chips

is reinventing its AI chips from scratch

by

Cambricon Technologies is an essential company in China’s plans to challenge the US for its leadership in artificial intelligence (AI). Although it is not as well known as Huawei or Moore Threads, this is one of the companies specialized in the design of accelerators for AI with greater growth potential. Be that as it may, these three companies are China’s clearest alternatives to Nvidia because all three have already managed to place competitive solutions on the market. The priority strategy of the Government led by Xi Jinping seeks to build a self-sufficient ecosystem capable of breaking Nvidia’s dominance in the market. However, as stated SCMPat the center of this rivalry is a fundamental design debate: should China continue betting on GPUs or is it preferable for it to make the leap towards ASIC technology (Application-Specific Integrated Circuit or application-specific integrated circuit)? ASIC chips are designed to perform a single specific task, unlike GPUs and CPUs, which are general purpose. Its main advantage is efficiency. And since they are optimized for a specific function, they consume less energy and are faster in that task. Even so, they have a disadvantage: their rigidity. They cannot be reprogrammed to carry out another function, so the debate we raised a few lines above makes perfect sense. Convergence seems inevitable Large Chinese technology companies that choose ASIC chips for AI gain performance in their specific models, but are tied to an architecture that does not adapt well if the type of workload changes. This is the problem with this approach. a report prepared by Morgan Stanley and published on May 8, makes the market dynamics clear: it predicts that Huawei will capture 62% of the Chinese AI accelerator market in 2026, followed by Cambricon Technologies with 14%. ASIC chip heavyweights increasing relevance and volume in China Among the large technology companies with their own chips, Baidu and Alibaba are around 5% each. In any case, there is no doubt about one thing: the heavyweights of ASIC chips are increasing in relevance and its volume in China. And they are largely succeeding because the performance gap between Chinese chips and Nvidia GPUs allowed for export has narrowed noticeably. Morgan Stanley data reflect that Huawei’s Ascend 950 cards and Cambricon Technologies’ Siyuan 690 cards exceed the performance of Nvidia’s H20 GPU. Zhang Haijun, an expert semiconductor analyst, holds that as AI models become more complex the line between custom ASICs and flexible GPUs becomes increasingly blurred. This scenario suggests that the winning architecture could end up combining elements of both approaches. Su Lian Jye, the chief analyst at consulting firm Omdia, defend That companies with strong AI engineering capabilities and a clear roadmap benefit from ASICs, while those handling mixed workloads continue to lean toward general-purpose GPUs. For now, the market momentum in China clearly favors specialists. To companies that bet on ASIC technology. Partly by choice. Partly because the sanctions have left them no choice. Image | Enflame More information | SCMP In Xataka | The US remains committed to stopping China. Now it has targeted the second largest Chinese chip manufacturer

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

Gasoline hoses have a tiny hole at the end. Without chips involved, it is the smartest piece in the entire supplier

