TSMC

TSMC is on the ropes and its biggest problem is not competition: it is water

by

CC Wei, the president and CEO of TSMC, has just participated in the inauguration of the Pingtung technology park, southern Taiwan. The ceremony was held under very intense rain, and the head of the largest chip manufacturing company of the planet has celebrated it publicly: “Last month I was still wondering: what should we do with the water? Should we start using tankers?” Wei has compelling reasons to be concerned about water. In Taiwan it is a very scarce resource, and it is also essential for its semiconductor factories. The integrated circuit production industry faces five major shortages: water, energy, labor, land and talent. And water is probably the most precious resource of all and the most difficult challenge to overcome. Lai Ching-te, the president of Taiwan, has communicated CC Wei the Government’s plans to connect the island’s reservoirs together in a clear attempt to alleviate this problem. The water we are familiar with, such as that which comes from the tap, spring water, and even bottled mineral water, is full of impurities. It contains bacteria, dissolved gases, mineral salts and microscopic particles in suspension. This is not a problem for most of the everyday applications for which we usually use it, but this water is not suitable for making chips. Even the smallest impurity invisible to the human eye is pure poison when involved in the production of cutting-edge semiconductors, such as 2nm integrated circuits which are currently being manufactured by TSMC, Intel and Samsung. Droughts are the biggest threat to TSMC The integrated circuit manufacturing process requires cleaning silicon wafers dozens of times. Every time a geometric pattern is transferred to wafers using lithography, they need to be cleaned. Also after pouring chemical reagents and photoresist fluids on them. However, the water used to remove any residue that may have deposited on the wafer cannot have the slightest impurity. It must be absolutely pure. In fact, the industry standard calls for water with an electrical resistivity of 18.2 megaohms per centimeter, which is the theoretical limit of water purity at room temperature. The problem is that producing ultrapure water is not easy. And it is not because it is necessary to subject it to reverse osmosis in multiple stages and ion exchange treatments. It is also necessary to degas it under vacuum, eliminate any microorganisms that it can contain with ultraviolet light and filter it using membranes expressly designed to capture the slightest impurity. In this article we do not need to delve into these processes, but there is something that we cannot ignore: this treatment consumes energy and requires the use of a large amount of chemicals. Furthermore, a significant part of the water that is processed is not transformed into ultrapure water, so it cannot be used. In 2021 Taiwan faced the most aggressive drought in recent decades In 2021 Taiwan faced the most aggressive drought in recent decades. The reservoirs that supply the north of the island, which is the region where most of the integrated circuit manufacturing plants are located, reached a critical level. This scenario forced the Government to suspend agricultural irrigation in order to protect industrial supply. Even so, TSMC had to resort to tanker trucks to transport water from other areas of the island. This crisis passed, but it exposed a structural vulnerability of an industry that is deeply sensitive to extreme climate phenomena and for which water is as valuable as silicon or some rare earth elements. At the moment TSMC reuses 85% of water that it consumes in its processes, but its plants require so much water daily that the 15% that cannot be reused continues to be a very large amount. Paradoxically, some of the most important chip factories in the US reside in Arizona, which is the second driest state in the country only behind Nevada. Both Intel and TSMC have state-of-the-art plants in this region, and Samsung manages several state-of-the-art factories in Texas, another state where water is not exactly plentiful. Besides, Intel has several very important plants in Israelagain a country for which water is a scarce resource, and, therefore, of great strategic value. The choice of these locations responds to geostrategic and geopolitical factors, and not to planning that prioritizes sustainability. In addition, there is another unappealable reality that further complicates this panorama: the most advanced integrated circuit manufacturing nodes. consume more ultrapure water than mature nodes. A 3nm chip from TSMC goes through more than 1,000 individual stages, and many of them require washing with ultrapure water As transistors become smaller and chips become more sophisticated, the number of stages required for their production process is higher, making it necessary to clean the wafers with ultrapure water more times. A 3nm semiconductor from TSMC goes through more than 1,000 individual stages, many of which require washing with ultrapure water. 2nm chips are even more demanding, and those that will arrive in the future will surely multiply water consumption. The problem faced by TSMC, Intel, Samsung, SK Hynix, Micron, GlobalFoundries and other semiconductor manufacturers is daunting. These companies are developing solutions, but at the moment they are not enough. TSMC, Samsung and Intel have very advanced water recycling and reuse programs. In addition, these companies are developing dry cleaning systems for wafers using reactive plasmas instead of water for use in some steps. And they are even looking for new materials for the ultrapure water distribution pipes with the purpose of minimizing losses in the circulation system. The problem is that these solutions do not eliminate the deep dependence that chip factories have on water. The definitive solution has not yet arrived. And it is urgent. It is very urgent. Image | TSMC More information | Reuters In Xataka | ChatGPT blocking mode: what it is, what it is for, who can use it and how to activate it In Xataka | AI is replacing one of the most hated jobs in the world: the tailcoat collector

