Yesterday, United States President Donald Trump announced a set of tariffs imposed on US trading partners, imposing a series of 10%+ tariffs on partners, calling it a "Liberation Day." Today, we are calculating how much these tariffs will impact consumers and what is most important at TechPowerUp: semiconductors powering our GPUs and CPUs. According to one of the top investment banks, Goldman Sachs, semiconductors are exempt from the reciprocal tariffs that Trump has imposed on Taiwan. However, the semiconductor manufacturing equipment used by makers like TSMC is not exempt and is expected to be hit with the 32% tariffs. This is only half of what Taiwan imposes on imports of US-made goods. For TSMC, the number one maker of GPUs and CPUs, tariffs can be tricky to navigate. While its existing manufacturing facilities use equipment sourced from Dutch ASML and a few US companies like Lam Research and KLA Corporation, it shouldn't be a problem to ship new silicon to the US.

However, if TSMC wants to expand its manufacturing facilities in any country that is not the US, it will have to deal with 32% tariffs on US-sourced silicon manufacturing equipment. For EU-based ASML, things are looking a little different. If over 20% of the equipment is made up of US content, a tariff exemption might apply, potentially reducing import costs. If more than one-fifth of a product's components or value originates from US sources, the equipment may be eligible for tariff relief. ASML's machines include some US components, so determining whether these machines meet the 20% threshold is crucial. If they do, the tariff exemption could help lower costs associated with importing these advanced machines, reaching up to $380 million. For non-US-injected goods, EU entities are subject to 20% tariffs.

NVIDIA's CEO Jensen Huang made some interesting commentary for the Financial Times, stating that the company will procure over half a trillion US Dollars worth of electronics over the next four years, and it it plans to keep hundreds of billions from the supply chain procurement in the US. "Overall, we will procure, over the course of the next four years, probably half a trillion dollars worth of electronics in total. And I think we can easily see ourselves manufacturing several hundred billion of it here in the US," said Jensen for FT. NVIDIA currently manufactures its silicon at TSMC's facilities, as well as electronics like motherboards and servers at Foxconn. However, the geopolitical situation is making NVIDIA reconsider its supply chain dependencies, and the company is looking for more US-based manufacturing.

NVIDIA confirmed that its latest "Blackwell" series of GPUs, including the latest Blackwell Ultra, are being manufactured at TSMC's Arizona facilities. TSMC announced a $100 billion investment in its Arizona expansion, and NVIDIA is ready to take up more of TSMC's capacity to meet its ever-growing demand for GPUs. During the GTC 2025 event, Jensen noted that only four cloud service providers will use 3.6 million GPUs this year. That is without any AI labs and enterprises, which are massive consumers of GPUs (xAI only has 200,000 GPU clusters). To continue manufacturing excellence so customers won't suffer, NVIDIA is also looking at other options for supply chain manufacturing partners. Intel, the only US-based company capable of producing advanced silicon, is a potential target for NVIDIA. "We evaluate their foundry technology on a regular basis, and we are ongoing in doing that... We look for opportunities to be a customer of theirs... I have every confidence that Intel can do it," added Jensen, who also stated that NVIDIA is interested in silicon manufacturing and chip packaging services, as Intel's Foveros 3D packaging and other technologies are attractive for Team Green.

According to a recent Center for Strategic and International Studies report, Huawei got its hand on approximately two million Ascend 910B logic dies through shell companies that misled TSMC. This acquisition violates US export controls designed to restrict China's access to advanced semiconductor technology. The report details how Huawei leveraged intermediaries to procure chiplets for its AI accelerators before TSMC discovered the deception and halted shipments. These components are critical for Huawei's AI hardware roadmap, which progressed from the original Ascend 910 (manufactured by TSMC on N7+ until 2020) to the domestically produced Ascend 910B and 910C chips fabricated at SMIC using first and second-generation 7 nm-class technologies, respectively. Huawei reportedly wanted TSMC-made dies because of manufacturing challenges in domestic chip production. The Ascend 910B and 910C reportedly suffer from poor yields, with approximately 25% of units failing during the advanced packaging process that combines compute dies with HBM memory.

Despite these challenges, the performance gap with market-leading solutions still remains but has narrowed considerably, with the Ascend 910C reportedly delivering 60% of NVIDIA H100's performance. Huawei has executed a strategic stockpiling initiative, particularly for high-bandwidth memory components. The company likely acquired substantial HBM inventory between August and December 2024, when restrictions on advanced memory sales to China were announced but not yet implemented. The semiconductor supply chain breach shows that enforcing technology export controls is challenging, and third parties can still purchase silicon for restricted companies. While Huawei continues building AI infrastructure for both internal projects and external customers, manufacturing constraints may limit its ability to scale deployments against competitors with access to more advanced manufacturing processes. Perhaps a future domestic EUV-based silicon manufacturing flow will allow Huawei to gain access to more advanced domestic production, completely circumventing US-imposed restrictions.