by

If you’ve ever paid attention to the pump while filling up, you may have noticed that it has a small hole located near the tip of the metal nozzle. That little hole is, possibly, the most ingenious piece of the entire set. And it is responsible for the hose “knowing” when to stop adding fuel and stopping on its own with that characteristic click. What exactly is it. This small hole is located at the end of the pipe (the part that you insert into the tank) and is connected to a thin, secondary tube that runs inside the nozzle parallel to the main fuel line. The nozzle uses the fuel itself that is being pumped to create the effect that activates the automatic cut. So to speak, the little hole does not pour gasoline, but rather breathes air. How it works. The key is in a physical principle called the Venturi effect. While the fuel flows at high speed through a narrowing of the duct, a low pressure zone is generated that sucks air through that small hole in the tip. The Venturi effect occurs because The density of gasoline is greater than that of airand it is precisely this phenomenon that causes the dispenser to turn off automatically when the tank is full. The moment of cutting. When the gasoline level inside the tank rises to cover that hole, the tube stops being able to suck in air. When the airflow is cut off, the suction is triggered and creates a vacuum that pulls on a flexible membrane (a diaphragm) housed in the handle of the nozzle. That movement releases a lever mechanism that slams the main valve shut, stopping fuel instantly. The pressure change causes the diaphragm to “jump”releasing the mechanical lever that closes the valve and ending with a click. And as you may have already noticed, the cut occurs even if you continue to pull the trigger. 100% mechanical. This entire system is purely mechanical. There are no electronic sensors, no chips, no batteries. The handle simply generates a slight vacuum at the tip of the pipe, and if that point becomes clogged, a mechanism closes the valve. It is basic physics applied to this little invention that we use in our routine, and that is capable of detecting even a small amount of fuel, blocking the hole to prevent it from overflowing. Security and cuts. This system prevents gasoline from overflowing from the tank, something that would be dangerous (risk of fire) and polluting. But its usefulness goes beyond safe filling. This extraordinary sensitivity is also the cause of those premature and repeated cuts when the jet turns off even though the tank is not full. The most common cause of these annoying cuts is simply a little gasoline splashing back and covers the hole momentarily, activating the mechanism ahead of time. In cars with short filler tubes, a rapid flow can easily flood that column, so the first recommended remedy is usually to reduce the filling rate. The position of the nozzle and the temperature of the fuel also play a role. In Xataka | The United States has the best electric car chargers in the world. Europe has something more important

China can’t buy the best Nvidia chips. So Alibaba has decided to connect theirs and sell them as if they were one

by

Alibaba does not want its infrastructure artificial intelligence (AI) continues to depend on Nvidia technologies. Little by little, the largest technology companies in China are assuming the request that Xi Jinping’s government made them at the beginning of October 2024: as far as possible They had to use chips produced in China. Ten months later this recommendation became a requirement. And the data centers that belong to the State throughout the country had to use at least 50% Chinese integrated circuits on their servers. This scenario especially favors Huawei, Moore Threads and Cambricon Technologies because they are Top AI GPU Manufacturers from China, but it also works great for Alibaba. In fact, Alibaba Cloud, its cloud computing subsidiary, has taken a very important step forward. A few days ago it presented a new chip for AI, the Zhenwu M890, and made official a very ambitious itinerary that describes what solutions it will develop over the next three years. This GPU has been designed by T-Head, the semiconductor division that Alibaba founded in 2018. It incorporates 144 GB of HBM3 memory and achieves an interconnection transfer speed between chips of up to 800 GB/s. As we are about to discover, this last feature is essential in the strategy that Alibaba has developed to compete in the AI ​​hardware market. Alibaba is going to spend $53 billion on its infrastructure According to Alibaba, the performance of its Zhenwu M890 chip is triple that of its predecessor. Additionally, it has been designed to perform well both during training of cutting-edge AI models and during inference. An important note: inference is broadly the computational process carried out by language models with the purpose of generating responses that correspond to the requests they receive. Alibaba wants to compete face to face with Nvidia in the deployment of infrastructure for data centers However, there is another relevant fact that is worth not overlooking: in medium precision operations (FP16) the Zhenwu M890 chip reaches 0.6 petaflops, a performance comparable to that of Nvidia’s A100 GPU and three times higher than that of the H20 chip. On the other hand, the ICN Switch interconnection chip allows link up to 128 GPUs M890 so that they work in unison. Alibaba assures that this architecture makes these GPUs work as a single chip, which, on paper, will allow it to compete head-to-head with Nvidia in the deployment of infrastructure for data centers. Regarding the itinerary that will follow until 2028, this Chinese company has anticipated that it plans to launch the Zhenwu V900 during the third quarter of 2027. According to Alibaba, it will implement its own significantly improved parallel computing architecture, will have three times the performance of the M890 chip, will be supported by 216 GB of memory and will reach an interconnection transfer speed of 1,200 GB/s. The Zhenwu J900 will arrive during the third quarter of 2028 with another major architectural leap. This roadmap It reflects that Alibaba goes all out. In fact, it has also announced that it will support this plan with an investment in 380 billion yuan (about $53 billion) over the next three years. Is the largest engagement of its kind in history of the company. Additionally, T-Head is planning its IPO to fund a more aggressive infrastructure investment program, which would put it in direct competition with Cambricon Technologies and Huawei’s Ascend line in the domestic AI chip market. Image | Alibaba More information | Alibaba | ChinaDaily In Xataka | Nvidia has to deal with the absolute distrust of several US legislators. Your plan in China is in danger In Xataka | The US wants to end Chinese AI chips sold abroad. And China knows how to defend itself

