Chinese tech companies Tencent and Baidu have outlined their strategies to remain competitive in the global AI sector amid tighter United States export controls on advanced semiconductors.

During a recent earnings call, Tencent’s president, Martin Lau, reported that the company has amassed a substantial inventory of graphics processing units (GPUs).

Lau said that Tencent is optimizing its software and exploring smaller AI models to lessen its dependence on extensive computing power.

Baidu, which operates a search engine and offers cloud services, stressed its “full-stack” AI capabilities in its earnings call.

The strategies of both companies highlight a focus on efficiency and domestic innovation to address challenges arising from limited access to advanced US technology.

🔗 Source: CNBC

🧠 Food for thought

1️⃣ China’s semiconductor dependency drives strategic adaptation

China’s tech giants are responding to a long-standing national vulnerability in semiconductor independence.

In 2018, China imported over $300 billion worth of computer chips, exceeding its crude oil imports, revealing the scale of its technology dependency on foreign suppliers 1.

Tencent’s approach of stockpiling GPUs while simultaneously developing more efficient AI models demonstrates how Chinese companies are creating short-term buffers while working on long-term solutions to reduce reliance on restricted technologies.

The strategy of optimizing software to maximize performance from existing hardware resources represents a pragmatic response to supply constraints, potentially accelerating innovation in efficiency-focused AI development.

Baidu’s emphasis on “full-stack” capabilities follows the same pattern, developing integrated technology stacks where they control more components to mitigate vulnerability to future supply disruptions.

2️⃣ Software optimization emerges as competitive differentiator under constraints

The emphasis both Tencent and Baidu place on software optimization reveals how export controls are reshaping competitive advantages in AI development.

Tencent’s Martin Lau explicitly challenges the Western assumption that expanding GPU clusters is necessary for AI advancement, stating they can achieve “good training results with a smaller group of such chips” through more efficient approaches.

Recent Tencent developments