Daily: Tariff rollercoaster and semiconductors; Trump bans EDA to China; Nvidia earnings beat expectations
9 min read.
Thank you for your patience as I returned from Jeju this week.
Highlights
Tariff rollercoaster and semiconductors. As I’m sure you all know, the Court of International Trade ruled that Trump’s ‘Liberation Day’ tariffs, which include tariffs on China, as well as levies on Canada and Mexico, exceeded the established authority of the President. A quick note on the semiconductor angle:
First, (some) semiconductors were exempt under the original tariff decisions.
Second, there is an ongoing Section 232 investigation into semiconductors and other electronics. Under the court ruling, Section 232 tariffs, like the 25% tariffs on steel, aluminium, and automobiles, remain valid. This is because those tariffs are grounded in a different legal framework that grants the President the requisite authority when national security is at risk. So, we will likely see a (probably) legally valid tariff on semiconductors in the coming month or two.
Trump bans EDA to China. Trump orders U.S. EDA companies to stop selling to China, as Trump’s chip restrictions on China grows tighter. At least Synopsys has already begun complying with the rule. The U.S. is leveraging its chokepoints in the semiconductor value chain - the vast majority of EDA software and advanced AI chips come out of the United States. China has domestic EDA companies like Empyrean Technology, but cutting off EDA will be tough on China.
Nvidia earnings beat expectations. Nvidia earnings revenue jumps 69& to US$44.1 billion, beating investor expectations. Net income rose 26% to US$18.78 billion, a little shy of market predictions.
Thanks for reading.
1. Policy and Geopolitics
1.1
FT (05/30): Donald Trump orders US chip software suppliers to stop selling to China
The Trump administration has told US companies that offer software used to design semiconductors to stop selling their services to Chinese groups, in the latest attempt to make it harder for China to develop advanced chips.
Several people familiar with the move said the US Department of Commerce had told so-called electronic design automation groups — which include Cadence, Synopsys and Siemens EDA — to stop supplying their technology to China.
The Bureau of Industry and Security, the arm of the US commerce department that oversees export controls, issued the directive to the companies via letters, according to the people.
1.2
WSJ (05/29): Senators Bash Nvidia’s Plans for Facility in China
Nvidia is facing fresh accusations from U.S. lawmakers that the company is too close to China, criticism that could signal new challenges for the richly valued chip maker.
Recent plans for an Nvidia facility in Shanghai risk giving China access to cutting-edge technology, Sen. Jim Banks (R., Ind.) and Sen. Elizabeth Warren (D., Mass.) wrote in a recent letter to Chief Executive Jensen Huang, a copy of which was viewed by The Wall Street Journal.
The facility “raises significant national security and economic security issues that warrant serious review,” they said in the letter. The senators demanded that Nvidia provide a comprehensive timeline and description of its plans for the facility, including details on specific research and engineering projects and any financial incentives.
1.3
FT (05/30): Chinese tech groups prepare for AI future without Nvidia
China’s biggest technology companies have begun the long and difficult process of switching their development of artificial intelligence to homegrown chips, as they contend with a dwindling stockpile of Nvidia processors and tightening US export controls.
Alibaba, Tencent and Baidu are among those starting to test alternative semiconductors to meet surging AI-related internal demand and client needs, according to industry executives.
They have been forced to step up contingency planning as growing US-China trade tensions led Donald Trump’s administration last month to clamp down on sales of their preferred chip: Nvidia’s H20, a watered-down product tailored to comply with Joe Biden-era curbs.
1.4
Nikkei (05/31): Huawei invested in 60-plus China chip firms since US sanctions
Huawei Technologies has bought into more than 60 Chinese chip-related companies since it became subject to U.S. sanctions in 2019, Nikkei has found, as it works to overcome a lack of access to Western technology with a homegrown supply chain.
Huawei has been expanding investments in the companies through Hubble, a wholly owned investment company it set up in 2019 when Washington began to restrict the Chinese telecom leader's access to U.S. technology and its market.
1.5
FT (06/01): Can the Gulf really become an AI superpower?
