Daily: Top DRAM maker SK Hynix; Intel's 20% layoffs; Huawei's Ascend 920 chips
6.5 min read.
Highlights
Top DRAM maker SK Hynix. SK Hynix is expected to beat Samsung as the world’s top DRAM producer for the first time, as SK Hynix releases their earnings report on Thursday. SK Hynix is expected to have 36% market share to Samsung’s 34% and Micron’s 25%.
Intel’s 20% layoffs. Intel is expected to cut 20% of its workforce this week, or some 21,000 employees. This comes after a round of layoffs of ~15,000 last year. We will get more colour on Intel’s latest strategy during their earnings release on Thursday.
Huawei Ascend 920 chips. Huawei will release their Ascend 920 chips, which are roughly equivalent to Nvidia’s H20 chips that just got banned under the Trump admin. The chips of 6nm calibre, will be in addition to yesterday’s news that Huawei will mass produce H100 equivalents. The gap between US and China is narrowing so quickly.
Thanks for reading.
1. Policy and Geopolitics
1.1
Nikkei (04/23): ByteDance, Alibaba and Tencent stockpile billions worth of Nvidia chips
China's top internet companies have stockpiled billions of dollars' worth of Nvidia's H20 artificial intelligence chips this year before the U.S. cut off shipments of the components in April, Nikkei Asia has learned.
The H20 graphic processing unit (GPU) was specifically designed for sale in the Chinese market to comply with U.S. export controls, but ByteDance, Alibaba and Tencent began preparing last year for the possibility that Washington could curb their shipments, too, sources said.
2. Economy, Finance, and Business
2.1
Bloomberg (04/23): SK Hynix Likely Unseated Samsung as DRAM Leader With AI Boost
SK Hynix Inc.’s upcoming earnings results will likely show it replaced Samsung Electronics Co. as the world’s top DRAM vendor for the first time in the two companies’ decades-long rivalry, thanks to demand for AI.
SK Hynix, a key supplier of high-bandwidth memory to Nvidia Corp., captured 36% of the DRAM market in the March quarter, compared with Samsung’s 34%, data from Counterpoint Research show. This comes just after operating profit at Samsung, which held the DRAM crown for more than 30 years, fell below SK Hynix’s for the first time in the December quarter.
On Thursday, SK Hynix is projected to report a 38% quarterly rise in sales and a 129% surge in operating profit in the three months to March, according to analysts polled by Bloomberg.
2.2
Bloomberg (04/23): Intel to Announce Plans This Week to Cut Over 20% of Staff
Intel Corp. is poised to announce plans this week to cut more than 20% of its staff, aiming to eliminate bureaucracy at the struggling chipmaker, according to a person with knowledge of the matter.
The move is part of a bid to streamline management and rebuild an engineering-driven culture, according to the person, who asked not to be identified because the plans are private. It would be the first major restructuring under new Chief Executive Officer Lip-Bu Tan, who took the helm last month.
The cutbacks follow an effort last year to slash about 15,000 jobs — a round of layoffs announced in August. Intel had 108,900 employees at the end of 2024, down from 124,800 the previous year.
2.3
Reuters (04/23): Intel results to spotlight new CEO Tan's strategy to reverse years of missteps
Intel's quarterly results will offer the clearest look yet at new CEO Lip-Bu Tan's turnaround strategy for the embattled American chipmaker, and investors are hoping for early signs that he is reversing years of strategic lapses.
The company is set to post its fourth consecutive quarterly revenue drop on Thursday.
2.4
TrendForce (04/23): Samsung Faces High Tariff Risks as It Reportedly Hesitant to Bring in Chipmaking Tools for Taylor
Samsung has denied rumors that its Taylor plant timeline will be pushed to 2027, but concerns are growing over the mounting hurdles in its U.S. investment. According to the Chosun Daily, the company has been reluctant to bring in key equipment like EUV machines to the Taylor facility — a move that could result in hefty import tariffs down the line.
KXAN suggests that Samsung still aims to start operations by the end of 2026. As noted by the Chosun Daily, Samsung’s new fab in Taylor, Texas, is 99.6% complete. However, unlike the usual pace, equipment hasn’t been brought in yet, reportedly due to Samsung’s hesitation in placing orders.
2.5
Bloomberg (04/23): Chinese Buyout Firm Is Said to Consider Sale of Chip Tester UTAC
Wise Road Capital Ltd. is considering selling semiconductor assembly and testing company UTAC Holdings Ltd., according to people familiar with the matter.
