Highlights
U.S. to limit chips to Malaysia and Thailand. Bloomberg has reportedly seen a draft Commerce Dept rule that would restrict advanced chip sales to Malaysia and Thailand. This comes as part of a broader effort to contain China’s access to AI chips. The Trump administration scrapped the Biden-era AI Diffusion Rule, which split every country into three tiers of AI chip access. This is the beginning of the Trump admin’s strategy on AI chips and China.
Samsung strategy shifts. Most worryingly, the Korean company expects a 39% decline in Q2 operating profit, following their delays in HBM sales to Nvidia and the like. This is especially significant in a memory landscape where SK Hynix is tearing away at the lead and even historical number 3, Micron, is catching up.
For logic, Samsung has shifted their strategy to double down on 2nm and 4nm in, instead of the latest edge 1.4nm. Their 1.4nm, which was scheduled to launch in 2027, is now slated for 2029. This is unlike Intel, which has reportedly prioritised 14A over 18A. Moreover, Samsung aims to discount their 4nm, 5nm, 7nm chips by 30%, to better compete with market leader TSMC.
TSMC prioritises US fabs. TSMC, meanwhile, will double down on their US fab construction at the expense of their Japan fabs. The Taiwanese foundry is eager to expedite US-based production with significant tariffs expected on the import of chips.
Thanks for reading.
1. Policy and Geopolitics
1.1
Bloomberg (07/04): US Plans AI Chip Curbs on Malaysia, Thailand Over China Concerns
President Donald Trump’s administration plans to restrict shipments of AI chips from the likes of Nvidia Corp. to Malaysia and Thailand, part of an effort to crack down on suspected semiconductor smuggling into China.
A draft rule from the Commerce Department seeks to prevent China — to which the US has effectively banned sales of Nvidia’s advanced AI processors — from obtaining those components through intermediaries in the two Southeast Asian nations, according to people familiar with the matter. The rule is not yet finalized and could still change, said the people, who requested anonymity to discuss private conversations.
Officials plan to pair the Malaysia and Thailand controls with a formal rescission of global curbs from the so-called AI diffusion rule, the people said. That framework from the end of President Joe Biden’s term drew objections from US allies and tech companies, including Nvidia.
2. Economy, Finance, and Business
2.1
WSJ (07/04): TSMC to Delay Japan Chip Plant and Prioritize U.S. to Avoid Trump Tariffs
Taiwan Semiconductor Manufacturing is delaying construction of a second plant in Japan partly because it is pouring funds more quickly into U.S. expansion ahead of potential Trump administration tariffs, people familiar with the plans said.
The revised schedule is the latest example of how President Trump’s aggressive stance on trade is pulling some investment toward the U.S. at the expense of allies. Major technology companies have committed to expand U.S. production of artificial-intelligence servers that are currently made in places such as Mexico and Taiwan.
2.2
Reuters (07/07): Samsung Electronics second-quarter profit likely to drop 39% on weak AI chip sales
Samsung Electronics is expected to forecast a 39% plunge in second-quarter operating profit on Tuesday, weighed down by delays in supplying advanced memory chips to artificial intelligence chip leader Nvidia.
The world’s biggest maker of memory chips is projected to report an April-June operating profit of 6.3 trillion won ($4.62 billion), its lowest income in six quarters, according to LSEG SmartEStimate.
The prolonged weakness in its financial performance has deepened investor concerns over the South Korean tech giant's ability to catch up with smaller rivals in developing high-bandwidth memory (HBM) chips used in artificial intelligence data centres.
2.3
TrendForce (07/07): Samsung Reportedly Bets on 4–7nm with 30% Price Gap over TSMC, Eyeing Markets China Hasn’t Entered
While Intel has shifted focus to 14A from 18A in a strategic trade-off, Samsung has reportedly made its compromises by prioritizing on 2nm and 4nm instead of 1.4nm, as per ZDNet. Meanwhile, Chosun Biz reveals that the struggling semiconductor giant also plans to boost demand in sub-7nm processes—still beyond the reach of Chinese rivals—by pricing the nodes about 30% below TSMC’s.
Chosun Biz suggests that for its 4nm process, Samsung aims to win orders by enhancing power efficiency by about 20% with the SF4U. According to Samsung, SF4U is a premium 4nm variant using optical shrink technology to enhance PPA (power, performance and area), with mass production planned for 2025.
Although Samsung’s 4nm, 5nm, and 7nm nodes trail TSMC in chip performance, the company competes by pricing them about 30% lower and maintaining yields above 70% amid fierce price battles, the report notes. According to an industry insider cited by Chosun Biz, Samsung’s 4nm and 5nm processes have improved enough to rival TSMC, giving it a key advantage as Chinese competitors have yet to develop these nodes.
2.4
Chosun Daily (07/04): Samsung delays 1.4nm node, doubles down on 2nm process enhancement
Samsung Electronics’ foundry division is struggling to secure large orders from major tech companies for its advanced sub-3 nanometer chip processes. While yield rates for the 2nm and 3nm nodes both exceed 40%, making them sufficient for commercial production, overall chip performance including power efficiency is reportedly falling short compared to competitors such as TSMC. Samsung has reportedly been conducting evaluations of its 2nm process with Nvidia and Qualcomm.
Samsung had planned to launch its 1.4nm process in 2027, but the timeline is now expected to slip to 2029. In the meantime, the company is refining its second-generation 2nm process (SF2P) and aims to begin mass production of a third-generation version (SF2P+) next year, with over 20 percent improvement in performance. Unlike TSMC, which has secured major orders from Nvidia, AMD, Apple, and Qualcomm, Samsung currently relies on its System LSI division and a few domestic and overseas startups for foundry demand.
3. Technology
3.1
TrendForce (07/07): UMC Reportedly Lands Qualcomm Deal for Advanced Packaging, Mass Production in 1Q26
As UMC looks beyond mature nodes to drive growth, Economic Daily News suggests that the Taiwanese foundry has landed a major advanced packaging order from Qualcomm. Its self-developed high-end interposer has passed Qualcomm’s verification and is now nearing mass production, the report adds.
UMC’s first batch of interposers—tailored to meet Qualcomm’s IC and memory capacitance needs—has passed electrical testing and entered trial production, according to supply chain sources cited by Economic Daily News. Featuring 1500nF/mm² capacitors, the interposers are reportedly expected to enter mass production as early as Q1 2026.
-