The first week of March 2026 delivered a rapid-fire mix of chip industry drama, market-shaping AI demand, and fresh policy pressure on China-linked semiconductors. From a high-profile investigation involving advanced process technology to looming memory price hikes that could ripple through PCs, smartphones, servers, and data centers, these were the stories drawing the most attention across Asia’s semiconductor supply chain from March 2 to March 8, 2026.
Former TSMC executive investigated over alleged disclosure of cutting-edge chip technology
A major development is unfolding around Wei-Jen Lo, formerly a senior vice president of corporate strategy development at TSMC. Lo retired in 2025 and later accepted an invitation from Intel to take an executive vice president role. Authorities have now formally opened an investigation and completed evidence collection tied to allegations that sensitive information was leaked.
Taiwan’s National Science and Technology Council (NSTC) Minister Cheng-Wen Wu said that after the incident, prosecutors worked with the Hsinchu Science Park Bureau to confirm whether the sub-2nm process technology involved qualifies as national core key technology. The case is closely watched because sub-2nm manufacturing is among the most strategically valuable capabilities in global semiconductors, influencing everything from next-generation AI accelerators to advanced mobile processors.
TSMC maintains foundry leadership as four China-based companies reach the global top 10
New industry figures show just how quickly foundry revenue has expanded. According to ChipInsights, combined revenue for 29 pure-play foundries worldwide hit CNY1.149 trillion in 2025, rising 25.46% year over year and surpassing CNY1 trillion for the first time. The rankings also spotlight the increasingly competitive field, with TSMC continuing to dominate while four Chinese players reportedly moved into the top 10, reflecting growing capacity and momentum in China’s semiconductor manufacturing sector.
Samsung highlights faster 2nm yield progress and pushes HBM growth plans
Samsung Electronics signaled stronger-than-expected progress in its advanced manufacturing roadmap, saying its 2nm process yield is improving faster than anticipated. That kind of yield improvement matters because it can translate into better cost efficiency, higher output, and faster ramps for next-generation chips.
The company also pointed to upcoming wafer production plans at its Taylor, Texas foundry, while reiterating confidence in high-bandwidth memory (HBM). Samsung’s stated goal to triple HBM revenue underscores how central AI infrastructure has become to memory strategy, as HBM is increasingly viewed as a must-have component for training and running large-scale AI models.
AI demand stays red-hot as trade tensions strengthen Nvidia and TSMC’s strategic position
Despite geopolitical uncertainty and tightening export controls, AI computing demand continues to show remarkable resilience. Nvidia posted another earnings beat for the fourth quarter of its fiscal 2026, reinforcing the idea that AI infrastructure spending remains a priority across major markets.
For the full fiscal year, Nvidia delivered more than US$120 billion in profit and reported a gross margin of 71.1%. Those numbers highlight the extraordinary profitability of AI hardware at scale, even as trade pressures intensified during the first year of US President Donald Trump’s administration. In this environment, companies controlling critical AI compute platforms and leading-edge manufacturing capacity stand to gain more pricing power and long-term strategic influence.
US moves to block federal purchasing of certain China-linked chips
Policy is also changing the competitive landscape. The US is moving to bar federal agencies from buying specific semiconductors linked to major China-based chipmakers, expanding procurement restrictions. The timing is significant: these tighter rules arrive as the electronics supply chain is already feeling strain from memory shortages and rising prices, adding another layer of complexity for procurement teams and system builders trying to secure components.
DRAM price increases loom as Samsung and SK Hynix reportedly begin customer negotiations
Memory pricing is back in the spotlight. Samsung Electronics and SK Hynix have reportedly notified customers about planned DRAM price increases for the second quarter of 2026 and have already started negotiations. Sources suggest the shift is being driven by surging AI-related demand and evolving purchasing practices that increasingly favor the biggest buyers with the strongest volume commitments.
For many companies, DRAM is an unavoidable cost input across product lines, from consumer devices to servers. That means price hikes can pressure margins, raise end-user prices, or force product configuration changes.
DRAM prices could jump as much as 70% in 2Q26 as structural shortages deepen
One of the most eye-catching projections from industry sources is that DRAM prices may surge up to 70% in the second quarter of 2026. The key takeaway is that AI-driven data center expansion is pushing the memory market beyond traditional supply-and-demand patterns, creating what sources describe as a structural shortage.
A widening gap between spot and contract DRAM prices—estimated at 40–50%—is increasing pressure on memory suppliers to raise contract prices in multiple rounds to narrow the spread. With AI events and industry momentum continuing to energize the memory market, the near-term outlook points to sustained tightness rather than a quick reversal.
