Tencent and Alibaba Accelerate AI Spending as Global Cloud Giants Face Rising Costs

China’s technology leaders are stepping up investment in artificial intelligence infrastructure, signaling a new phase in the global race for computing power. Tencent reported first‑quarter 2026 revenue of RMB 196.46 billion, up 9% year‑on‑year, with capital expenditure rising 16% to RMB 31.94 billion. Executives said AI‑related spending will increase further this year, particularly in the second half.

On the same day, Alibaba CEO Wu Yongming declared that “almost no card in Alibaba’s servers is idle,” underscoring the intensity of demand. He argued that heavy investment in AI data centers will deliver certain returns over the next three to five years, and suggested capital expenditure could exceed the previously announced RMB 380 billion. He likened AI development to manufacturing, requiring two factories — one for training and one for inference — both built on massive data‑center infrastructure.

The announcements triggered a rally in U.S.‑listed Chinese stocks. 21Vianet surged more than 25%, Kingsoft Cloud rose over 17%, Alibaba and Meis Intelligent gained more than 8%, while Baidu and JD.com climbed more than 7%.

Tencent acknowledged that its early AI capabilities were not strong, but said it has been strengthening talent, management, and training to close gaps and is now entering a growth trajectory. Wu Yongming had already warned last November that even the fastest supply‑chain and server deployment could not meet demand, suggesting the RMB 380 billion figure was too conservative.

Rising hardware costs are adding pressure across the industry. ByteDance announced plans to spend more than RMB 200 billion this year, 25% above its initial plan, due to higher memory‑chip prices. Microsoft said USD 25 billion of its USD 190 billion capital expenditure is earmarked to offset component cost increases.

The surge is driven by AI Agents, AI programming, and other applications that are dramatically increasing token usage. Baidu founder Robin Li noted that token consumption has become the closest thing to an industry consensus metric, though it measures cost rather than revenue. Morgan Stanley reported that soaring token demand — up 350% — has led to a sharp revision of hyperscale cloud providers’ 2026 capital‑expenditure forecast, from USD 450 billion to USD 800 billion.

Chinese brokerages see the trend as reinforcing. China Galaxy Securities said North American CSPs are raising spending forecasts to build AI infrastructure, with the arms race entering a white‑hot phase and downstream demand remaining strong. Huayuan Securities highlighted that Q1 earnings from North American CSPs showed rapid growth in inference demand and large backlogs of expansion orders, confirming long‑term demand and easing ROI concerns. Huaxi Securities added that short‑term AI industry trends are likely to remain upward, with hardware leading the cycle. The market has shifted from broad gains led by cloud providers and chips to a phase of diffusion and differentiation driven by bottleneck segments such as storage and optical modules. Pricing now focuses less on sheer capital‑expenditure expansion and more on order certainty, profit realization, cash‑flow pressure, and investment returns.