Abu Dhabi has announced a gargantuan data centre cluster for OpenAI and other US companies as part of their “Stargate” project. The Emirate, which manages $1.7tn in sovereign wealth funds, is investing billions through AI fund MGX, and its Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) is opening a centre in Silicon Valley.
Gulf states “have the capital, the energy, the political will”, said Sam Winter-Levy, fellow at the Carnegie Endowment for International Peace. “The only thing these countries didn’t have were chips and talent. Now [after Trump’s visit] they may have chips.”
The region’s vast AI ambitions could still face challenges, experts warn. Both countries lack the skilled workforces of Silicon Valley or Shanghai, and research output has trailed other nations.
Meanwhile, human rights groups have raised ethics concerns over autocratic monarchies’ use of surveillance, while US experts have cautioned about leaks of American technology to China.
2. Economy, Finance, and Business
2.1
NYT (05/28): A.I. Chipmaker Nvidia’s Revenue Jumps 69% to $44.1 Billion
Nvidia’s business prospects have been whipsawed by the U.S. government lately. Last month, the government blocked the sale of artificial intelligence chips to China. Weeks later, it approved the sale of similar chips to the Middle East.
Amid the turmoil, Nvidia still maintained its breakneck growth as the leading provider of the computer chips used for building artificial intelligence.
Nvidia said on Wednesday that sales in its most recent quarter rose 69 percent from a year earlier to $44.1 billion. Its net income rose 26 percent to $18.78 billion.
The company exceeded Wall Street’s expectations for sales of $43.28 billion, but fell short of predictions for a profit of $19.49 billion.
Nvidia also projected that revenue in the current quarter would rise 50 percent from a year ago to $45 billion, as it expands sales of its newest A.I. chip, Blackwell. The sales forecast is in line with Wall Street’s prediction of $45.75 billion, suggesting that the tech industry’s embrace of artificial intelligence is in its early stages, with ample room to run.
2.2
WSJ (05/28): TSMC to Set Up Chip Design Hub in Germany
Taiwan Semiconductor Manufacturing Co. said it would establish a new chip-design center in Munich, Germany, a win for the European Union as Brussels seeks to become more self-sufficient in the production of semiconductors.
TSMC, the world’s largest contract chip maker, said the facility would design high-performance and energy-efficient chips to be used across the automotive and industrial sectors as well as artificial intelligence.
A company spokesperson said Munich was selected for its proximity to European clients. The hub, expected to open in the third quarter this year, will add to TSMC’s global network of nine design centers spanning Taiwan, mainland China, Japan, Canada and the U.S.
2.3
Bloomberg (05/31): TSMC Evaluates Building Advanced Chip Plant in the UAE
Taiwan Semiconductor Manufacturing Co. is evaluating building an advanced production facility in the United Arab Emirates and has discussed the possibility with officials in President Donald Trump’s administration, according to people familiar with the matter, a potentially major bet on the Middle East that would only come to fruition with Washington’s approval.
The company has had multiple meetings in recent months with Steve Witkoff, the US special envoy to the Middle East, and officials from MGX, an influential investment vehicle overseen by the UAE president’s brother, the people said. Those conversations are a continuation of talks that began under President Joe Biden’s administration but had died down by the end of his term.
The project being discussed is a substantial investment in what’s called a gigafab — a complex of six factories similar to what TSMC is building in Arizona. The total cost of such a facility in the UAE is unclear. TSMC plans to spend $165 billion on its Phoenix project, which also includes research and packaging facilities.
2.4
Reuters (05/30): Marvell forecasts second-quarter revenue above estimates on strong demand for custom AI chips
Marvell Technology forecast second-quarter revenue above Wall Street estimates on Thursday, betting on strong demand for its custom chips powering artificial intelligence workloads in data centers.
Demand for custom AI chips continues to fuel growth, while networking chips and electro-optics have also seen robust order momentum. These advancements help hyperscalers seeking to scale their infrastructure to support AI workloads.
3. Technology
3.1
TrendForce (05/29): Samsung’s 3nm Yield Reportedly Stuck at 50%, Far Behind TSMC’s 90%
While Samsung is gaining traction at mature nodes like 7nm and 8nm—with orders reportedly from Nintendo—it continues to struggle at the advanced 3nm level. According to South Korean media outlet Chosun Biz, even after three years of mass production, its 3nm yields remain at just 50%.