The Beijing-based private equity firm has engaged an adviser and could seek about $3 billion in a potential sale, the people said, asking not to be identified discussing private deliberations.
UTAC has production facilities in Singapore, where it is headquartered, as well as China, Indonesia and Thailand, with a sales network focused primarily on the US, Europe and Asia, the company’s website shows.
3. Technology
3.1
TrendForce (04/23): The TC Bonder Drama: Could Hanmi’s Supply Shift the HBM Race Between Micron, SK hynix, and Samsung?
As the HBM race heats up among memory giants, TC (thermal compression) bonders are taking center stage. Notably, recent developments involving key player Hanmi Semiconductor, Micron, and SK hynix have sparked attention—and could potentially reshape the memory landscape, according to reports from The Elec and DealSite.
As per The Elec, Hanmi Semiconductor has reportedly landed a major order from Micron for around 50 TC bonders lately, which would be significantly larger than the dozens of units shipped to the U.S. memory maker in 2024.
However, Hanmi’s closer alignment with Micron and rumored price hikes for SK hynix point to growing tensions in their long-standing partnership. According to The Elec, Hanmi increased TC bonder prices for SK hynix by 25–28% and withdrew all engineers from its Icheon HBM line to focus more on Micron.
3.2
SCMP (04/23): No need for Nvidia: iFlytek touts reasoning model trained entirely with Huawei’s AI chips
Chinese voice-recognition firm iFlytek said that training its large language models (LLM) entirely with Huawei Technologies’ computing solutions has increased its growth potential amid the intensifying US-China tech war, after the Trump administration moved to restrict the export of Nvidia’s H20 artificial intelligence (AI) chips to China.
iFlytek on Monday boasted that its Xinghuo X1 reasoning model, a “self-sufficient, controllable” LLM trained with home-grown computing power, had matched OpenAI o1 and DeepSeek R1 in overall performance following an upgrade, according to a company blog post published on WeChat.
At the end of last year, the efficiency of Huawei’s Ascend 910B AI chip was only 20 per cent that of Nvidia’s solution for the training of reasoning models, but iFlytek and Huawei have jointly increased that to nearly 80 per cent this year, iFlytek founder and chairman Liu Qingfeng said on Tuesday during an earnings call with investors.
3.3
SCMP (04/22): Huawei to roll out AI chips in second half as potential alternative to Nvidia H20: report
Huawei Technologies will roll out its new Ascend 920 artificial intelligence (AI) chip later this year, as the company steps into the role as China’s alternative supplier to US giant Nvidia, which is subject to even tighter US export restrictions, according to media reports and industry insiders.
The Ascend 920, which was at the advanced 6-nanometre node level, is expected to begin mass production in the second half of this year, with analysts saying it could become an alternative to Nvidia’s H20 AI chips that were just banned from sale to China, according to a report by Taiwanese media Digitimes.
3.4
EE Times (04/22): After Three Years, Modular’s CUDA Alternative Is Ready
Building a CUDA alternative was never going to be an easy task.
Chris Lattner’s team of 120 at Modular has been working on it for three years, aiming to replace not just CUDA, but the entire AI software stack from scratch.
Modular, as a software-only company, is better positioned to build a stack that works for all hardware, according to Lattner.
Modular’s AI inference engine, Max, launched in 2023 with CPU support for x86 and Arm CPUs, and support for Nvidia GPUs was added recently. This means Modular now has a full-stack replacement for CUDA, including the CUDA programming language and the LLM serving stack that builds on top of it.
Crucially, Lattner said Max can meet the performance of CUDA for Nvidia A100 and H100 GPUs.
3.5
The Verge (04/23): Nvidia’s GPU drivers are a mess
Nvidia’s GPU drivers have been a disaster over the past four months. It all started when Nvidia released its drivers for the RTX 50-series cards in January, and introduced black screen issues, game crashes, and general stability problems for new and existing graphics cards. Now, yet another new hotfix has emerged to fix even more issues introduced by Nvidia’s buggy drivers.
Nvidia GPU owners have been struggling for months with a variety of issues reported on Reddit and in Nvidia’s own support forums, with most people fixing their issues by rolling back to the December 566.36 driver before the RTX 50-series. Those fortunate enough to secure one of Nvidia’s latest GPUs aren’t able to roll back to the old drivers as they simply don’t support the new RTX 50-series, so Nvidia has been issuing a series of updates to try and address the problems.
-