The United States promised to be very happy manufacturing its own chips. Nvidia just spent 150 billion in Taiwan

by

Houston, we have a problem. A couple of days ago the CEO of Nvidia stood on the stage at Computex in Taipei and said an inconvenient truth for the United States: “Taiwan is the epicenter of the AI ​​revolution. This is where chips and packaging are made. This is where systems are created. This is where AI supercomputers were created.” The setting was Computex 2026, Asia’s biggest tech event, and it wasn’t a compliment to the host, it’s a real depiction of how the industry works. It may sound paradoxical for an American company and at a time when The United States wants to reindustrialize with chipsbut he needs it. It is a structural issue. The harsh reality of profitability. Nvidia plans spend 150,000 million dollars a year in Taiwan, much more than the 100,000 million they spend now and with an abysmal difference compared to the 10,000 and 15,000 million five years ago. If it sounds silly, it’s because it is, but so is its billing: in the first fiscal quarter of 2026 billed 81.6 billion dollars, 85% more than the previous year in that same period. Also its benefit it’s already going off the charts: 58.3 billion, more than triple compared to the same period last year. That this money goes to Taiwan and not to the United States is due to technical and objective issues: Taiwan produces 90% of the most advanced chips in the world, according to a study by the Stimson Center. Of that Taiwanese production, TSMC controls 70% and is going to invest between 52,000 and 56,000 million this year. Bottom line: If Nvidia wants cutting-edge manufacturing capability, it has to be there. Why is it important. The best way to see it is to put Vera Rubin on the table, who In Huang’s words it is “probably the biggest product launch in Taiwan’s history.” Each system contains about two million parts and is assembled with 150 suppliers, almost all Taiwanese. This mechanism is not assembled by decree or in a legislature: it requires years and putting billions of dollars on the table. There is no factory in Arizona that can do something like this at least until 2030. Constellation will be Nvidia’s new headquarters in Taipei and will come to stay permanently: 4,000 engineering professionals will work in that center that according to Huang It will be operational by 2030. It is no longer that it buys in Taiwan, it is that the most valuable semiconductor company in the world is building the heart of its R&D in that core, an island 10,000 kilometers from the United States. A splash of cold water on Trump’s aspirations. Context. In January 2026, Taiwanese companies they committed to invest $250 billion in semiconductors and AI in the United States, as part of a trade agreement with Washington. Because Taiwan and the US are a symbiosis: each needs the other to maintain its position in the race for AI. The investment of a private company like Nvidia is another expression of this pact. In fact, Nvidia is not the only one: AMD is doing exactly the same: associate with Taiwanese manufacturers such as ASE, SPIL and Wiwynn with their Helios AI platform on the horizon (expected for the second half of 2026). That the two largest AI chip designers in the world strengthen ties with Taiwan is confirmation that the island’s industry is strategically necessary for the entire industry, not a particular bet by one firm. The elephant in the room: China. China’s role in this story is twofold: it is a threat and also a client. According to Reutersin 2026 Chinese companies have placed orders for more than two million units of the H200. Trade restrictions have made the operation difficult, but they have not been able to prevent it. One of the last cases point upon the arrival of a shipment of Nvidia AI chips to China via Japan. Nvidia lives in a contradiction from which it cannot escape: Its supply chain is on an island that China considers its own. China, which is its largest potential market, is blocked. Washington prohibits him from selling to Beijing while asking for independence from Taipei. And judging by his statements, Jensen Huang has bet everything on continuing to walk that wire. Yes, but. The Nvidia CEO forgot one problem in his speech: Taiwan makes the overwhelming majority of the world’s most advanced chips, but TSMC’s diversification into Arizona, Japan and Germany will not be ready before 2028 at best. That is to say, there are almost four years ahead in which Nvidia depends totally on Taiwan, a country that matters 97% of your energy. Furthermore, the atmosphere in the Strait of the same name is increasingly heated. Concentrating the production of its most critical component in a geographically hot spot is dangerous to say the least: if something explodes, there is no plan B. The closure of the Strait of Hormuz has reminded us of this the hard way. In Xataka | Huawei has found a way to counteract US sanctions: overcoming Moore’s Law In Xataka | US companies have always had a hard time making a lot of money in China. One industry is the exception: chips Cover | freepik and Jimmy Liao

