Below are the most-read DIGITIMES Asia stories from the week of April 13-19, 2026:
Samsung launches voluntary retirement amid chip-device earnings gap
Samsung Electronics has launched a broad voluntary retirement program as it seeks to manage rising costs and structural imbalances between its highly profitable semiconductor division and its underperforming device business, according to South Korean media reports. The initiative, rolled out in phases and targeting older employees across divisions, offers enhanced compensation packages and reflects efforts to reduce labor expenses amid weaker margins in smartphones, TVs, and appliances, where higher memory costs and intensifying competition have weighed on profitability.
While the semiconductor unit continues to drive the bulk of earnings on strong AI-related demand, the Device eXperience division lags significantly, prompting workforce adjustments, internal reassignments, and limited hiring. The restructuring also aligns with anticipated policy changes in South Korea that could extend the retirement age, potentially increasing long-term labor costs, leading Samsung to take preemptive steps to streamline its workforce and improve operational efficiency.
Qualcomm moves into custom DRAM with CXMT for smartphones
Qualcomm is reportedly exploring a partnership with Changxin Memory Technologies (CXMT) to co-develop custom DRAM for smartphones, as tightening memory supply and rising costs force changes across the mobile ecosystem. With DRAM production increasingly prioritized for AI-driven high-bandwidth memory (HBM), conventional mobile DRAM has become more expensive, squeezing margins in mid-range and entry-level devices where memory accounts for a significant share of total costs.
The potential collaboration would allow Qualcomm to optimize memory architecture, stabilize supply, and better control costs, particularly for Chinese smartphone makers, while also aligning with China's broader push to localize its semiconductor supply chain. The move reflects a wider industry shift toward customized, vertically integrated solutions as memory constraints reshape smartphone design and production strategies.
CPU shortage more acute than memory; industry awaits Intel 18A yield improvement
A global CPU shortage is severely disrupting PC and industrial computing supply chains, with processors from both Intel and Advanced Micro Devices (AMD) reportedly unavailable even at premium prices, while memory remains scarce but still purchasable. The crunch has hit notebook and industrial PC makers particularly hard, with Intel's dominant presence in industrial segments amplifying the impact.
Legacy chips such as Raptor Lake are especially difficult to source, prompting a shift toward newer generations like Arrow Lake, though higher costs remain a concern. While both chipmakers have raised prices, availability has become the key constraint, and relief is expected to depend largely on improved yields from Intel's 18A process. Industry observers warn that supply tightness could persist into 2026, potentially dampening consumer demand and forcing manufacturers to manage inventories cautiously amid ongoing uncertainty.
CCL price hikes to extend through 2026 with rising glass fiber and copper foil costs
Rising demand for AI servers and networking equipment is driving a sharp increase in high-end PCB production, creating supply bottlenecks and pushing up costs for key upstream materials like copper-clad laminate (CCL). Shortages of fiberglass cloth and a widening gap in HVLP copper foil supply are fueling sustained price hikes, with some materials expected to see annual increases of 20-30% or more through 2026.
Major manufacturers, including Panasonic and Taiwan Union Technology, have already announced significant price increases, passing rising costs down the supply chain. As material constraints intensify and cumulative price adjustments build, PCB makers are raising quotes for customers, highlighting how AI-driven demand is reshaping cost structures across the electronics manufacturing ecosystem.
TSMC hints at next-gen LPU bid, stokes speculation that Samsung's Groq order may not last
TSMC chairman C.C. Wei revealed the company is collaborating with a customer on next-generation LPU development, a move widely interpreted as targeting business currently held by Samsung and potentially linked to Nvidia. Although the customer was not named, the timing suggests TSMC is positioning itself to capture future inference chip production.
Analysts note that Samsung's current role stems from an earlier partnership with Groq prior to Nvidia's involvement, whereas Nvidia's broader ecosystem already relies heavily on TSMC's more advanced process nodes and packaging capabilities. As a result, any next-generation LPU designed under Nvidia's direction could naturally shift to TSMC, signaling intensifying competition between the two foundries and raising doubts about the longevity of Samsung's foothold in AI inference manufacturing.
TSMC enters 2nm mass production, scales 3nm capacity
TSMC outlined an aggressive roadmap for advanced-node expansion at its first quarter 2026 earnings call, with Wei emphasizing Taiwan as the core hub for cutting-edge production. The company has already begun mass production of its 2nm process, driven by strong smartphone and AI demand, while expanding 3nm capacity globally across Taiwan, the US, and Japan to support applications ranging from AI to automotive and IoT.
At the same time, TSMC is shifting its mature-node strategy toward higher-value specialty technologies while reallocating resources to advanced nodes. Looking ahead, its next-generation A14 process is on track for 2028 mass production, underscoring TSMC's focus on sustaining technology leadership amid surging demand for advanced semiconductors.
Global chip equipment sales reach record US$135 billion as AI drives investment surge
Global semiconductor equipment spending reached a record US$135.1 billion in 2025, up 15% year-over-year, according to SEMI, as strong AI-driven demand fueled investment across advanced logic, memory, and packaging technologies. Growth was particularly robust in back-end segments, with test equipment surging 55% and packaging tools rising 21%, reflecting the increasing complexity of AI chips and HBM.
Regionally, Asia further consolidated its dominance while Europe and North America saw declines after prior investment cycles. The data underscores how AI is accelerating both capacity expansion and technological upgrades across the semiconductor manufacturing ecosystem.
Article edited by Jack Wu
