• Tencent AI operations remain unaffected by US GPU restrictions, claims sufficient chip inventory.
• Software optimisation signals reduction in American semiconductor dependency.

Chinese internet conglomerate Tencent has dismissed concerns over US GPU export restrictions, with executives stating the company’s AI infrastructure possesses adequate processing power for ongoing operations. This comes as Washington and Beijing continue negotiations over semiconductor trade policies.

During Tencent’s Q2 2024 earnings call, company president Martin Lau addressed questions about potential consequences of recent US decisions that allow Nvidia and AMD to resume limited GPU sales to China. Lau’s response suggested the policy changes hold minimal significance for the company’s immediate operations.

“We do not have a definite answer on the import situation yet. There are a lot of discussions between the two governments,” Lau stated, according to The Register. However, he emphasised the company’s current position: “From our perspective, we do have enough chips for training and continuous upgrade of our existing models. We also have many options for inference chips.”

Strategic shift toward software optimisation

Tencent’s approach extends beyond developing stockpiles. “We are executing a lot of software improvements to drive efficiency in inference so we can put more workloads on the same number of chips,” Lau said during the earnings call.

The optimisation-first strategy represents a shift in how Chinese technology companies respond to ongoing US export controls. Rather than scrambling for alternative hardware sources, Tencent appears to be focusing on extracting maximum performance from current resources.

Implications for US semiconductor giants

The company’s self-sufficiency claims present challenges for American chip manufacturers hoping to capitalise on renewed access to Chinese markets. Both Nvidia and AMD had anticipated revenue increases from resuming sales to China, particularly given the massive scale of Chinese AI development projects.

Lau’s remarks suggest Tencent may not contribute to the expected purchasing surge, potentially dampening revenue projections for US semiconductor companies. The situation becomes more complex when considering the Trump administration’s reported plans to claim a percentage of GPU sales to China.

Moreover, Tencent has indicated flexibility in sourcing inference chips from non-US suppliers, suggesting the company has developed alternative supply chains that could reduce American market share in Chinese AI infrastructure.

Financial performance amid AI investment concerns

Despite robust financial results, Tencent’s earnings call revealed challenges it’s facing recouping AI investments. The company reported Q2 revenue of RMB 184.5 billion (US$25.7 billion), representing 15% annual growth, while net profit reached RMB 64.8 billion (US$9 billion), up 11%.

However, Lau acknowledged persistent difficulties in balancing AI infrastructure costs with revenue generation. “Depreciation costs from AI will continue to go up,” he noted. “But we continue to reap the benefits of AI. The issue is these two may not match each other completely, but both are moving in the same general direction.”

The comments suggest Tencent, like many technology companies globally, continues struggling to translate substantial AI investments into proportional revenue streams.

Diversification beyond GPU-dependent services

Tencent’s cloud division has adapted its strategy to reduce dependence on GPU availability, according to Lau’s statements. The company is actively pursuing opportunities in CPU-based computing and database services, areas less affected by current US export restrictions.

“Our cloud strategy is not dependent on GPU,” Lau emphasised, adding, “We are also growing in CPU and database.” The tendency to diversification allows Tencent to maintain growth in cloud services regardless of semiconductor supply chain disruptions, and potentially reduce exposure to future trade policy changes.

Consistent messaging on hardware independence

This is the third consecutive quarter where Tencent has communicated to investors that additional GPU purchases are unnecessary for current operations, which suggests the company’s position on hardware sufficiency is not reactive to recent US policy developments.

The company’s core social media platforms continue showing strong user engagement, with numbers of Weixin and WeChat monthly active users reaching 1.411 billion, an annual growth of 40 million users or a 3% increase.

As US-China technology trade relationships remain in flux, Tencent’s stated hardware independence and optimisation focus may serve as a template for other Chinese technology companies navigating similar export control challenges. The long-term effectiveness of its approach to maintain competitive AI capabilities will likely influence industry-wide strategies in China’s technology sector.

Want to learn more about cybersecurity and the cloud from industry leaders? Check out Cyber Security & Cloud Expo taking place in Amsterdam, California, and London.

Explore other upcoming enterprise technology events and webinars powered by TechForge here.