The world was tired of depending on TSMC to manufacture all its chips. This is what is causing Intel’s great resurrection

by

Who has seen you and who sees you, Intel. The legendary semiconductor firm seems to be leaving behind its painful journey through the desert, and the latest news points to a true resurrection. The signature has achieved a spectacular contract to manufacture three million Google TPUs, and Nvidia is also studying the possibility of use Intel 18A node for future multi-die GPU designs. This is spectacular news for the company. Promising future, at last. The agreement with Google’s cloud division is a huge boost for the chip manufacturing business (foundry) from Intel. This deal will see Intel produce millions of AI chips at its advanced 3-nanometer node. With it, the firm achieves a decisive step to compete with TSMC, which until now was the absolute reference for those who wanted to access advanced semiconductor manufacturing processes. There is another crucial geopolitical factor here: part of these chips will be produced in the US, which helps in the objective of not depending so much on Asian countries for this process. Flirting with Nvidia. But Nvidia also seems to be interested in Intel’s 18A photolithographic process. The company led by Jensen Huang is considering the use of this node for its future multi-die architectures for its GPUs. Nvidia has managed to become TSMC’s main customer, but this manufacturer cannot satisfy Nvidia’s demand, so this company is looking for plans B, and Intel is serving it one on a plate. The signature by the way, already bought 4% of Intel in September 2025, so it is the first interested in Intel doing well. The PowerVia revolution. There are two big technical arguments that are apparently convincing Google and Nvidia. The first, the transistors RibbonFET. The second, even more important, PowerVia technology. This system is a qualitative leap because it physically separates the power and signal lines from the transistors, which avoids bottlenecks and improves both performance and efficiency of the CPUs that use this technology. Chip sovereignty. This decision by Google and Nvidia’s plan respond in part to the pressures that the US government is doing—and boosting with its CHIPS Act— to recover technological sovereignty and avoid dependence on foreign countries. Both companies know that 90% of the planet’s advanced chips depend on that island called Taiwan, and taking advantage of Intel’s renewed capacity is a great opportunity for kill two birds with one stone. They reduce their dependence on TSMC, and comply with the demands of the US government. War makes strange allies. The current situation is unique, because it is causing companies that competed fiercely in the field of hardware (Intel and Nvidia) to now be forced to collaborate out of pure necessity. Intel needs clients of this type to demonstrate to investors that its division foundry can operate independently of its consumer processor or server division. And Google and Nvidia in turn need Intel to break manufacturing monopoly of semiconductors that TSMC had. Intel finally resurrects. The big winner of these agreements is Intel, which has gone through a really compromising stage but has for a year has not stopped growing. We can see it in its valuation on the stock market. A year ago its shares were trading at $20.68, and now they are trading at $107.04 and with these agreements that value may continue to improve. Good for Intel. Image | Intel In Xataka | Bill Gates has X-rayed Intel. And his diagnosis is overwhelmingly accurate.