Telecinco faces one of the toughest summers in its history. He has bet all his chips on Paz Padilla

by

Last August, Telecinco closed the month with an 8% audience share, the worst monthly figure in its history. In May 2026 it averages 9.2%: somewhat better, but it certainly does not provide a much better diagnosis of the state of suspended animation that the channel lives. Mediaset has been setting annual historical lows for four consecutive years, and the response for this summer has a name and surname: old acquaintances such as Carlos Lozano or Paz Padilla return to the channel, in a decision that makes it clear that Mediaset’s bet is not based on risk. Four years in free fall. The channel signed its historic low audience for the fourth consecutive year in 2025, with a 9.5% annual feeafter 2024 had already been the first below 10%. The problem is not only with the public, but with everything that comes with it: the net advertising income of Mediaset España fell 8% in 2025. For the first time, the group was surpassed in advertising billing by Atresmedia. To this we can add a structural crisis of the environment: Daily linear television consumption time fell by 7.7% in 2025, with an especially pronounced contraction among viewers aged 14 to 49. Telecinco competes, like all of them, against the changing habits of a generation more attentive to streamingTikTok and video games. Lozano, Padilla and other old acquaintances. Under the motto “Open for holidays”, Telecinco has presented its summer strategywhich includes the dating show ‘Love or whatever arises’ (with Carlos Lozano returning to the channel eight years after leaving it, reviving the spirit of ‘Women and men and vice versa’ in the same afternoon slot) or the daily version of ‘De friday’ called ‘Monday to Friday’with Santi Acosta and Beatriz Archidona. Finally, the entertainment format ‘The Peace Show’ with Paz Padilla at the helm, the weekend afternoons and reducing the ‘Party’ time. The presenter was fired by the channel in 2021 in quite tense circumstances, which says a lot about how both parties need each other, in a cessation of hostilities pact that must not have been easy. The changes will start in the second half of June. What goes away and what resists. To adjust this whole grid, there are programs that go on vacation. ‘Jorge’s Diary’ takes a break until September, coinciding with Jorge Javier Vázquez’s vacation. Hold ‘Allá tú’ with Juanra Bonet in the afternoons from Monday to Friday thanks to its good audience of close to 10% of share. The same thing happens with ‘The Fair Price’ by Carlos Sobera, which will also try out on the weekends, filling the gap left by the resounding failure of ‘Seen what has been seen‘. The goldmine of dating programs. Along with the realitiesand sometimes incurring almost unnatural matings between both variants, the dating shows They are the only thing that now invariably works for Telecinco. ‘The island of temptations‘has maintained competitive audiences during this four-year crisis; ‘Married at first sight’ was last season’s surprise. ‘Love or whatever arises’ tries to get away from this trend, and also with a specific format that has been tested previously with success. However, even if all these proposals work, they will be nothing more than mere patches to liven up the summer of a chain that needs a deeper reinvention. The return of Lozano or Padilla are symptomatic of the lack of originality of the resources of a Telecinco that survives on the basis of patches and proper names that drag audiences down. Because an ‘Island of Temptations’ is something that only happens from time to time: to get there you have to keep taking risks and, possibly, try formats that don’t smell like you’ve already seen them. In Xataka | The last bullet that Telecinco has left in the audience’s gun is a promo generated by AI and based on a TikTok success

