Daily: Jensen in China; No TSMC/Intel JV; Intel CEO's new management
5 min read.
Highlights
Jensen in China. Nvidia CEO Jensen Huang went to Beijing to meet the Vice Premier, the DeepSeek CEO, among others. It reminds me of when he visited China during the Lunar New Year celebrations, which had coincided with Trump’s inauguration. In both visits the message was something to the effect of ‘China is an important market’ and ‘Nvidia will continue to operate in China.’ Jensen, with his lower political profile (compared to fellow tech billionaires), is trying very hard to walk the tightrope between U.S. and China. I suspect it’ll be a game of cat and mouse, where Nvidia tailors chips for the Chinese market, while U.S. chip restrictions get tighter and tighter. Meanwhile, domestic Chinese chips will only get better and better.
No TSMC/Intel JV. TSMC denies a reported joint venture with Intel, where TSMC would have helped Intel’s fabs become more efficient. The whole saga back and forth over the past few months was a bit blurry, but it seems that Trump administration officials encouraged the idea and there was never really much buy in from the companies.
Intel CEO’s new management. Intel’s new CEO Lip-bu Tan launches his new management strategy. He is tightening the leadership structure, with more direct reports centralised under the CEO and a new AI Chief. Intel stock is down 1.5%, likely from the fallout of the TSMC JV. But the leadership shakeup should be good news for Intel, which has become a corporate behemoth since the company became trapped under the Innovator’s Dilemma and was managed by a series of businessmen rather than the original batch of engineers.
Thanks for reading.
1. Policy and Geopolitics
1.1
NYT (04/17): Nvidia C.E.O. Meets With Chinese Trade Officials in Beijing
A day after the U.S. government opened an investigation into whether Nvidia, America’s leading chipmaker, violated rules with its sales to China, its chief executive, Jensen Huang, met on Thursday with Chinese trade officials in Beijing.
Mr. Huang had been invited to meet with the China Council for the Promotion of International Trade, a state-backed trade body, according to state media. He also met with He Lifeng, China’s vice premier for economic policy.
The visit was covered by Chinese state media, which reported that Mr. Huang said U.S. controls on Nvidia’s sales to China had a significant impact on the company’s business. Nvidia, the report said, will “continue to spare no effort” to make products that comply with regulations and “unswervingly serve the Chinese market.”
2. Economy, Finance, and Business
2.1
FT (04/17): TSMC denies chip tie-up in prospect with struggling Intel
Taiwan Semiconductor Manufacturing Company (TSMC), the world’s biggest chipmaker, has dismissed reports of a tie-up with its ailing US rival Intel, in its first public comment on an idea pushed by Trump administration officials.
TSMC was “not engaged in any discussion with other companies regarding any joint venture, technology licensing or technology transfer and sharing”, chief executive CC Wei told investors on Thursday.
On Thursday, TSMC maintained its bullish forecast of close to 25 per cent growth this year, driven by booming artificial intelligence chip revenue, but warned that US trade policy would further dilute profit margins and could damage demand.
2.2
Reuters (04/18): Exclusive: Intel CEO Lip-Bu Tan flattens leadership structure, names new AI chief, memo says
Intel's new CEO, Lip-Bu Tan, is flattening the semiconductor giant's leadership team, with important chip groups reporting directly to him, according to a memo from Tan seen by Reuters.
Intel has also promoted networking chip chief Sachin Katti to be chief technology officer and artificial intelligence chief, according to the memo.
Despite acquiring several AI chip startups, Intel failed to settle on a coherent strategy to challenge Nvidia, and in January shelved its most recent attempt, a chip called Falcon Shores. Developing the new AI strategy will fall to Katti.
2.3
Reuters (04/18): Micron rejigs business units to highlight AI data center demand
Micron Technology said on Thursday it is reshuffling its business segments to focus on the artificial intelligence-linked demand for its memory chips from large-scale cloud providers.
The new segment, called "cloud memory business unit", will focus on products used by hyperscalers, as well as its high-bandwidth memory (HBM) chips that help perform data-intensive AI tasks quickly.
3. Technology
3.1
Axios (04/18): TSMC says it will manufacture 30% of most advanced chips in Arizona
The Taiwan Semiconductor Manufacturing Co. will produce 30% of its most advanced chips in Arizona when its six Phoenix plants are operational, the company announced on an earnings call Thursday.
Why it matters: Chairman and CEO CC Wei told investors the scope of the company's Phoenix investment will create "an independent leading-edge semiconductor manufacturing cluster in the U.S."
3.2
TrendForce (04/18): Samsung Reportedly Dismisses Taylor Fab Delay Rumors, Confirms Operations Set to Begin in 2026
Following market rumors that Samsung may slow down its U.S. investment and further delay the timetable for its Taylor plant to 2027, a latest report from KXAN suggests that Samsung pushed back, claiming it still aims to start operations by the end of 2026.
According to South Korean media outlet The Elec, when Samsung’s Taylor investment was announced in late 2021, production was set for 2024—but the timeline has been pushed back multiple times since. Citing weak customer demand, the outlet adds that Samsung has reportedly cut its on-site workforce to a quarter of its original size, fueling talk of further delays.
3.3
Nikkei (04/18): Tenstorrent to boost its chip design engineers in Japan over sixfold
Tenstorrent is preparing to take contracts for designing advanced chips in Japan, with plans to increase engineers there more than sixfold, the Toronto-headquartered contract semiconductor design company said Thursday.
Tenstorrent has been training Japanese engineers at its U.S. sites with support from the Japanese Ministry of Economy, Trade and Industry. It currently has about 15 engineers in Japan, a number it expects to increase to more than 100 by the end of the year.
3.4
TrendForce (04/18): New Breakthroughs in China’s RISC-V Chip Industry
Recently, RIVAI officially unveiled its next-generation high-performance processor, Lingyu, in Qianhai, Shenzhen. As China’s first fully self-developed high-performance RISC-V server chip, Lingyu is up to international mainstream standards in computing power, energy efficiency, and interface configuration, making it suitable for scenarios such as high-performance computing (HPC), all-flash storage, and large open-source language models like DeepSeek.
Lingyu processor is built on RIVAI’s independently developed CPU core IP and on-chip network IP, achieving advanced out-of-order execution, high-speed data paths, and a mesh interconnect architecture. Through a co-optimization approach combining hardware-software co-design and process design, the chip introduces innovations across product engineering, EDA toolchain, physical design, and wafer manufacturing processes, significantly improving energy efficiency and reducing total cost of ownership (TCO).
-