TSMC chairman challenges Chinese chipmakers

by

TSMC leads the integrated circuit manufacturing industry overwhelmingly. The current market share of this Taiwanese company is approximately 70%, according to the consulting firm TrendForce. Samsung is the second largest producer of chips for third parties, although with a market share of 7.2% is positioned very far from the leader of this industry. And the Chinese company SMIC (Semiconductor Manufacturing International Corp) is hot on his heels in third position with a share of 5.32%. However, there was a time when Intel dominated the semiconductor industry with a force comparable to that currently held by TSMC. Its rise to leadership began with the 1981 agreement with IBM to supply the processor for the original PC, making the x86 architecture the standard. de facto of personal computing. Over the next three decades, Intel set the pace of technological development in the semiconductor industry, but its decline began in 2015 when it began accumulating delays in the transition to the most advanced nodes. In 2010, probably few analysts in this sector would have predicted that TSMC would take over from Intel both from a technological point of view and from a strictly commercial perspective. Its market share of 70% says it all. Now it is the Chinese chip manufacturers who are slowly beginning to intimidate. SMIC, as we have just seen, is already hot on Samsung’s heels in the fight to manufacture semiconductors for third parties. And Hua Hong and Huawei are pushing harder and harder. So much so, in fact, that TSMC shareholders are starting to get restless. Huawei wants to change the rules of the game that TSMC is winning CC Wei, the current president of TSMC, has assured that his company “is not afraid” of competition from China. This assertion is not accidental. In fact, this executive spoke these words in response to a shareholder’s question during the annual meeting that was held just a few hours ago. wei has pointed out that competition has been a constant throughout the company’s forty years of history. And it’s true. But it is also true that no company has led the chip sector forever. Fairchild Semiconductor, Texas Instruments, NEC, Toshiba, Hitachi, and, of course, Intel, have led. And all of them have fallen. If we look towards China, the companies that seem to worry TSMC the most are SMIC and Huawei. In fact, their alliance has allowed SMIC to manufacture 7nm integrated circuits using the equipment deep ultraviolet photolithography (UVP) of ASML and without the need to resort to the most advanced machines of extreme ultraviolet lithography (EUV). Also, as we told you last week. Huawei has presented a new scaling law and a new chip architecture capable, on paper, of taking its semiconductors to a lithographic process node equivalent to 1.4 nm by 2031. Huawei’s plan is to continue improving the performance and density of its chips despite the restrictions At the moment the most advanced integrated circuits that TSMC, Intel or Samsung produce are 2 nm. Huawei’s plan is to continue improving the performance and density of its chips despite restrictions that limit China’s access to the most advanced semiconductor manufacturing equipment. And the heart of their strategy is the “tau scaling law.” This principle seeks to reduce the time it takes for signals and data to travel through computer chips and equipment. It proposes a paradigm shift that replaces the traditional geometric miniaturization of transistors with temporal scaling (τ), hence its name. It seems like a very complicated strategy, but it’s actually reasonably simple. We can easily understand what it is by referring to this example. Let’s imagine that we have a city (the chip) with many buildings (transistors) connected by roads (wires). Moore’s Law says: “Make buildings smaller to fit more in the same space“. Huawei, however, proposes: “buildings can no longer be much smaller, so instead let’s make cars (electrical signals) travel faster on the roads, and redesign the urban layout so that they travel less distance.” τ (tau) is, precisely, the time it takes a car to go from one building to another, and Huawei’s bet is to reduce it as much as possible. Huawei’s LogicFolding architecture plays an essential role in this approach. And, if we continue with our example, it proposes a new design of the roads on which cars circulate, so that the chip will perform better without the need to build smaller buildings. Huawei has anticipated that its next generation of Kirin chips, which will arrive next fall, will be the first to implement the LogicFolding architecture. Whatever its Chinese competitors do, TSMC will continue to do very well in the short and medium term. But in the long term his current leadership is not guaranteed. Image | TSMC More information | SCMP In Xataka | The condemnation that afflicts China: after decades of manufacturing a competitive desktop processor, it is six years behind

AI chips have always wanted to become more and more powerful. TSMC has just pointed out the true limit: efficiency