This has made it harder for Samsung to win big tech’s trust, with Chosun Biz reporting that Google’s Tensor G5 is shifting to TSMC’s 3nm, moving away from Samsung. As highlighted by 9to5Google, the search engine giant has reportedly locked in Tensor chip production with TSMC for the next 3 to 5 years, covering Pixel models through at least the Pixel 14.
Chosun Biz reports that Google isn’t alone—Qualcomm and AMD are also leaning on TSMC, drawn by its solid 90%+ yield at 3nm. The report notes that Apple (M5), Qualcomm (Snapdragon 8 Elite Gen 2), NVIDIA (Rubin), and MediaTek (Dimensity 9500) are expected to use TSMC’s third-generation 3nm (N3P) process, with plans to shift to 2nm starting in 2026.
3.2
Nikkei (05/31): SoftBank, Intel work on AI memory chips that use half as much power
SoftBank and Intel are developing a type of memory for artificial intelligence expected to consume much less electricity than current chips, helping to build efficient AI infrastructure in Japan.
The companies plan to develop a structure for stacked DRAM chips that uses a different wiring structure than current advanced high-bandwidth memory, slashing power consumption by roughly half.
3.3
TrendForce (06/02): TSMC’s 2nm Wafers Rumored to Soar to $30K Per Unit, Yet CSP Giants Reportedly Rush to Adopt by 2027
As TSMC prepares for 2nm mass production in 2H25, wafer prices have reportedly hit $30,000 per unit for 2nm and could soar to $45,000 at future nodes, according to Commercial Times. Despite the steep cost, major CSPs are set to follow AMD, NVIDIA, and Broadcom in adopting the node in the next 2 years, the report adds.
Commercial Times, citing supply chain sources, suggests that developing a single 2nm chip—from project kickoff to final output—costs a staggering $725 million. Still, top players are diving in—AMD’s next-gen EPYC “Venice” was the first HPC chip taped out on TSMC’s N2 in April, with MediaTek set to follow in September (likely Dimensity 9600), as noted by the report.
3.4
Bloomberg (05/31): Amazon Pledges to Continue Aggressive Data Center Expansion
Amazon.com Inc.’s cloud services chief says the company is aggressively expanding server farms globally and looking to boost access to the latest artificial intelligence chips from Nvidia Corp.
Amazon Web Services, the world’s largest seller of rented computing power and data storage, opened a cluster of data centers in Mexico earlier this year, AWS Chief Executive Officer Matt Garman said Friday in an interview with Bloomberg Television. The cloud division is also building out new facilities in Chile, New Zealand, Saudi Arabia and Taiwan, Garman said.
3.5
Bloomberg (05/28): Japan Backs AI Chip Startup EdgeCortix in Boost to Defense Tech
EdgeCortix Inc., a Tokyo-based AI chip startup, is riding a wave of interest to foster Japanese semiconductors with defense applications.
EdgeCortix, which has won a contract tied to the US Department of Defense, on Wednesday secured government subsidies of ¥3 billion ($21 million) to develop energy-efficient AI chiplets for commercialization in 2027. The contract may help revenue more than double this year, founder Sakyasingha Dasgupta said.
The products, designed to help robots make real-time decisions and fill the country’s labor shortages, target mass production at Taiwan Semiconductor Manufacturing Co.’s plant in Japan. The subsidies are on top of ¥4 billion in support the semiconductor designer won in November to make chips for next-generation communication systems.
3.6
Bloomberg (05/28): Qualcomm Releases Study Showing That Its Modems Beat Apple’s C1
A study commissioned by Qualcomm Inc. found that its modem chips worked better than a rival component developed by Apple Inc., especially for cellular customers in dense urban areas.
The iPhone 16e, the first smartphone to use Apple’s in-house C1 modem, was slower to download and upload information on T-Mobile’s 5G network in New York City than Qualcomm-powered Android devices, according to a report by Cellular Insights Inc. Qualcomm paid for the study and made the findings available to Bloomberg.
-