Yazılar

Malaysia to Tighten Semiconductor Regulations Amid U.S. Pressure

Malaysia plans to impose stricter regulations on the movement of semiconductors, particularly those from Nvidia, as part of efforts to curb the flow of advanced chips to China under U.S. pressure. The United States has expressed concerns over the potential diversion of these critical chips to China, where they could be used in the development of artificial intelligence (AI) technologies.

Trade Minister Zafrul Aziz revealed that the U.S. government has asked Malaysia to monitor shipments of high-end Nvidia chips and ensure that they are not diverted to unauthorized destinations, particularly China. The U.S. is concerned that servers containing these chips may end up in Chinese data centers instead of the intended locations, and is pushing Malaysia to track every shipment of Nvidia products entering the country.

Malaysia’s investigation into the situation also ties into a broader inquiry regarding a fraudulent transaction case in Singapore, involving the illicit shipment of U.S. servers to Malaysia. These servers may have contained advanced chips covered by U.S. export controls. The case, which involves Singapore-based firms accused of supplying these servers fraudulently, is valued at $390 million. There are concerns that the shipments may have been intended for Chinese AI company DeepSeek, which gained attention for its AI model performance earlier this year.

The U.S. government is also probing whether DeepSeek has been using banned U.S. chips, as part of a wider investigation into the potential violations of export controls on semiconductor technologies.

Tencent Unveils T1 Reasoning Model Amid Intense AI Competition in China

Tencent, the Chinese tech giant, has officially launched its T1 reasoning model, marking a significant step in the intensifying competition in China’s artificial intelligence (AI) sector. The official version of T1, announced on Friday night, offers faster response times and enhanced capabilities for handling extended text documents. The company highlighted that the model’s content logic remains clear, and the text is neat and clean, with a notably low hallucination rate.

This release comes at a time when China’s AI landscape is becoming increasingly competitive, particularly after DeepSeek introduced models that are said to offer comparable or even superior performance to Western counterparts at much lower costs. Tencent had previously released a preview version of T1 via platforms such as its AI assistant app, Yuanbao, but the official version is now powered by Tencent’s Turbo S foundational language model, which was unveiled late last month.

According to a comparison chart shared in the announcement, Tencent’s T1 model outperformed DeepSeek’s R1 model on certain knowledge and reasoning benchmarks. This new development is part of Tencent’s broader push to accelerate its AI investments, which include plans to significantly increase capital expenditures in 2025 following a strong focus on AI spending throughout 2024.

Beijing Supports AI Startup Manus in Bid for Global AI Dominance

Chinese AI startup Manus has made significant strides, with its China-facing AI assistant now officially registered and receiving notable state media attention, as Beijing continues to promote domestic AI companies. The startup, which recently garnered global attention for releasing what it claims is the world’s first general AI agent capable of making decisions and executing tasks autonomously, is being positioned as a key player in China’s ambition to rival global AI leaders.

Manus’ breakthrough moment came when the company went viral on social media platform X, following the introduction of its AI agent, which offers a more advanced and independent functionality compared to current AI chatbots like ChatGPT and the AI model DeepSeek. Beijing’s state-run CCTV aired a segment showcasing Manus, highlighting the AI agent’s unique capabilities, and comparing it to DeepSeek’s AI chatbot, which also gained recognition for offering competitive performance at a fraction of the cost of its U.S. counterparts.

The Chinese government has supported Manus’ development, with Beijing’s municipal government approving the registration of Manus’ earlier AI assistant, Monica, which is a necessary step for launching generative AI apps in China. This regulatory approval aligns with Beijing’s strategy of bolstering the domestic AI sector while maintaining tight control over content deemed sensitive by the authorities.

In addition to government backing, Manus secured a strategic partnership with the team behind Alibaba’s Qwen AI models, further strengthening its position in the competitive AI landscape. Manus’ AI agent is currently available through an invite-only system, with a waitlist reportedly exceeding 2 million users.