Tencent has reportedly signed its largest overseas cloud leasing agreement with Oracle, securing access to about 100,000 advanced artificial intelligence chips through data centers in Southeast Asia as Chinese technology companies seek additional computing capacity outside the country.
The five-year agreement, reported by the Financial Times on Wednesday, citing people familiar with the matter, is estimated to be worth about $7 billion and would involve multiple Oracle data centers across Southeast Asia. Tencent is also expected to make an upfront payment of about 30%, according to the report.
If confirmed, the scale of the arrangement would highlight the extraordinary computing requirements emerging from China’s AI race, as Tencent and other major technology companies invest heavily in training and deploying more capable models.
Register for the next Tekedia Mini-MBA.
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
More importantly, the reported structure illustrates how Chinese technology companies are seeking access to advanced AI computing capacity through infrastructure outside mainland China as U.S. restrictions limit the availability of leading AI processors inside China.
But Tencent’s reported decision to lease computing capacity in Southeast Asia rather than simply expand its domestic infrastructure points to a broader challenge facing China’s AI industry.
The United States has imposed export controls restricting China’s access to some of the most advanced AI chips, particularly processors from Nvidia and other U.S. suppliers. Beijing has responded by accelerating development of domestic alternatives, while Chinese companies have simultaneously sought other ways to obtain computing capacity.
A five-year Oracle agreement worth roughly $7 billion would mark a substantial commitment to overseas AI infrastructure. The reported 100,000 chips would give Tencent access to a large pool of advanced computing resources that are unavailable to it domestically.
The arrangement would also shift part of Tencent’s AI infrastructure footprint outside China.
This is considered a huge shift because, among other things, training large language models requires enormous amounts of computing power, while running AI products at scale also requires sustained access to inference capacity. As models become more capable and AI applications gain users, the computational burden increasingly extends beyond the initial training phase.
Tencent has been expanding its AI ambitions across consumer and enterprise products, increasing the need for both model-training infrastructure and computing capacity for commercial deployment.
The company recently released a preview version of a new AI image-generation model aimed at professional creators. The model supports text-to-image and image-to-image generation, adding to Tencent’s growing portfolio of AI applications.
The Economics of The Chip Squeeze
The reported deal also illustrates how export restrictions can change the economics of AI development.
For Chinese companies, access to cutting-edge processors is not simply a question of purchasing chips. Computing capacity can be obtained through cloud providers that operate data centers in jurisdictions where certain processors can legally be deployed.
Tencent would reportedly be leasing computing capacity from Oracle rather than importing the processors directly into China. The structure could allow the company to use advanced chips without those processors being physically deployed inside mainland China.
The reported arrangement thus underpins the growing importance of cloud infrastructure as an intermediary between semiconductor restrictions and AI development. It also reveals why Washington has increasingly focused not only on direct chip exports but on the possibility that restricted Chinese companies could obtain access to advanced computing remotely through overseas data centers.
For cloud providers, meanwhile, the surge in AI demand creates an enormous infrastructure opportunity. Oracle has been expanding its cloud capacity to serve AI developers and has signed large computing agreements with technology companies seeking access to scarce advanced processors.
Tencent’s reported commitment is expected to add another major customer to that trend.
China’s AI Race is Becoming An Infrastructure Race
The deal comes as Chinese technology companies compete to develop more capable AI models while Beijing encourages greater reliance on domestic technology.
China’s strategy is producing two parallel efforts.
One is the development of domestic processors and software ecosystems that can reduce reliance on Nvidia and other foreign suppliers. The other is securing access to advanced computing resources wherever they remain available.
Tencent is one of China’s largest technology companies and has substantial financial and engineering resources to devote to AI. Its willingness to reportedly commit billions of dollars to overseas cloud capacity underscores how valuable advanced computing has become.
The scale of the reported transaction also puts the economics of the AI boom into perspective. A $7 billion commitment over five years would amount to roughly $1.4 billion a year, before accounting for other infrastructure, energy, networking, personnel and model-development expenses.
That spending underlines why AI is becoming an increasingly capital-intensive business even for companies that already possess extensive cloud and data-center infrastructure.
Nevertheless, the investment is expected to provide Tencent with the computing capacity needed to accelerate model development and support large-scale deployment. For Oracle, a deal of this size would reinforce the importance of cloud infrastructure providers in the global AI supply chain.
But the reported agreement also underscores the limits of China’s current domestic chip ecosystem. If Tencent needs to secure tens of thousands of advanced processors through overseas infrastructure, it suggests that domestic alternatives have not yet fully eliminated the computing gap created by U.S. restrictions.
China’s technology industry is therefore pursuing two tracks simultaneously: developing its own AI hardware and software while finding ways to access global computing capacity. The Tencent-Oracle agreement, if confirmed, would be one of the clearest examples yet of how those two pressures are reshaping the geography of AI infrastructure.