NASA prepares chips 100 times more powerful

by

Human beings explore because they need to understand what lies beyond. We have done it by crossing oceans, climbing mountains and, for decades, sending machines to places where we cannot yet be. But a space mission has more to do than get there. For example, collecting data, interpreting it and sending it to Earth to do science. This is where the great challenge appears, because space requires computers capable of functioning for years in an environment that punishes electronics like few others. High Performance Spaceflight Computing. This is the name of the response that NASA is preparing. According to the agencythe project seeks to develop together with Microchip Technology a new space processor capable of offering up to 100 times more computing capacity than current space computers. We are not talking about a chip designed for a laptop or a mobile phone, but rather a system on a chip, or SoC, called to be integrated, once certified for space flight, in future ships, orbiters, rovers, manned habitats and deep space missions. SoC, a familiar term. This is the type of architecture that is common in our smartphones and tablets: small devices that concentrate essential elements of a computer in a single piece. The difference compared to an isolated processor is precisely there. An SoC is not limited to executing instructions, but can integrate CPUs, computing support units, advanced networks, memory, and input and output interfaces. On Earth we use it to gain efficiency and reduce size. In space, moreover, it has to survive. The challenge. As we say, space punishes electronics in a way that we rarely see down here. According to NASA, a processor intended for real missions must withstand electromagnetic radiation, extreme temperature fluctuations and high-energy particles capable of altering the operation of the systems. We are not just talking about losing performance, but about errors that can force a ship to enter “safe mode”, with non-essential operations turned off until mission teams resolve the incident. A key phase. Now comes the time to check if everything that is promised on paper holds up when taken to the physical field. JPL began testing in February and will maintain the campaign for several months, with radiation tests, thermal cycles, shocks and functional evaluations. The agency ensures that the processor is working as designed and adds a striking fact, although still within the framework of the tests: the first indications show that it operates with a performance 500 times higher than the radiation-hardened chips currently in use. More autonomy away from home. Space exploration has a limit that is not resolved with a larger antenna: distance. Between Earth and Mars, ua signal may take a while between 3 and 22 minutes to travel in one direction, depending on the position of both planets in their orbits. That means we can’t drive a rover like someone drives a remote-controlled car. We have seen it in the Martian landings, the famous “seven minutes of terror”, when a ship enters, descends and lands, executing a choreography by itself that from Earth we can only know when it has already happened. On-board computing. NASA proposes that this type of processor will allow future ships to use artificial intelligence to respond in real time to complex situations, analyze large volumes of data, store it and transmit it more quickly. Let’s remember the case of Perseverance which already combined orbital data of Mars, its panoramic camera and a Snapdragon 801 to compare what he saw with information obtained from space and refine his position on the Martian surface. If we want to continue exploring Mars and look further, we will need more and more systems capable of making decisions without always waiting for an order from Earth. Technology that returns. The history of space exploration is also the history of ideas that are born to solve very specific problems and then find a place on Earth. In this case, NASA points to possible adaptations for sectors such as aviation and automotive, in potential uses such as drones, electrical networks, medical equipment, communication services, artificial intelligence and data transmission. It does not mean that we will see this processor in a consumer product tomorrow, but it does mean that the effort to make it more powerful, efficient, scalable and resistant can go beyond a ship on its way to deep space. Images | POT In Xataka | The biggest problem with living on the Moon is its nights. NASA believes it has found the solution to avoid running out of electricity