by

More performance? It is the first thing we usually ask of a new chip, almost without thinking about it. We have done it for years with the processors in our devices and we do it now with the chips that support much of the deployment of AI. More computing power, more speed, more scope to do things that previously seemed out of reach. But this logic begins to encounter a very specific limit: energy. What is making its way now is a less flashy idea, but increasingly difficult to ignore: progress will not only be measured by how much a chip calculates, but also by how much energy it needs to do it. The clearest clue comes from TSMC. We are talking about the largest contract chip manufacturer in the world, a company that does not sell processors under its own brand, but rather produces semiconductors designed by other players in the industry. According to ReutersKevin Zhang, senior vice president of business development, explained at a conference in Amsterdam that his customers are paying more and more attention to performance improvements that do not increase consumption. The pressure comes from very different profiles, from smartphone manufacturers to AI data center operators, all with a concern that we have seen growing in recent times: electricity cost and energy availability. The key is in the manufacturing. TSMC has not simply described a change in priorities. He has also placed it on his technological calendar with A14a future manufacturing technology planned around 2028. The firm expects that this process offers more than a 20% improvement in performance and, at the same time, reduces consumption by up to 30% compared to N2, the process that the company takes as a reference in that comparison. The key is that we are not talking about a specific processor, but rather the method with which subsequent chips can be manufactured. Not everything is about miniaturizing. For decades, reducing the size of transistors has been one of the great ways to gain performance and efficiency in chips. That logic doesn’t go away: transistor density remains within TSMC’s roadmap. What Zhang points out is that in the face of energy pressure from AI, other solutions, such as advanced packaging, chip stacking, and photonics, are also gaining weight. In parallel, as we pointed out a few weeks agoTSMC has decided not to use High-NA EUV, the lithography associated with ASML’s most advanced and ambitious equipment, in its A13 and A12 processes planned for 2029. The battle is also in the data. Huawei enters this conversation with Tau Scaling Lawa proposal that seeks to improve performance by accelerating the movement of data within the chips. The idea shifts part of the focus from the transistor to architecture and integration, two areas that gain weight when manufacturing smaller components is not enough. Along the same lines appears LogicFolding, which Huawei presents as a possible step beyond traditional 3D stacking, but which will depend on new design tools for folded architectures and better dissipation solutions for devices ranging from smartphones to AI data centers. Where are we going? TSMC does not speak for the entire industry, but its position makes the message carry. The firm suggests that, at least in its roadmap and in conversations with its clients, energy efficiency is gaining prominence that was previously more hidden behind performance. And it’s not a concern limited to AI data centers. Huawei, for its part, shows that the problem is also being addressed from architecture and integration, not just from the manufacturing process. The common point is not a closed conclusion, but an increasingly visible tension: chips will have to continue to be more capable, but each leap will be more difficult to justify if it increases consumption, heat or costs. Images | Xataka with Nano Banana In Xataka | Samsung has just achieved a milestone that has not been recorded for eight years. The problem is that it is a mirage

Intel has a plan to stand up to TSMC in 2027. First it has to survive 2026

by

During his almost four years at the helm of Intel, Pat Gelsinger stated on several occasions how important the semiconductor manufacturing business was for this company. In fact, many of the decisions he made pursued strengthen your competitiveness in a sector strongly dominated by Taiwanese integrated circuit manufacturer TSMC. However, the continuous delays in the manufacturing processes, the million-dollar losses and the drop in the stock market value of this company they ended up hastening his departure. Lip-Bu Tan, the current CEO of Intel, assumed leadership of the company in March 2025. At that time, a very uncertain future loomed over this company, and it was not at all clear what role Intel’s chip factories would play in the company’s global strategy. In fact, the leaks that predicted the possible segregation of semiconductor production plants into an independent company suggested that Lip-Bu Tan was willing to do without its chip factories. Fourteen months after his arrival to the general management, the panorama is very different. Integrated circuit manufacturing plants once again occupy a central position in Intel’s strategy. has confirmed it recently Lip-Bu Tan without the slightest ambiguity on CNBC’s Mad Money. From his statements it is clear that he aspires to consolidate Intel as the Western alternative to TSMC. And its cutting-edge nodes and Apple play a fundamental role in this ambitious plan. Node 18A is Intel’s best hope The most advanced integration technology Intel currently has in large-scale production is 18A lithography. In theory it is comparable or slightly superior to 2nm nodes from TSMC and Samsung. When Tan took the reins of the company, the performance of the 18A node it was not good. In fact, the outlook looked so bad that was forced to ask for help to some of its partners in the integrated circuit manufacturing ecosystem to analyze the data it had and find a way to optimize production and increase its competitiveness. “Performance” evaluates what percentage of the chips produced are working correctly. A low one triggers million-dollar losses An important note: in this context “performance” evaluates what percentage of the chips produced work correctly. Low performance triggers million-dollar losses. Tan has explained that the industry standard requires improving that performance by 7 to 8% each month, and has confirmed that now Intel is reaching that figure. There is no doubt that it is an unmistakable sign that the situation is changing. So much, in fact, that customers are starting to knock on the door. Intel has already closed agreements chip manufacturing with Tesla and Google. AND, as we told you at the beginning of this monthApple is exploring the possibility of Intel and Samsung manufacturing the advanced chips for their devices in the US. In all likelihood, the loss of influence and priority in the TSMC production chain that it has maintained for more than a decade has led to this decision. Now Nvidia has these privileges. There are several compelling reasons why Apple may be interested in Intel manufacturing its integrated circuits in the US. Or Samsung in its state-of-the-art plant in Texas. Or you could even work with both companies simultaneously and not completely break its business relationship with TSMC. Either way, this diversification strategy would allow Apple to effectively protect itself from supply chain disruptions triggered by geopolitical instability. And also the shortage of some components caused by the massive construction of data centers to artificial intelligence (AI). The next step will be node 14Athe integration technology with which Intel hopes to be able to compete head-to-head with TSMC in 2027 and 2028. Tesla has already confirmed which will order chips with this photolithography from Intel for its electric vehicles and robotics projects. Image | Intel More information | DigiTimes Asia In Xataka | Bill Gates has X-rayed Intel. And his diagnosis is overwhelmingly accurate.

