images Take-Two Interactive CEO Strauss ...

Read more
News

Savings Bonds Offer Enticing Rates. You Need to Prove Your Identity First.

November 20, 2026
News

 How the Streamers Stack Up in Subscribers, Revenue and Profits | Analysis

November 20, 2026
News

New generations are taking over Congress. They expect transparency, not Schumer’s backroom deals

November 20, 2026
News

Is There Room for Ron DeSantis in a Post-Trump World? He Seems to Think So.

November 20, 2026
Dizzying link unearthed between Musk and ‘shamelessly’ nominated Trump judge: report

Dizzying link unearthed between Musk and ‘shamelessly’ nominated Trump judge: report

November 20, 2026
Prince Harry shares how Archie and Lilibet are coping with ‘culture shift’ move to UK — and daughter’s sweet accent change

Prince Harry shares how Archie and Lilibet are coping with ‘culture shift’ move to UK — and daughter’s sweet accent change

November 20, 2026
China’s A.I. Ambitions Are Running Into a Money Problem

China’s A.I. Ambitions Are Running Into a Money Problem

November 20, 2026

DNYUZ © 2026

No Result
View All Result

DNYUZ © 2026