Intel Products today launched the Intel Assured Supply Chain (ASC) program, designed to provide additional transparency and assurance in the silicon manufacturing process. This specialized client system-on-chip (SoC) solution, initially offered on specific Intel Core Ultra Series 2 mobile and desktop SKUs, provides a digitally
attestable chain of custody of each chip's progress through the chip manufacturing process, leveraging a dedicated chip manufacturing pathway through specific Intel
manufacturing locations. With the ASC program, Intel Products delivers added transparency into processor manufacturing, assuring customers about the locations of
their silicon supply chain.

"Intel has long been a leader in secure, transparent and reliable semiconductor manufacturing, and the Intel Assured Supply Chain program is another step forward in strengthening trust in the technology that powers our customers' critical operations," said Jennifer Larson, general manager, Commercial Client Segments, Client Computing Group, Intel.

A few days ago, we published a report about Intel's 18A yields being at an abysmal 10%. This sparked quite a lot of discussion among the tech community, as well as responses from industry analysts and Intel's now ex-CEO Pat Gelsinger. Today, we are diving into known information about Intel's 18A node and checking out what the yields of possible products could be, using tools such as Die Yield Calculator from SemiAnalysis. First, we know that the defect rate of the 18A node is 0.4 defects per cm². This information is from August, and up-to-date defect rates could be much lower, especially since semiconductor nodes tend to evolve even when they are production-ready. To measure yields, manufacturers use various yield models based on the information they have, like the aforementioned 0.4 defect density. Expressed in defects per square centimeter (def/cm²), it measures manufacturing process quality by quantifying the average number of defects present in each unit area of a semiconductor wafer.

Measuring yields is a complex task. Manufacturers design some smaller chips for mobile and some bigger chips for HPC tasks. Thus, these two would have different yields, as bigger chips require more silicon area and are more prone to defects. Smaller mobile chips occupy less silicon area, and defects occurring on the wafer often yield more usable chips than wasted silicon. Stating that a node only yields x% of usable chips is only one side of the story, as the size of the test production chip is not known. For example, NVIDIA's H100 die is measuring at 814 mm²—a size that is pushing modern manufacturing to its limits. The size of a modern photomask, the actual pattern mask used in printing the design of a chip to silicon wafer, is only 858 mm² (26x33 mm). Thus, that is the limit before exceeding the mask and needing a redesign. At that size, nodes are yielding much less usable chips than something like a 100 mm² mobile chip, where defects don't wreak havoc on the yield curve.

Infineon Technologies AG today announced that the company has succeeded in developing the world's first 300 mm power gallium nitride (GaN) wafer technology. Infineon is the first company in the world to master this groundbreaking technology in an existing and scalable high-volume manufacturing environment. The breakthrough will help substantially drive the market for GaN-based power semiconductors. Chip production on 300 mm wafers is technologically more advanced and significantly more efficient compared to 200 mm wafers, since the bigger wafer diameter offers 2.3 times more chips per wafer.

GaN-based power semiconductors find fast adoption in industrial, automotive, and consumer, computing & communication applications, including power supplies for AI systems, solar inverters, chargers and adapters, and motor-control systems. State-of-the art GaN manufacturing processes lead to improved device performance resulting in benefits in end customers' applications as it enables efficiency performance, smaller size, lighter weight, and lower overall cost. Furthermore, 300 mm manufacturing ensures superior customer supply stability through scalability.

According to sources close to Bloomberg, Intel plans to cut 10,000 jobs from its global workforce. The news comes amid heavy pressure on the semiconductor giant, which has been on a steady decline over the years, while other industry rivals like AMD and NVIDIA have been rising and taking market share in various areas from Intel. It is reported that Intel currently has 110,000 employees globally, and reducing the workforce by 10,000 would net Intel around 100,000 global employees left. These figures exclude employees from spun-out units like Altera FPGA company, which is under Intel's ownership. Intel's aim to reduce its workforce is expected to come with a significant cost benefit to the company, with projected savings of $10 billion by 2025.

