As AI computing power surges exponentially, electricity has become a critical strategic resource for the tech industry. With traditional power supply models struggling to meet the rapidly growing demand from data centers, US tech giants are transforming from mere consumers into active investors and traders in energy.
China's leading DRAM maker CXMT has unveiled new DDR5 and LPDDR5X chips with performance now viewed as comparable to Samsung Electronics and SK Hynix. The launch has unsettled South Korea's memory sector, especially since CXMT reached this level in under a year after pivoting from low-cost DRAM to premium development in early 2025.
As the surge in AI compute drives a shift in data-center power design, backup battery units (BBUs) have secured adoption from at least three cloud service providers, boosting their presence in high-end AI servers and prompting interest in electric double-layer capacitor (EDLC) solutions within passive components.
The rapid expansion of artificial intelligence (AI) workloads is forcing a fundamental redesign of data center infrastructure, driving historic memory shortages and opening a rare market window for alternative memory technologies, according to Chuck Sobey, founder of ChannelScience and general chair of the Chiplet Summit.
3Peak, a leading Chinese analog chipmaker, plans to acquire Ningbo Aura Semiconductor through new share issuance and/or cash. The move advances 3Peak's push to become a global platform-based provider of analog and mixed-signal solutions.
To enhance artificial intelligence (AI) performance, Meta and Nvidia are advancing plans to embed GPU compute cores directly into the base die of high-bandwidth memory (HBM). This innovation blurs the lines between memory and system semiconductors, presenting new opportunities and challenges for South Korea's semiconductor industry.
Competing with leading players, Cheng Mei Instrument Technology recently delivered the first CoPoS measurement and inspection equipment to the ASE Group, aiming to break KLA Corporation's near-monopoly in the equipment market. Cheng Mei Chairman Tony Tsai stated that in recent years, the company has been aggressively entering the advanced packaging field. He estimates that in 2025, measurement and inspection equipment in the semiconductor sector will account for roughly 60–70% of revenue, with 2026 expected to exceed 70%.
Rapidus plans to begin construction on a second fabrication plant in Chitose, Hokkaido, in fiscal 2027 to produce 1.4nm and eventually 1nm chips, according to reports from Nikkei and Yomiuri Shimbun. The move underscores Japan's push to return to the front of advanced logic manufacturing, even as Rapidus works to complete its 2nm process and secure the funding required to sustain its roadmap. However, the company has since denied this reported timeline.
Google's expanding Tensor Processing Unit (TPU) strategy is emerging as a serious challenge to Nvidia's long-running dominance in AI accelerators, particularly after a report from The Information revealed that Meta is in talks to begin using Google TPUs in its data centers in 2027 under a potential multibillion-dollar agreement.
Topco Scientific (TSC) expects semiconductor material demand to remain robust through 2026, driven by AI-led advanced process capacity utilization and strong memory market momentum. The company is optimistic about next year's performance in key product lines, including photoresist, silicon wafers, and quartz, with photoresist showing the strongest growth potential amid booming advanced node requirements.
Recent market rumors indicate that Meta is negotiating to deploy Google's seventh-generation TPU "Ironwood" in its data centers by 2027, with procurement potentially reaching tens of billions of US dollars. This move signals confidence that Google will become a leading ASIC chip supplier, challenging Nvidia's dominance and securing the second spot in the AI accelerator market.
More coverage
