This week in China’s artificial intelligence sector highlights a landscape defined by rapid technical achievements, significant regulatory hurdles, and intensifying geopolitical friction. DeepSeek continues to disrupt the global frontier model market with extreme pricing and performance, while Baidu faces massive domestic regulatory roadblocks in the autonomous driving sector. Meanwhile, U.S.-China technology tensions have reached new heights over export controls and intellectual property disputes.

Here is your professional analysis of the major developments for the week ending June 27, 2026.

DeepSeek: Continued Disruption and New Paradigms

DeepSeek remains the most aggressive disruptor in the AI ecosystem following the late-April release of its DeepSeek V4 and V4-Pro models.

The V4-Pro, boasting 1.6 trillion total parameters (with 49B active), continues to achieve parity with or surpass closed-source frontier models in math, STEM, and agentic coding benchmarks. The accompanying V4-Flash model (284B total / 13B active) provides an ultra-fast and cost-effective inference alternative.

Notably, DeepSeek has established a 1 million token context window as the default baseline across its official services, powered by DeepSeek Sparse Attention (DSA) and advanced token-wise compression. To compound the pressure on global competitors like OpenAI (GPT-5.5), Anthropic (Claude 4.7), and Google (Gemini 3.1 Pro), DeepSeek executed a massive 75% price cut on its V4-Pro first-party API. As older models (DeepSeek-chat and DeepSeek-reasoner) are slated for permanent retirement by July 24, 2026, the company is forcing a rapid industry-wide transition to its highly compatible, low-cost API infrastructure.

Baidu: Autonomous Driving Setbacks and Global Pivots

Baidu’s heavy investment in integrating its Ernie 5.0 foundation model into physical robotics and autonomous vehicles hit a significant regulatory wall this week.

In a massive blow to the domestic robotaxi industry, China has suspended the issuance of new autonomous driving permits. This regulatory freeze follows a widespread outage in March 2026 that stranded dozens of Baidu’s Apollo Go robotaxis in Wuhan. The suspension effectively halts domestic expansion for Baidu and its competitors.

Facing severe roadblocks at home, Baidu is executing an aggressive pivot toward international markets. The company is leaning into global deployment partnerships—including previously announced collaborations with Uber and Lyft—and actively targeting expansions in Dubai, Germany, and the UK.

Alibaba and Tencent: IP Controversies and Ecosystem Integration

Alibaba (Qwen)

Alibaba is prioritizing massive integration and enterprise cloud dominance, but it is currently embroiled in severe international intellectual property allegations. In June 2026, Anthropic formally accused Alibaba’s Qwen lab of conducting a massive “distillation campaign,” allegedly utilizing nearly 25,000 fake accounts to siphon Claude’s advanced capabilities into Qwen’s training pipelines. The fallout resulted in a notable stock drop and heightened scrutiny in Washington.

Simultaneously, Alibaba is aggressively targeting Western developers by slashing prices by up to 80% on its Qoder coding platform during U.S. working hours. On the hardware front, Alibaba Cloud introduced the Qwen-Robot embodied AI model, signaling a strategic pivot toward full-stack physical AI and robotics.

Tencent (Hunyuan)

Tencent continues to focus on 3D generation, efficiency, and deep integration into its expansive WeChat/WeCom ecosystem. The global launch of the Hunyuan 3D Creation Engine allows developers to generate high-quality 3D assets from text, images, or sketches in minutes. Open-source versions have already surpassed 3 million downloads, with over 150 enterprise clients adopting the platform via Tencent Cloud.

Additionally, Tencent is rolling out its AI workplace assistant, “Dayuan,” inside WeCom, powered by the highly efficient Hunyuan-A13B model.

Geopolitical Friction and Escalating Export Controls

The geopolitical landscape surrounding AI hardware and software has grown increasingly hostile, with major policy moves from both Washington and Beijing.

U.S. Export Controls

The U.S. Department of Commerce officially closed a critical loophole that allowed Chinese companies to procure advanced AI hardware through international subsidiaries. The new directives mandate export licenses for all high-end chips destined for Chinese and Macau-based companies, regardless of whether the transaction is routed through a third-country affiliate. Furthermore, a bipartisan U.S. congressional coalition introduced legislation aimed at preventing Chinese firms from training frontier models via Western cloud providers (AWS, Azure, GCP).

China’s Retaliation and Domestic Governance

In direct retaliation, China’s Ministry of Commerce added approximately 10 U.S. firms (including key rare earth miners) to its export control list on June 22, 2026, restricting their operations within mainland China.

Domestically, China is moving to tighten AI governance. The deployment of the “Artificial Intelligence Agent Interconnection” system will create a unified identity management framework for AI agents to track activity and mitigate enterprise deployment risks. Regulators also rolled out strict new AI guidelines for the banking and insurance sectors, focusing on compliance, transparency, and data localization, as Premier Li Qiang emphasized the necessity of state oversight to prevent “losing control” over frontier technology.

This weekly briefing provides a snapshot of the fast-paced developments in China’s AI sector, tracking the intersection of technological breakthroughs, regulatory frameworks, and global geopolitical dynamics.