The news isn't yet official, but it is expected to see the light of the day as soon as this week. As Intel's CEO Pat Gelsinger invests heavily into the fab construction and development of next-generation products, there have been a few notes that Intel would have to overcome some challenges shortly to reach its long-term goals like more advanced silicon manufacturing facilities and new products for AI/HPC and client sector. One of those short-term measures is reducing the workforce to cut down expenses. Intel has reduced its workforce before. In 2022, the company announced reduced spending in non-critical areas and reducing the workforce, and in 2023, cut the workforce by 5% to 124,800 employees last year, only to be left with 110,000 employees in 2024.

Tenstorrent, a company that sells AI processors and licenses AI and RISC-V IP, announced today that it selected Samsung Foundry to bring Tenstorrent's next generation of AI chiplets to market. Tenstorrent builds powerful RISC-V CPU and AI acceleration chiplets, aiming to push the boundaries of compute in multiple industries such as data center, automotive and robotics. These chiplets are designed to deliver scalable power from milliwatts to megawatts, catering to a wide range of applications from edge devices to data centers.

To ensure the highest quality and cutting-edge manufacturing capabilities for its chiplet, Tenstorrent has selected Samsung's Foundry Design Service team, known for their expertise in silicon manufacturing. The chiplets will be manufactured using Samsung's state-of-the-art SF4X process, which boasts an impressive 4 nm architecture.

In a Samsung Electronics business report released on March 12, it was revealed that the giant electronics manufacturer will begin mass production of new chips through a 4 nm 2.3-generation process. The newly established manufacturing process is set to start by the end of H1 2023, which means that mid-June would be the expected commencement date. Samsung Semiconductor has managed to produce a satisfactory yield of wafers with the new generation chipsets.

Samsung Electronics has experienced significant problems with the production of previous generation 4 nm chips, and industry insiders have been surprised by the sudden announcement of the third generation version, given rumors pointing to the Hwaseong factory struggling to reach yields at the 60% mark. Qualcomm famously dropped Samsung in favor of TSMC as a source of 4 nm chipsets in 2022, due to disappointing yield figures.

The US Department of Commerce is in the process of increasing the stranglehold in tech exports directed to Chinese shores. The move is being made through the delivery of letters to US-based technology companies - namely KLA Corp, Lam Research Corp and Applied Materials Inc. - ordering them to stop the export of machines and equipment that can be used for sub-14 nm manufacturing. The move by the Department of Commerce only has validity for the companies that have been served by such a letter - at least until the Department codifies its newest regulations.

This means that only sellers with approved export licenses can keep doing business with Beijing, thus limiting the US companies China can work with as it aims to achieve at least a degree of self-sufficiency in the latest chipmaking tech. Perhaps the decision has come too late, however, as China's mainstay silicon manufacturing, SMIC, already manufactures chips at the 14 nm process (chips that have been deployed in China's Tinahu Light supercomputer already) and has even showcased manufacturing capability in the 7 nm field. It pays to remember that the US already had applied similar restrictions on equipment experts to China for the better part of two years - which apparently did little to stem China's capability to create increasingly denser semiconductor designs.

Rather unusually, Intel announced its latest chip fab plans not via a press release, but via an exclusive article in TIME magazine. It seems like an unusual strategy, as these kinds of things are normally not announced in this kind of fashion, especially as TIME doesn't exactly have close ties with Intel, nor the semiconductor industry as a whole, but maybe this is Intel's new way of trying to change the image of the company. Either which way, Intel is apparently planning on building no less than two fabs on the 1,000 acre (~4 square kilometer) site in New Albany, Ohio, which should be the workplace of some 3,000 people once it stands ready in 2025.

The article quotes Pat Gelsinger saying "Our expectation is that this becomes the largest silicon manufacturing location on the planet," as Intel has the option to double the land for its new fab site and apparently has plans for as many as eight fabs at the location. Additional fabs obviously depends on demand and crucially if Intel manages to full off its contract foundry business, since without it, it seems unlikely that Intel is going to need the additional six fabs in the foreseeable future, especially as Intel is in the final stages of finishing its new fab in Ireland, while also planning to announce a location for yet another fab somewhere in Europe and let's not forget Arizona. The TIME article goes into a lot more details as to what the new fabs mean for Ohio, but doesn't go into much more detail about Intel's plans for its future fabs there.

Update: Official press release below.

Semiconductor manufacturing is no easy task. Every company in that business knows that, and the hardships of silicon manufacturing have been felt by even the greatest players like Samsung and Intel. Today, according to the latest report from Business Korea, Samsung is again in trouble with its 5 nm node. It has been reported previously that Samsung is struggling with yields of its 5 nm node, however, we didn't know just how much until now. According to the sources over at Business Korea, Samsung's 5 nm semiconductor node is experiencing less than 50% yields. That means, for example, that out of 100 chips manufactured on a single silicon wafer, only half are functional. And that is not good at all.