China was supposed to be behind in chip-making equipment. Now its engraving technology is the standard and even TSMC uses it

by

Gerald Yin Zhiyao is the president and CEO of AMEC (Advanced Micro-Fabrication Equipment China), one of the largest chinese companies specialized in the design and production of the equipment involved in manufacturing integrated circuits. During a roundtable held at the end of July 2024 this veteran executive maintained that Chinese chip manufacturing equipment was at that time between 5 and 10 years behind its most advanced competitors in terms of quality and reliability. Yin Zhiyao is one of China’s leading experts in semiconductor production equipment manufacturing. He has not hesitated on several occasions to publicly adopt a critical stance when assessing the degree of development of the Chinese chip manufacturing machine industry, which is why his statements tend to be interesting to say the least. And the one he did last Sunday on Chinese state television, and which has been picked up by SCMPit was. According to Yin ZhiyaoAMEC’s ​​plasma etching technology has established itself as a standard in the integrated circuit industry and has been adopted by its major international rivals. In fact, according to the founder of AMEC, TSMC, the Taiwanese company that leads the chip manufacturing marketuses some of its machines in its production chain. It may seem like bravado, but it doesn’t have to be. It makes sense for TSMC to use semiconductor processing machines designed and manufactured by AMEC. What China has and what it doesn’t have Much of the sanctions deployed by the US Government seeks to put out of reach of Chinese companies the most advanced chip manufacturing equipment available on the market. In this scenario Yin Zhiyao holds something very reasonable: the US bans have accelerated the development of China’s semiconductor industry. In fact, at the end of 2023 the Xi Jinping Administration handed over to its main companies that are dedicated to the manufacture of semiconductor production equipment no less than 41 billion dollars. Photolithography and etching are two different stages that are repeated dozens of times during chip manufacturing Despite this effort, China still does not have extreme ultraviolet photolithography (UVE), which are suitable for manufacturing cutting-edge chips. At least not in large scale production. What it does have, as the head of AMEC states, are the engraving machines (etching) involved in the production of advanced integrated circuits. These devices are responsible for removing material from the exposed areas in order to physically sculpt the circuits on the silicon wafer. In this context, it is important that we keep in mind that photolithography and etching are two different and consecutive stages that are repeated dozens of times during chip manufacturing. Photolithography aims to transfer the geometric pattern that describes the circuit from a mask or template to the surface of the silicon wafer using extreme ultraviolet light. This is the stage that ASML machines on the edge nodes solve. Immediately after, the engraving process takes place, which can be wet or plasma. This last variant bombards the surface of the silicon wafer with an ionized gas that produces a plasma. This is precisely the process carried out by AMEC machines. Image | TSMC More information | SCMP In Xataka | China has responded to the US with a milestone: it already has an AI model capable of running on GPUs with different architectures

TSMC is focusing on the lucrative AI industry, so Apple is looking for a new girlfriend. Found it at Intel

