Last year, the titans of China’s technology industry received a rare invitation to meet with Xi Jinping, the country’s leader, in the ceremonial Great Hall of the People.
But when the entrepreneurs — including the heads of national stars like the telecommunications giant Huawei and rising companies like the robotics start-up Unitree — assembled in neat rows before Mr. Xi, an unexpected figure was among them: Jack Ma.
Mr. Ma, who founded the Alibaba e-commerce empire in a cramped Hangzhou apartment and became a rock star of China’s internet era, had all but vanished from public view since Beijing torpedoed the planned $34 billion public offering of one of his companies after he criticized regulators in 2020.
Now, however, Alibaba has re-emerged as an A.I. powerhouse, creating one of the world’s most widely used A.I. systems.
Alibaba made its most popular A.I. models open source, allowing others to use and modify them freely. That made its technology much cheaper to use than proprietary systems from U.S. competitors like Anthropic and OpenAI, helping the company attract users around the world.
But it also raised a difficult question — how to turn that global popularity into a profitable business. It’s a challenge that many companies with open-source technologies face and one that is dividing Alibaba’s A.I. team.
Alibaba became one of China’s most valuable companies on the strength of its globe-spanning e-commerce business. But in recent years, it has been outmaneuvered by lower-cost rivals at home and has struggled to expand overseas. After Mr. Ma faded from public view, a new generation of executives took over, adopting a lower profile while emphasizing Alibaba’s role within Beijing’s broader technology policies.
“Alibaba is quite aware of its position as part of this team of national champions that are leading China’s charge in A.I.,” said Kyle Chan, a fellow at the Brookings Institution. “But also they have this huge responsibility to not repeat what they did before.”
In hindsight, Alibaba had gradually laid the groundwork to become an A.I. heavyweight. As its online shopping and logistics businesses expanded, the company built data centers capable of processing vast volumes of customer data worldwide.
It also borrowed a page from Amazon, parlaying its e-commerce success and technological infrastructure into a major cloud computing business, giving Alibaba two essential ingredients for building artificial intelligence systems: data and computing power.
“Alibaba in many ways has been ahead of its time,” Mr. Chan said. “It made early bets in a lot of these areas that are now the hot topics.”
Alibaba introduced its family of A.I. models, Qwen, in 2023 and quickly made them open source. When DeepSeek, an A.I. start-up, burst into the spotlight with claims that it had built a powerful model for a small fraction of the cost of Western competitors, global investors rushed to capitalize on the excitement around China’s open-source A.I.
China has made open-source technology a pillar of its drive to become an A.I. superpower. Nearly all of its leading A.I. systems are open source.
By January, Qwen had become the world’s most downloaded open-source A.I. system. Its models were being downloaded about one million times a day, according to data from Hugging Face, which hosts many open-source A.I. projects.
But that reach has not translated into big money. In the first three months of this year, Alibaba reported $1.3 billion in revenue from A.I.-related products — less than 4 percent of its total revenue. That pales in comparison with the company’s plan to spend more than $55 billion by the end of next year to build out its A.I. infrastructure.
The push to turn its open-source success into a profitable business has quietly fractured the team behind Qwen.
In March, Lin Junyang, Qwen’s lead engineer, announced that he was leaving Alibaba. Several other key engineers left around the same time. Two people familiar with the team said it had become divided by disagreements over how best to commercialize Qwen.
Alibaba had long kept its largest, most advanced models proprietary, while releasing leading open-source models alongside them. Now, the company is signaling a broader shift away from widely used open-source models and toward closed ones that customers must pay to use. In April alone, Alibaba released three proprietary models within days of one another.
The Qwen team remains focused on keeping pace with the leading models coming out of Silicon Valley, but there is a growing recognition that technological leadership will not be enough if the company doesn’t make money, said a member of Alibaba’s A.I. research lab, which includes the Qwen team and others, who spoke on the condition of anonymity because the person was not authorized to discuss internal matters.
The challenge for Alibaba is apparent in its stock price. While A.I.-related stocks have surged in markets worldwide, Alibaba’s stock has fallen 37 percent this year in Hong Kong, where the broader market has declined 12 percent.
Building a top-performing A.I. model is expensive, requiring enormous investments in hangar-size data centers filled with computer chips that consume copious amounts of electricity.
Chinese A.I. companies face an additional hurdle: U.S. export controls limit their access to the most advanced chips. China’s leading A.I. start-ups and researchers routinely say shortages of computing power are the biggest thing holding them back, and they are spending heavily to secure it.
Those exorbitant costs are heaping financial pressure on Chinese A.I. companies. Start-ups like MiniMax and Z.ai have gone public to raise money from investors, and Alibaba is increasingly steering customers toward proprietary models even if that strategy upsets some of its top talent.
“It is a tough choice to make, and one that every open-model lab has to make at some point,” said Kevin Xu, the founder of Interconnected Capital, a hedge fund that invests in A.I. technologies.
Alibaba’s A.I. business faces mounting external challenges. The Pentagon recently placed the company on a blacklist of firms that it says support the Chinese military, a designation that Alibaba argues has already harmed its business. In a statement, Alibaba said it was not affiliated with the Chinese military.
Anthropic and OpenAI have also accused Chinese companies, including Alibaba, of improperly harvesting data from their A.I. systems to accelerate the development of their models.
Last month, Anthropic sent a letter to Senators Tim Scott, Republican of South Carolina, and Elizabeth Warren, Democrat of Massachusetts, accusing Alibaba of “brazenly” and “illicitly” trying to copy its technology using 24,000 fraudulent accounts. Alibaba declined to comment on the allegations.
Many in the technology industry compare today’s A.I. boom to the early days of the dot-com era, when internet companies were flourishing but no one had yet figured out how to build durable businesses around the technology.
Richard Lin, a vice president at the Silicon Valley company Datastrato who has long been involved in China’s open-source community, said China’s leading A.I. companies were all fighting for survival.
“There isn’t an A.I. company with a sustainable business model right now,” he said. “It’s not a healthy industry.”