Usually, for a node to go into high-volume manufacturing (HVM), the yielding rate needs to be around 95%. In case it is not at that level, manufacturing of that node is not very efficient and not very profitable. The V1 Line in Hwaseong, where this Samsung 5 nm is made, uses EUV tools to manufacture the new node. While the yields are currently below 50%, it is expected to improve as Samsung engineers tweak and tune the node and the tools that are running the facility. We can expect to hear more about the yields of this node in the coming months.

UMC has announced plans to invest $3.6 billion in increasing output from its 28 nm manufacturing facilities. This move comes amidst a global semiconductor shortage, and isn't the first time a semiconductor manufacturer "dust off" their older manufacturing processes as a way to remove pressure from more modern silicon manufacturing capabilities. In this case, UMC will be increasing manufacturing output from its 300 mm Fab 12A facility in Tainan, Taiwan.

UMC has entered agreements with some of its clients, who will be paying upfront for expected chip rollout in the future. In exchange, clients will get the benefits of preset pricing (thus avoiding any potential increases arising from increased demand or general price fluctuation), as well as UMC's assurance of certain manufacturing volume allocation towards their needs. Fab 12A currently manufactures 90,000 300 mm wafers per month (wpm). An additional 10,000 wpm is being installed this year and phase six will add another 27,500 wpm to the mix. The mature 28 nm tools will be installed in floors that already feature support for future tooling upgrades to 14 nm. UMC expects to hire around 1,000 additional employees as part of this expansion effort.

Manufacturing silicon is no easy task. You need to have all the right supplies available all the time. One of the most used ingredients in silicon manufacturing is water. Almost every process needs it and it needs to be constantly available to the manufacturer. According to the report coming from Reuters, Taiwan Semiconductor Manufacturing Company (TSMC) and United Microelectronics Corporation (UMC) are experiencing water shortages. The Taiwan island is in trouble, as the typhoon season has been rather mild and water supplies are at the historic lows. Water restrictions are in place all across the island and the reservoirs in the center and southern regions are at only 20% capacity.

The lack of water is a big problem for TSMC and UMC, as both companies rely on the constant income of it. With water restrictions in place, TSMC has to keep its facilities running and needs to solve the problem. That is why Taiwan's biggest silicon manufacturer is now making small orders of waters, delivered by a truckload. TSMC expects to compensate for the lack of water coming from its regular sources with truckloads of it. While we do not know the numbers of it, we can expect the water use to be very high if we take into account the number of wafers TSMC produces at its facilities.

Samsung has been one of the world's biggest foundries and one of three big players still left in the leading-edge semiconductor process development and manufacturing. However, the Korean giant is always seeking ways to improve its offerings, especially for Western customers. Today, it is reported that Samsung has reportedly talked with regulators in Texas, New York, and Arizona about building a $17 billion silicon manufacturing facility in the United States. The supposed factory is going to be located near Austin, Texas, and is supposed to offer around 1800 jobs. If the deal is approved and Samsung manages to complete the project on time, the factory is supposed to start mass production in Q4 of 2023.

What process is Samsung going to manufacture in the new fab? Well, current speculations are pointing out to the 3 nm node, with Samsung's special GAAFET (Gate All Around FET) technology tied to the new node. The fab is also expected to make use of extreme ultraviolet (EUV) lithography for manufacturing. Samsung already has a facility in the US called S2, however, that will not be upgraded as it is still serving a lot of clients. Instead, the company will build new facilities to accommodate the demand for newer nodes. It is important to note that Samsung will not do any R&D work in the new fab, and the company will only manufacture the silicon there.

It's been doing the rounds in the rumor mill that AMD is looking to expand its semiconductor manufacturing partners beyond TSMC (for the 7 nm process and eventually 5 nm) and Global Foundries (12 nm process used in its I/O dies). The intention undoubtedly comes from the strain that's being placed on TSMC's production lines, as most foundry-less businesses outsource their wafer production to the Taiwanese companies' factories and manufacturing processes, which are currently the industry's best. However, as we've seen, TSMC is having a hard time scaling its production facilities to the unprecedented demand it's seeing from its consumers. The company also has recently announced it may prioritize new manufacturing capabilities for the automotive industry, which is also facing shortages in chips - and that certainly doesn't instill confidence in capacity increases for its non-automotive clients.