by

In 2023, Apple completed its transition and made all Macs in its catalog work with its Apple M-series chips. It was the end of a relationship that began in 2005, when Steve Jobs announced the transition from IBM PowerPCs to Intel chips. Then things went wrong and Apple ended up separating itself from Intel in its products, but once again there has been an interesting twist of the events. Intel does not know how to be a foundry. The integration of the Apple M1 in 2020 was the real beginning of a logical strategy: Apple wanted to design its own chips for its Macs as it had already done in its iPhone or iPad, and the result was extraordinary. The curious thing is that Apple negotiated with Intel to manufacture the iPhone chips, but Intel rejected the opportunity. When Morris Chang, founder of TSMC, asked Cook why he hadn’t chosen Intel to make those chips, responded “Intel just doesn’t know how to be a foundry (chip factory)”. TSMC turns to AI. The relationship between Apple and TSMC has been one of the most important in the semiconductor industry. TSMC makes virtually all of Apple’s advanced chips, from iPhone processors to Mac M chips. That dependence, however, has become uncomfortable for two reasons: Chip shortage: The rise of AI has made the demand for this type of chips extraordinary. TSMC is prioritizing customers with the highest volume and margin in the most advanced nodes, and there Apple competes with NvidiaAMD and other manufacturers looking for 2 and 3 nm chips. Geopolitics: 90% of the world’s advanced chip production is in Taiwanand any escalation of existing tensions with China could paralyze the supply chains of Apple and the vast majority of technology firms on the planet. Diversifying manufacturers is therefore a strategic necessity. Intel gets interesting. It is true that Intel is not the only alternative that Apple was exploring, and Samsung was another candidate to work closely with the Cupertino firm. However Intel has a first important advantage with the 18A nodeits next-generation manufacturing process that experts consider comparable to TSMC’s 2nm process. Apple has been considering this node for entry-level M chips for months. Intel will not be manufacturing Apple’s most advanced chips at the moment, but this is a potential first step so that it can be verified that Intel can indeed accomplish the task and then also manufacture its most ambitious designs. Lip-Bu Tan turns the tables. Intel’s new CEO took over in early 2025and since then the company has taken promising steps when it was in a situation really worrying. has reached agreements with Nvidia to develop x86 chip sets with RTX graphics, for example. It also collaborates with Tesla to manufacture chips with an even more advanced node14A, for Elon Musk’s future TeraFab. Preliminary agreement. Official details of the deal are not yet known, but in The Wall Street Journal they claim that said agreement exists although it describes it as preliminary. It is not currently clear which chip or chips Intel will manufacture or in which photolithographic process. It is expected that the 18A node will be used for those entry-level M chips, but it is not ruled out that the 14A will not be part of this new commercial relationship. Be that as it may, if the agreement is closed as it seems, we would be facing a definitive boost to this new strategy of foundry traditional approach—manufacturing chips for third parties—that Intel is adopting. The circle closes. Intel lost the contract for iPhone chips because it refused to manufacture them for not having enough marginand thus passed up the opportunity to be a de facto partner in probably the most lucrative product in the history of technology. He then tried correct the errorbut he didn’t succeed. Then Intel would lose the Mac chip business, which would be another major setback. Now it seems to be taking flight again, and its promising future—along with other factors—have made Apple want to work with it again. It seems that Intel, after all, is learning to be a foundry. Image | Fortune CEO Initiative In Xataka | The US’s problem in the AI ​​and humanoid race is not China: it is all of Asia and it is greatly disadvantaged

