China's outsourced semiconductor assembly and test (OSAT) sector is accelerating capacity expansion and technology upgrades, with leading players ramping investment to capture rising demand from AI, high-performance computing (HPC), and automotive electronics.
MetaOptics, a Singapore-based metalens developer, is partnering with Taiwan's Pin-Jye Nano Technologies to scale production of next-generation optical components as the technology moves from research into early commercialization and gains new relevance in artificial intelligence and advanced sensing applications.
In response to strong demand for data security and on-premises AI deployment within South Korea's manufacturing and semiconductor sectors, Taiwanese AI company Spingence Technology is aggressively expanding into the South Korean market starting this year through its enterprise edge large language model (LLM) platform, Edgestar.
Elon Musk's SpaceX is facing production challenges at its newly built fan-out panel-level packaging (FOPLP) and printed circuit board (PCB) factories in Texas, delaying full-scale manufacturing until mid-2027. Industry sources say that despite nearly complete equipment installation, yields remain below expectations, prompting a postponement of mass production schedules.
As HPC and AI processors push computing performance to unprecedented levels, transistor density has reached a point where thermal behavior is no longer uniform. Instead of gradual, evenly distributed heating, modern chips exhibit sharp, localized hotspots that concentrate extreme thermal loads within small regions.
A newly expanded collaboration between Intel and Google signals a key shift in AI infrastructure: CPUs are back at the center of the conversation. Both companies emphasized that the partnership spans multiple generations of Intel's Xeon processors and includes co-development of custom infrastructure silicon. Financial terms and deployment timelines were not disclosed, but the scope points to a long-term alignment, not a one-off supply deal.
Amazon's custom chip business has surpassed an annual revenue run rate of more than US$20 billion, CEO Andy Jassy said in his annual shareholder letter, according to The Information. The figure reflects the rapid adoption of Amazon Web Services' (AWS) in-house silicon portfolio, including Graviton CPUs and Trainium AI accelerators, which are currently sold exclusively through AWS cloud services.
In recent years, MediaTek has steadily increased its influence in the global Wi-Fi chip market. Its market share gains have been driven not only by continued growth on the consumer side, but also by a strategic push into the telecom operator segment — where the company is securing fresh business across Europe and North America.
Global advanced packaging capacity is currently in severe shortage. Nvidia has already reserved most of TSMC's leading-edge capacity, particularly its CoWoS packaging technology. According to TSMC's disclosures in an interview with CNBC, demand for such advanced packaging is growing at a compound annual growth rate (CAGR) of 80%.
Anthropic is exploring the possibility of designing its own AI chips, though the effort remains at a very early stage, Reuters reported, citing three sources familiar with the matter. "The plans are in early stages, and the company may still decide only to buy AI chips and not design any," Reuters reported, adding that the startup has not yet committed to a specific architecture or assembled a dedicated semiconductor team.
As global corporations accelerate spending on artificial intelligence (AI) infrastructure, supply constraints are no longer limited to memory chips. Signs of tightening availability are now emerging in multi-layer ceramic capacitors (MLCCs), small but essential components used across a vast range of electronic systems. Lead times for these parts are lengthening across the industry, according to market data.
More coverage
