AI Infrastructure and Compute Power Breakthroughs
Shenzhen has achieved a significant breakthrough in high-quality AI development. On March 26, the 11,000P intelligent computing cluster, invested and constructed by the Shenzhen municipal government, was officially activated. Including the 3,000P cluster launched previously, the city's total intelligent computing power has now reached 14,000P. This marks the nation's first 10,000-card level, fully autonomous, and domestically controlled AI computing cluster built entirely with advanced domestic chips, signifying a crucial step in China's self-reliance in AI computing infrastructure.
Tesla also formally launched its TERAFAB project, which aims to achieve an annual compute output exceeding 1 terawatt (1 TW). This initiative is specifically designed to produce high-performance chips for Tesla's humanoid robots, anticipating a potential annual production of 1 to 10 billion units as projected by Elon Musk, which will necessitate a substantial increase in chip demand.
AI Security and Large Model Market Performance
In the realm of AI security, Ant Group's AI Security Lab recently conducted a specialized security audit of the open-source autonomous agent framework, OpenClaw. Over a three-day intensive review, the team submitted 33 security vulnerability reports. OpenClaw has since confirmed and patched eight of these vulnerabilities in its latest version, 2026.3.28, including one critical, four high-severity, and three medium-severity flaws. Ant Group affirmed its commitment to continuously monitor OpenClaw's security risks, supporting the industry in safely and robustly adopting AI agents.
The domestic large model sector has also seen notable success. Just one month after the release of the Kimi K2.5 model, Moonshot AI's annualized recurring revenue (ARR) has surpassed $100 million. According to insiders, following the K2.5 model's launch, API supply for Tokens Per Minute (TPM) quotas became rapidly constrained. Some clients even offered multi-million dollar consumption commitments and upfront guarantees to secure priority supply, underscoring the strong market demand for high-quality large model computing power.
Xiaomi Group is also intensifying its efforts in AI. Lei Jun announced the official launch of a special AI talent recruitment program, revealing that Xiaomi's R&D and capital expenditures in AI this year amount to 16 billion yuan. The company has made substantial progress in foundational large models, embodied AI robots, and mobile AI projects.
Storage Technology and Expanding AI Applications
Samsung Electronics continues to advance its NAND flash memory technology. The company has completed the first phase of process upgrades at its Xi'an factory, phasing out traditional 128-layer (V6) NAND flash and commencing mass production of 236-layer (V8) products. Samsung plans further upgrades to transition to 286-layer (V9) NAND flash within this year, which will provide more efficient and higher-density solutions crucial for AI data storage.
iQIYI has also officially opened its professional film and television production AI agent platform, "Nadou Pro," for external use, entering its pre-commercialization phase. This development signals a deepening application of AI agents in content creation and industrialized film and television production workflows.
Robotics and Data Governance Frontiers
Linghou Robotics announced the completion of its Series B funding round, raising hundreds of millions of RMB. Investors include notable institutions such as SMIC Juyuan, Houwei Capital, Sequoia China, Guotai Junan Innovation Investment, and Hefei Construction Investment Fund, with existing shareholders like Shanghai Zhiyuan and Boyuan Capital also increasing their investments. This reflects the capital market's continued optimism for intelligent robotics, particularly in areas integrating with AI.
Furthermore, the World Data Organization (WDO) was established in Beijing on March 30. As a professional, non-governmental, and non-profit international entity voluntarily formed by data-related organizations and individuals globally, the WDO aims to "bridge the data divide, unleash data value, and prosper the digital economy." The organization will be dedicated to promoting global data cooperation and governance practices, actively exploring efficient communication and rational utilization pathways for data on a compliant, secure, and trustworthy basis, providing solid support for global digital economic development.