Hormuz blockade is about to cause serious problems for Samsung and TSMC

by

The closing of Strait of Hormuz because of the conflict with Iran has turned the entire technology industry upside down and energy, beyond all the geopolitical tension that has been dragging on. It is an earthquake that runs through the entire semiconductor supply chainincluding key components that we do not have in mind a priori, but that are essential for the production of all types of microchips. From the most specialized gases to solvents, minerals and, essentially, all critical raw materials that are now much more complicated and expensive to obtain. Raw material. Apart from silicon, there are other essential raw materials for chip manufacturing that have recently been very difficult to obtain. Just like they count From Bloomberg, the production of these chips requires dozens of materials as specific as ultrapure gases, acids, solvents, resins… Many of which come from a very specific geographic region: the Middle East. The blockade of the Strait of Hormuz has suddenly cut off the supply of a good part of them, and although large manufacturers such as TSMC and Samsung have some accumulated inventory, the margin is narrowing with each passing week. Helium has no substitute. Helium is perhaps the most critical material of everyone. It is used to cool wafers during circuit etching, in EUV lithography processes, and to maintain the thermal stability of silicon. It has no substitute. Qatar produced about a third of the world’s supply, but Iranian attacks on its energy facilities in Ras Laffan and Mesaieed have paralyzed virtually all of its production. According to Bloomberga complete restoration could take up to five years. South Korea imported around 65% of its helium from Qatar, making Samsung and SK Hynix the most vulnerable manufacturers. Memory chips require much more helium than logic chips. bblunt, sulfuric acid and solvents. Beyond helium, the blockade is also affecting other equally critical materials. High-purity hydrogen bromide gas, essential for etching processes, is in short supply. High-purity sulfuric acid, used to clean wafers and remove photoresists once the circuits are printed, also is facing restrictions. Just like they explain In The Guardian, the Gulf exports about 45% of the world’s sulfur, the raw material from which it is obtained. And then there are solvents for photoresists, such as PGMEA, which is obtained from naphtha, a crude oil derivative that previously came largely from Iran. Inventories. The large manufacturers have come out to say that, for the moment, they have enough reserves to last several months. The South Korean government confirmed in April that bromine and helium inventories covered several months of production. TSMC, for its part, said it does not expect an immediate impact, although it warned that the prices of certain gases and chemicals will likely rise. The problem is that many of these materials have a limited useful life, since they cannot be stored indefinitely. Liquid helium evaporates during transport (especially now that ships must go around Africa), and photoresist solvents expire once opened. Jonathan Colehower, general manager of UST’s Global Operations and Supply Chain department, counted to PC Gamer that companies like Samsung “were operating with very tight inventories” following the just-in-time model, and that “this was not on their radar.” cgeographical concentration. One of the hardest lessons of this crisis is that the technology supply chain has very specific choke points. And it is not just about the Gulf producing oil; the thing is produces very specific materials in very specific installations that have no easy equivalent elsewhere. Jenna Ingram, Director of Proactive Intelligence at Exiger, counted told PC Gamer that manufacturers that previously bought helium from the Gulf are now competing for the same limited volume produced by Canada and the United States, which already had their own customers. It should be added that China has just restricted its exports of sulfuric acid and that Russia has imposed temporary controls on helium exports, making the picture even more complicated. Who will endure and who will not. In this scenario, size matters a lot. The big ones (TSMC, Samsung, SK Hynix) have enough purchasing power to sneak to the front of the queue when there is a shortage, pay premiums for high purity materials and draw on strategic relationships if necessary. The smallest ones, no. According to GartnerIf the situation continues, it could also affect the AI ​​industry. For consumers, the forecast is not at all hopeful. Supply priorities will most likely favor AI infrastructure over consumer electronics. How long will this last? “I think at best we are looking at another 12 or 18 months of difficulties. I don’t think this will reset overnight,” explained to the media Derek Lemke, Senior Vice President of Product Intelligence at Exiger. Colehower, for his part, explains that “a good part of the damage is not only an interruption of supply, but damaged infrastructure” that must be rebuilt. And, above all, he emphasizes that “prices are sticky. They go up, but they rarely go down.” Cover image | Harrison Broadbent In Xataka | China takes off in quantum computers: it already has the first dual-core and 200 qubits on the planet ready

wants to be the “TSMC” of data for robots

by

In recent years we have seen how artificial intelligence advanced on a relatively abundant raw material: text, images, videos and code published on the web. With robots, the terrain changes completely. We are not just talking about answering a question well or generating a convincing image, but about acting in the physical world, moving pieces, grabbing objects and doing so without everything being perfectly prepared. That difference explains why part of the next AI race may play out away from the usual focus. The investment. Settings It has not attracted the attention of just any investor, but of some of the large business groups in South Korea. According to Foley Hoagwhich legally advised Config on the operation, the startup, based in Seoul and San José, has closed a seed round of $27 million led by Samsung Venture Investment. ZER01NE Ventures, the investment arm of Hyundai Motor, LG Technology Ventures and SKT America have also participated. The operation values ​​the company at more than 200 million dollars and brings its total financing to 35 million. The “TSMC” of robots. The simile is not about chips, but about position in the value chain. Config aspires to position itself at a point similar to that of TSMC in semiconductors: not competing with its end customers, but rather supplying a part that others need to create their own products. In their case, that piece is not wafers or processors, but rather data for foundational robotics models. That approach is gaining traction as large manufacturers look to develop their own robotic AI without relying entirely on third-party vendors. Key difference. In a language model, the big cost is processing enormous amounts of digital information; in robotics, as Config CEO Minjoon Seo explained to TechCruncheach piece of data must be collected physically. That means having robots, spaces where they can be tested, and human teams that make them work. As companies look for more capable machines, data collection and labeling can quickly become more expensive, because we are no longer talking about information that waits on the web, but rather actions that occur in the real physical world. The key is in the conversion. The signature is based on an idea that is somewhat less obvious than the simple accumulation of data. Many robotics teams train their models with human motion data and then try to adapt them to machine behavior. The startup advocates another path: transforming that data before training begins so that it better fits the way robots move and interact with the environment. They have already started. Config has almost 300 people working on producing that data. The startup claims to have gathered more than 100,000 hours of human movement data, compared to roughly 3,000 hours for AgiBot Worldwhich the source presents as the largest comparable open set. The difference, more than 30 times, helps explain why the company is so insistent on the scale of its data operation. What’s coming. The next step will be to expand this machinery even more. Config wants to scale its operations in Vietnam and Seoul to reach one million hours of data collected, a goal that fits with its idea of ​​becoming an infrastructure provider for third parties. The company also aims to take its enterprise platform to $10 million in annual recurring revenue by the end of 2027. The third front is to launch a Robot-as-a-Service product in the cloud, designed so that companies can use the foundational Config model without depending on hardware integrated into the robot itself. Looking to the future. What this movement leaves is a fairly clear snapshot of where part of robotics may go in the coming years. Not everything will depend on the robot that we see in a factory, in a warehouse or in the field, but on all the previous work that allows human actions to be converted into useful learning for a machine. Config is still a young startup and its great promise has yet to be demonstrated at scale, but interest from Samsung, Hyundai, LG and other big names points to an idea with potential. Images | Config | Igor Omilaev In Xataka | Nvidia’s CEO is in China. And the future of your company is at stake there.

