Broadcom reported record revenue in the fourth quarter of fiscal year 2025, driven by soaring AI semiconductor demand, and issued an upbeat outlook for the first quarter of fiscal year 2026 with expectations that AI chip revenue will double year over year.

Broadcom is rapidly scaling its custom AI chip business, with CEO Hock Tan detailing a US$73 billion AI backlog, major new ASIC customers, and multi-billion-dollar XPU orders. The company expects to accelerate AI revenue through fiscal year 2026 as hyperscalers expand large-scale model training and inference infrastructure.

Oracle's fiscal second-quarter report for 2025 showed increased borrowing and shrinking free cash flow as the company expands from software into artificial intelligence (AI) hardware. The firm's remaining performance obligations (RPO) linked to AI surged to US$523 billion, marking a 438% year-over-year increase and prompting an increase in capex guidance from US$35 billion to US$50 billion for the year. Despite concerns about growing financial leverage, supply chain partners such as Foxconn and MiTAC Holdings maintain confidence, citing steady customer orders and capacity growth.

As the demand for cloud artificial intelligence (AI) infrastructure rapidly grows, multilayer ceramic capacitors (MLCCs) have emerged as one of the top cost contributors in AI server bills of materials (BOM), second only to graphics processing units (GPUs) and memory modules. This surge is driven by rising average selling prices and increased per-unit usage of MLCCs in next-generation AI server designs.

The global power equipment supply chain is facing a transformer shortage due to the unrelenting AI data center construction and demand for renewable energy. Delivery times for key equipment have been significantly extended, with some large transformers requiring as long as 24 months. This not only tests grid resilience but also creates challenges for major tech firms urgently seeking new power capacity.

he Malaysian government announced that companies engaged in midstream and downstream production of rare earth elements (REE) are eligible for related incentive policies. According to reports from Bernama and Nanyang Siang Pau, the Ministry of Investment, Trade and Industry (MITI) stated that under the 1986 Promotion of Investments Act, these REE midstream and downstream operators can apply for Pioneer Status, which grants tax exemptions on investments.

French artificial intelligence startup Mistral AI has launched two new open-source models and a command-line interface tool designed for software developers, expanding its focus beyond general language models. The offerings include the Devstral 2 series models, Devstral Small 2, and the Mistral Vibe CLI tool, aimed at the growing vibe coding market.

Advantech is accelerating deployment of its end-to-end intelligent transportation system (ITS) platform as global demand for smart traffic infrastructure rises. The company has expanded its footprint across South Korea through projects in Busan, Jeju, Incheon, Gwangju, Anyang, and Daegu, delivering systems for pedestrian–vehicle coordination, school zone safety alerts, and intersection conflict detection.

The Executive Yuan in Taiwan has proposed a NT$1.25 trillion (US$40 billion) special defense budget aimed at enhancing seven critical areas, including air defense, electronic warfare, and cybersecurity. A central focus of the "Taiwan Shield" initiative is to develop a multi-layered air defense system spanning high, medium, and low altitudes to improve national security.

Meta Platforms is reportedly delaying the release of its next-generation artificial intelligence (AI) model, Avocado, to the first quarter of 2026, according to sources cited by CNBC. The move accompanies a strategic shift from open-source to closed-source AI development and a major restructuring of Meta's AI division.

Chinese artificial intelligence startup DeepSeek is under scrutiny for allegedly circumventing US export restrictions by illegally acquiring thousands of Nvidia's Blackwell architecture GPUs. Nvidia acknowledged the situation but said it had no concrete evidence and would investigate all available leads.