Taiwan's semiconductor manufacturers continue to confront challenging market conditions for mature process technologies, with reports that TSMC may slow expansion plans at its Kumamoto facility. Industry sources indicate that ASE Technology and Intel have also delayed capacity growth in Malaysia, despite maintaining public optimism about future demand.

Samsung Electronics Chairman Lee Jae-yong recently made a high-profile visit to China, meeting with leading Chinese automaker executives and President Xi Jinping. South Korean analysts suggest the trip was more than diplomatic optics—it reflects Samsung's bid to capitalize on China's surging artificial intelligence (AI) demand and revive its faltering semiconductor division.

Hiroshi Kawamoto, Division Officer, Finance Division of Tokyo Electron Limited (TEL), one of Japan's leading semiconductor equipment manufacturers, projects that demand for high-bandwidth memory (HBM) equipment—vital for powering generative AI—will continue to grow. However, he anticipates that sales of front-end semiconductor manufacturing processes in 2025 will likely remain on par with 2024 levels.

SEMICON China 2025 opened at the Shanghai New International Expo Centre to large crowds and packed booths, a full display of China's forward momentum for its semiconductor industry. Beneath the prestige, however, domestic chip equipment makers are embroiled in a brutal price war, with some reportedly cutting prices to just 40% of list value in a scramble for market share.

A senior executive from Kaynes Semicon revealed plans for the company to become the first in India to package chips by July, noting the delay in Micron's ATMP project.

Between the 1960s and 1980s, the semiconductor industry primarily used a vertically integrated model covering all aspects, from equipment to materials. This approach disadvantaged latecomers such as Taiwan and South Korea, which concentrated on process technology rather than developing their own equipment and material capabilities.

Amid the AI boom, much of the spotlight has focused on Chinese chips like Huawei's Ascend. Yet, general-purpose computing power, still dominated by CPUs, remains a critical foundation. Most enterprise servers continue to adopt a "CPU plus AI accelerator" setup. Even Nvidia, known for its GPUs, is developing its own Arm-based Grace CPU to meet this demand.

President Donald Trump signed an executive order to establish an office within the Commerce Department to facilitate and speed up investments of more than US$1 billion in the US and administer federal funds from the Chips and Science Act.

Nvidia's relationship with Chinese AI startup DeepSeek has evolved dramatically, affecting Nvidia's stock price and driving strong demand for its export-controlled H20 GPU in China. Since late January 2025, major Chinese technology companies have been aggressively purchasing servers to deploy DeepSeek AI models, creating unprecedented demand for Nvidia's China-specific H20 chips.

On March 31, at his first public appearance as CEO during the Intel Vision conference in Las Vegas, Lip-bu Tan outlined his strategy for Intel, which includes spinning off non-core assets, enhancing AI-related products, and collaborating with the US government to attract 2 to 3 key customers for its foundry business.

TSMC officially marked its entry into the 2-nanometer technology era with a ceremony at its Kaohsiung Fab 22 P2 plant. During the event, Executive VP and Co-Chief Operating Officer Y.P. Chyn revealed plans for three additional plants beyond the two already in operation.