China is neither nor does it want to be in the 2nm war between TSMC, Intel and Samsung. Your plan to win is different

by

“Many people believe that competition in the semiconductor industry comes down to the advanced nodes and that we will only achieve success when we reach 2 or 3 nm. This is a misunderstanding“. This statement was made by Richard Chang Rujing, the founder of SMIC (Semiconductor Manufacturing International Corp), the largest Chinese semiconductor manufacturer with a global market share of about 5%. Rujing has spoken these words with a very clear intention: he wants China to strengthen its supply chain and its position in the global integrated circuit market by developing its manufacturing capacity for mature chips. Currently the most advanced integration technology that SMIC has in production is 7nm photolithography due to their inability to access equipment extreme ultraviolet lithography (VVE) of ASML. And there is no doubt that it and other Chinese chip manufacturers would benefit greatly from having the capacity to produce 5, 3 and 2 nm semiconductors. In this way they could compete on equal terms with TSMC, Intel and Samsung. However, there is something very important that we should not overlook: advanced nodes represent less than 20% of the world market of integrated circuits by product volume, while more than 80% of demand It comes from the segments of mature nodes and specialized processes. Rujing wants SMIC and the other Chinese chipmakers to invest more in their mature nodes, and it makes sense. After all, this is the strategy that is allowing China resist US pressure. Mature chips are the medicine the Chinese industry needs During the first two months of 2026, China exported integrated circuits worth $43.3 billionwhich represents an increase of 72.6% compared to the same period in 2025. This information comes directly from Chinese customs records, so it is presumably reliable. However, the most astonishing thing is that this country’s exports as a whole have grown by 21.8% during January and February, so it is evident that the semiconductor industry has been stimulated with much more intensity than other sectors. More than 80% of demand comes from the mature nodes and specialized processes segments Domestic demand has stimulated the growth of the Chinese chip industry in recent years, but the figures I have collected in the previous paragraph show that external demand is also very strong. In this context it is reasonable for us to ask what type are integrated circuits that Chinese manufacturers are mass producing. And the answer is very revealing: these are chips derived from mature integration technologiesusually 28 nm or less advanced. After all, the semiconductors that we mostly find in electronic devices, household appliances or cars, among other products, have been produced using them. Many Chinese chip manufacturers, such as Hua Hong Semiconductor, China Resources Microelectronics or Guangzhou ZenSemi, are manufacturing 28 nm integrated circuits or with even more mature technologies. And the Beijing Yandong Microelectronics (YDME) company is going to build a 4.6 billion dollar plant expressly to produce 28nm semiconductors on 300mm wafers. It is evident that these companies would not focus on the manufacturing of mature chips in this way if it were not a profitable strategy, and, above all, necessary to sustain the Chinese integrated circuit industry at a time as critical as the current one. Image | TSMC More information | SCMP In Xataka | China is preparing for the worst scenario: it fears that the US will prevent TSMC from delivering chips for cars and smartphones

Log In

Forgot password?

Forgot password?

Enter your account data and we will send you a link to reset your password.

Your password reset link appears to be invalid or expired.

Log in

Privacy Policy

Add to Collection

No Collections

Here you'll find all collections you've created before.