Looking like a real miura in the gym is of no use if you do something afterwards: stuff yourself with chips

by

Something that is already quite internalized among society, precisely, is that abusing food ultra-processed is closely linked to obesity, type 2 diabetes and cardiovascular problems mainly due to the large presence of low quality fats. But although we think that this bad diet only translates into an increase in subcutaneous fat, medical technology has shown us that they are also filling our muscles with fat. A new study. This is precisely what a new study has pointed out. published in the magazine Radiology, which, Thanks to images obtained by magnetic resonance, researchers have discovered that the consumption of these products is related to an increase in intramuscular fat in the thighs. And it doesn’t matter how much we go to the gym for this. The evidence doesn’t lie. Much of nutritional studies are based on surveys and general measurements such as BMI or weight on the scale at certain times of the day. But here science has chosen to take an image to have completely objective data thanks to magnetic resonance imaging that objectively quantifies the internal composition of the muscle. The results. Here they could see that people who ate a higher proportion of ultra-processed foods had a significantly higher fraction of fat infiltrated into the thigh muscles. But most interesting of all is that, even after adjusting the results for total calorie intake, physical activity levels and other demographic variables. That is, it is not simply that people who eat ultra-processed foods eat more calories or exercise less, but there is something in the very nature of these foods, such as additives or the lack of fiber, which favors fat to be deposited directly between the muscle fibers. It’s clinical. This phenomenon of infiltration of adipose tissue into the muscle is clinically known as myosteatosis and, to visualize it, imagine the difference between a lean cut of meat and a marbled or fat-marbled steak. Why are we worried? Muscles are not only “motors” that allow us to move, but they are metabolically very active organs that are essential for regulating our blood glucose and even for the functioning of our brain. In this way, when fat infiltrates them, muscle quality plummets. And this is a big problem because logically we would begin to lose strength, sarcopenia would develop and there would be a risk of suffering a metabolic disease. Furthermore, in the context of this study, a poorer quality of the thigh muscles, especially the quadriceps, translates into greater overload of the knee joint and increases the chances of having knee osteoarthritis. There are nuances. As is usual, the researchers themselves point out that this is a cross-sectional analysis, so we do not have a “photograph” of the current moment and this means that there is no cause-effect relationship. That is, this study was done in older adults and people with previous illnesses, so these results cannot be generalized to the younger population and an independent study must be done with this sample to see exactly if a pizza from the supermarket can be too harmful. Images | senivpetro in Magnific In Xataka | Drinking coffee is not harmful, but for science there is a very clear limit that should not be exceeded

