Tencent President Says AI Spending May Ease After Infrastructure Buildout

Tencent President Martin Lau said the company’s heavy spending on artificial intelligence is partly a matter of building infrastructure that can be used for years. Speaking on a November 18 earnings call, he compared the investment to building a factory for model training and token production.

Lau expects Tencent’s financial position to improve markedly in 2028 as it continues using infrastructure bought earlier. He said equipment prices are high in 2026 and 2027 but expects comparable equipment to cost substantially less in 2028. He also expects gains in computing efficiency to increase the training and token output possible from the same facilities. These are management expectations, not reported results.

The scale of the spending was evident in Tencent’s third-quarter figures, reported on November 19, 2026. The company recorded net debt of 13.7 billion yuan, compared with net cash of 58.2 billion yuan in the second quarter. Tencent attributed the change chiefly to large computing-capacity purchases, partly offset by 8.4 billion yuan in net cash inflows as disposals from its investment portfolio exceeded new investments.

Third-quarter free cash flow was negative 69.1 billion yuan. Tencent reported 2.6 billion yuan in net cash from operating activities, more than offset by 65.1 billion yuan in capital-expenditure payments, 4.6 billion yuan in media-content payments and 2.0 billion yuan in lease-liability payments. Operating cash flow included substantial prepayments for computing capacity. Tencent said free cash flow would have been positive 16.9 billion yuan excluding those prepayments.

Capital expenditure reached 81.433 billion yuan in the quarter, up 527% from 12.983 billion yuan a year earlier. For the first three quarters of 2026, it totaled 166.153 billion yuan, up 179% from 59.566 billion yuan in the comparable period of 2025.

Lau said computing-capacity constraints were a short-term obstacle to model training for companies in China and could delay Tencent’s next model. He argued that domestic supply chains would mature and said efficient models such as Hy4 Flash could run entirely on domestically made chips. On returns from AI investment, he said Tencent first looks at whether it can build valuable products and businesses, with financial results following later. He acknowledged that its Yuanbao product had not achieved the success expected and said Tencent had substantially reduced spending on it to focus on capabilities. Lau gave no timetable for when AI products would generate profits.

An unnamed Tencent executive also attributed pressure on core-business profit to slower growth in fintech services, changes in marketing-services growth and higher component costs. The executive said DRAM prices had risen by about 400% year on year earlier in 2026. Although memory prices had since stabilized, the executive expected the unusually high costs to affect the core business over the next several quarters.

Further reading