That's what originated form the rumor mill. Speculating, this could mean that AMD would be looking to outsource products with generally lower ASP to Samsung's foundries, instead of trying to cram even more silicon manufacturing onto TSMC's 7 nm process (where it already fabricates its Zen 3, RDNA 2, EPYC, and custom silicon solutions for latest-gen consoles). AMD might thus be planning on leveraging Samsung's 8 nm or even smaller fabrication processes as alternatives for, for example, lower-than-high-end graphics solutions and other product lines (such as APUs and FPGA production, should its acquisition of Xilinx come through).

The possibility barely exists to account for all the silicon manufacturing processes currently in development; TSMC themselves are rolling out 5 nm, 4 nm, 3 nm, and 2 nm processes at various points in time in the future. Now, the company has announced that it will be rolling out a revision of the 3 nm manufacturing process, named 3 nm Plus, come 2023. According to DigiTimes, the Taiwanese manufacturer's first client for this process will be Apple.

There is no information on what exactly 3 nm Plus leverages and offers over the "vanilla" 3 nm process. It could be anything from higher transistor density, lower power consumption, or higher operating frequency - or maybe a mixture of the three. The original 3 nm manufacturing process is set to offer a 15% performance gain over the current top-of-the-line 5 nm node, with 30% decreased power use and up to 70% density increase. Interestingly, TSMC is keeping their FinFet manufacturing technology, on grounds of better implementation costs and higher power efficiency compared to the more exotic GAA (Gate-All-Around) technology that its rival Samsung, for one, aims to implement in 3 nm.

When the US decided to impose sanctions on all US-made technology use in foreign countries (China), the Chinese semiconductor manufacturing industry seemed at the time that it would just completely stop. Without the tools to manufacture silicon, Chinese manufacturers would need to turn to other countries to search for a possible solution. That, however, turned out impossible as the US administration has decided to stop the silicon from going into the hands of Chinese companies, by making a condition that any US-made technology can not get to China. Many of the parts for silicon manufacturing are designed in the US, so they have the power to restrict the use.

Today, in a surprising turn of events, we have information that Shanghai Micro Electronic Equipment (SMEE) has developed a deep ultraviolet (DUV) lithography scanner that is set for delivery in 2021. With a plan to deliver it in the fourth quarter of 2021, SMEE has designed this DUV scanner for the production of 28 nm node. While not being the most advanced node available to date, it is a significant start for Chinese technology independence. ASML, the producer of such scanners, used to be one of the few options there, however, it just gained a competitor. China will deliver its new silicon on a 28 nm process at the end of 2021. Pictured below, you can see how the scanner from SMEE looks like.

Silicon manufacturing is starting to get harder and harder every day, with new challenges appearing daily. It requires massive investment and massive knowledge to keep a silicon manufacturing company afloat. No company can survive that alone, so some collaborations are emerging. Today, thanks to the sources of Nikkei Asia, we have information that Taiwanese Semiconductor Manufacturing Company (TSMC) is collaborating with Google to push the production of 3D chip manufacturing process, that is said to overcome some of the silicon manufacturing difficulties. The sources also say that AMD is involved in the process as well, making Google and AMD the first customers of the advanced 3D chip design. The two companies are preparing designs for the new way of creating silicon and will help TSMC test and certify the process.

TSMC will deploy the 3D silicon manufacturing technology at its chip packaging plant in Miaoli, which is supposed to do mass production in 2022. With Google and AMD being the first customers of new 3D technology, it is exciting to see what new products will look like and how they will perform. The 3D approach is said to bring huge computing power increase, however, it is a waiting game now to see how it will look like.

The semiconductor manufacturing industry is a cutthroat competition mostly played between established forces. One need only look to AMD's decision to spin-off its manufacturing arm to create Global Foundries to see how even a grand company can hit manufacturing issues (though not only manufacturing issues hit AMD at that time, obviously) can threaten to shutter operations. Intel's recent issues with 10 nm and 7 nm fabrication also come to mind. as such, it comes at no great surprise that Chinese company Wuhan Hongxin Semiconductor Manufacturing Company (HSMC) has reportedly run out of cash. What's a little more surprising is how this company was actually backed by the Chinese government, and yet it still failed - proof of the semiconductor industry's technical and investment liquidity requirements.

HSMC back in 2017 announced plans to bring online a manufacturing plant in the central Chinese province of Hubei. The aim was to manufacture 14 nm and 7 nm chips as early as 2019/2020, funded by a $20 billion loan and numerous funding rounds. HSMC's ex-CEO Chiang Shang-yi (who previously served as former head of R&D at TSMC) said to EETimes that "Investors ran short of cash." And that was it for the company's aspirations. The company has now been absorbed by the municipal government in the central Chinese province of Hubei, and what will come of that (and the company's future) remain uncertain.