China's AI chip industry is moving beyond a contest over process nodes and into a broader race involving memory, advanced packaging, chip interconnects and system architecture.
The 2026 World Artificial Intelligence Conference (WAIC) opens in Shanghai on July 17, with Chinese President Xi Jinping set to attend and deliver a keynote speech.
AI technologies and applications have matured rapidly in recent years. That has driven trends such as edge AI and physical AI. The shift has also affected industrial computer (IPC) makers, which have long focused on end-point applications on the factory floor and at other field sites.
CXMT's STAR Market IPO has become more than a fundraising exercise. The strategic placement roster shows how China's largest DRAM maker is using the capital market to connect semiconductor suppliers, AI cloud providers, device brands, automakers, and state-backed investors, reinforcing a domestic memory ecosystem.
Reports of longer chip lead times have been mounting. ADI has notified customers that, as recovering demand tightens supply, lead times for some analog chip products have stretched to six months, and it has urged customers to place orders early to avoid delivery delays. Meanwhile, a channel player said STMicroelectronics (ST) MCU lead times have extended to 52 weeks, prompting distributors to begin asking customers about demand for all of 2027.
Generative AI is accelerating demand for computing power, memory and data bandwidth, shifting semiconductor innovation beyond front-end processes toward advanced packaging, silicon photonics (SiPh) and Co-Packaged Optics (CPO). Benjamin Hein, CEO Electronics and Executive Board member at Merck, said advanced packaging materials are poised to outgrow front-end process materials and the broader materials market, while Near-Packaged Optics (NPO) will bridge the industry's transition to commercial CPO.
South Korean President Lee Jae-myung unveiled the country's Three Mega Projects for AI and Semiconductors in late June 2026, an ambitious national strategy designed to strengthen South Korea's global leadership in artificial intelligence and semiconductors. The initiative centers on three pillars—semiconductors, physical AI, and AI data centers—and aims to double the nation's DRAM output within five years while expanding capabilities in high-bandwidth memory (HBM), advanced packaging, AI processors, and next-generation memory technologies. It also seeks to extend South Korea's semiconductor footprint beyond the Seoul metropolitan region.
At a humanoid robotics summit in Tokyo in May 2026, I saw a consulting firm's global labor automation map for Physical AI. After returning, I recreated the same map using the firm's research on Digital AI job functions. Placing the two side by side revealed something unexpected.
In the second quarter of 2025, DIGITIMES visited the rapidly expanding Johor-Singapore Special Economic Zone (JS-SEZ) and Wiwynn's massive AI server system integration (SI) factory. Microsoft also granted a media interview, using the occasion to discuss Singapore's role as the hub of the broader ASEAN AI ecosystem.
US President Donald Trump recently claimed that Taiwan's TSMC will double the size of its Arizona fab project, reviving attention on his goal of raising the US share of the global chip market to 50% before the end of his term. TSMC declined to comment on the report, but investors may press the company on the issue at its second-quarter 2026 earnings call.
The AI race is expanding from computing power to data transmission, making optical interconnects a critical battleground for next-generation AI infrastructure.
Reports that Meta is considering leasing out idle AI computing capacity have rattled investors. But treating Meta's predicament as a warning sign for the entire AI industry is a classic case of overgeneralization.
As the world enters an AI-centric era, the global race for technological leadership is no longer defined only by who can build the most advanced models. It is increasingly shaped by who can secure compute, deploy infrastructure at scale, reduce energy constraints, and turn research into commercial capability.
The world's most powerful AI models are encountering a new constraint beyond chips, data and engineering talent: governments increasingly want a say in when frontier systems are released, who may access them and which capabilities should remain restricted.