Corning has the solution to accelerate Nvidia chips even more

by

There are good reasons why a company of Nvidia’s stature would want to collaborate with a company like Corning, specialists in manufacturing the glass that protects our mobile phones. Corning offers more products than your Gorilla Glassand that is precisely what Jensen Huang’s company is interested in. And it is that Nvidia is going to invest about 3.2 billion dollars at the glass manufacturer with the intention of multiplying the optical connectivity production capacity on US soil tenfold. What’s on the table. The financial scope of the agreement has been revealed in parts. It was initially announced that Nvidia would receive warrants (stock purchase rights) to acquire up to 15 million Corning shares at a price of $180 per share, representing a potential investment of up to $3.2 billion. Added to this is a pre-financed warrant for another additional 500 million. But the CEO of Nvidia confirmed on CNBC that the company has also made “a prepayment of several billion dollars” to finance the construction of the new factories, a figure that was not part of the initial official announcement and whose exact amount has not been made public either. Fiber optics are the thing. The data centers that power AI They house hundreds of thousands of GPUs that must communicate with each other continuously and at high speed. For decades, this communication has been carried out using copper cables, and in fact for short-distance connections within the rack (from the server to the switch), they are still used, but fiber optics end up being superior in everything, both in terms of speed, energy consumption and lower signal loss. Which Nvidia has in mind. The technical term at the center of this agreement is co-packaged optics, which refers to the integration of glass fiber directly into chip systems, progressively replacing copper cables. Inside Nvidia rack systems (such as the Vera Rubin) there are currently about 5,000 copper cables that could be replaced by Corning fiber optics. Already at last year’s GTC, Huang rated this technology “essential for the deployment of AI.” The company has been preparing the ground for months: in March invested $4 billion in Coherent and Lumentumtwo companies specializing in lasers and components that convert data between light and electrical signals, which then travel through Corning fiber cables. Who else is in the race. Nvidia is not the only one betting on this technology. Its competitors Broadcom and Marvell They have already launched similar productsand Intel also develops its own co-integrated optics solutions. For its part, Corning already had Meta as a reference client. In fact, Zuckerberg’s company announced an agreement of up to 6 billion dollars for Corning to expand its optical cables plant in Hickory, North Carolina. The alliance with Nvidia now adds three more facilities and multiplies the company’s optical connectivity manufacturing capacity in the United States by ten, in addition to increasing its fiber production by more than 50%. ““Made in America”. The agreement comes precisely at a time when the Trump administration is pushing to relocate technological supply chains that have been built for decades in Asia (Taiwan, China or Vietnam). Huang counted to CNBC that “it is an extraordinary opportunity to reinvest and revitalize American manufacturing for the first time in generations.” According to the CEO of NVIDIA, the tech sector would not be the only one to benefit, since the construction and operation of these data centers generates demand for electricians, construction workers, chip manufacturing operators and infrastructure specialists. “The skilled worker shortage and demand are incredibly high,” mentioned Huang in the middle. Converted company. Corning has become another of those companies that have seen their business benefit from the AI ​​boom. And the signature accumulates an increase of more than 300% in the last year, driven by its repositioning towards the AI ​​market and moving away from its best-known image as a manufacturer of glass for mobile screens. In Xataka | If the question is whether using ChatGPT or Claude in English is more efficient and saves tokens, the answer is: yes

that Intel and Samsung manufacture their advanced chips in the US

by

Apple is no longer TSMC’s largest customer. It has been since 2014 and has maintained this position for just over a decade, but in 2025 Nvidia established itself as the main customer of this Taiwanese semiconductor manufacturer. Until December 31, 2025, Nvidia generated 19% of TSMC’s revenuecompared to 17% for Apple. Even so, there is no doubt that the Cupertino company continues to be a priority customer for TSMCalthough this scenario could change in the short or medium term. And, according to BloombergApple 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. However, to understand why Apple is considering leaning on Intel and Samsung, we also need to look in another direction. Betting on Intel or Samsung makes sense, although it involves risks 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 sever your 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). Apple is a candy for Intel and Samsung. There is no doubt about that In fact, last week Apple recognized during the presentation of its economic results that its current supply chain is not very flexible. And in the current situation of instability, having Intel and Samsung as partners is a very attractive option. Furthermore, for these two companies it is very important get Apple as a client. In 2024 this last company represented approximately 25.2% of annual income from TSMC. And in 2025 this figure dropped to 17%, allowing Nvidia to overtake Apple with 19% of income generated for TSMC. Still, Apple is a candy for Intel and Samsung. There is no doubt about that. For Intel, having Apple as a client would represent a definitive boost for its integrated circuit manufacturing division for third parties. And for Samsung to manufacture a part of Apple’s SoCs would be a very important boost that would probably allow it to attract other clients in the fight it has had with TSMC for years to dominate the cutting-edge chip manufacturing market. Whatever Apple executives decide, it is highly unlikely that this company will abandon TSMC. This Taiwanese chipmaker has proven for years that it is perfectly capable of manufacturing cutting-edge semiconductors on a large scale and with indisputable consistency. You don’t have to prove anything, but Intel and Samsung do because it’s currently unclear whether they can reliably deliver the kind of production and scale that has made them TSMC in the dominant custom chip maker. Apple may end up outsourcing part of the manufacturing of its SoCs to Intel or Samsung, but nothing indicates that it will break its already long-standing commercial alliance with TSMC in the medium term. Image | Apple More information | Bloomberg In Xataka | Apple had been able to maintain prices despite the crazy rise in RAM. That’s over

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.