data

The camera that watches you while you drive is already mandatory in new cars. And no one guarantees what happens with that data.

by

Since yesterday, July 7, no new car can be registered in the European Union without the ADDW system (Advanced Driver Distraction Warning or Advanced Driver Distraction Warning), the final phase of the General Safety Regulation. We already had recently What does this system technically consist of? and why the car is going to beep at us more now than before. The melon that we had not yet fully opened was about what happens to the images captured by that camera that looks at your face while you drive. About what goes the regulations. He text of the regulation requires ADDW to function as a warning tool, not a recording tool. It analyzes the driver’s image in real time, decides whether or not the driver’s gaze is on the road and generates only a binary signal: attentive or distracted. In theory, the raw video never leaves the vehicle. Article 6(3) of the General Security Regulation itself establishes that the system must not record or retain data beyond what is strictly necessary to fulfill its function. The problem is what the standard does not say. “The regulations do not mandate any independent audit mechanism to verify that installed ADDW systems actually operate in a closed loop,” account the specialized media All About Cookies. It also does not define precisely what is considered “necessary” or how long that information can be retained before being deleted. In practice, the guarantee that your biometric data does not leave the car depends, today, on the manufacturer’s word. It is not the first time that word fails. The recent history of the sector does not help to generate confidence. An investigation by The New York Times revealed in 2024 that “several manufacturers, including General Motors, shared information about their drivers’ behavior with data brokers such as LexisNexis, which in turn they transferred it to insurance companies”, which came to translate into premium increases for drivers who did not even know they were being evaluated. For its part, a special investigation by Reuters documented that, between 2019 and 2022, “Tesla employees privately shared, through an internal messaging system, especially sensitive videos and images recorded by their customers’ car cameras,” including scenes of accidents, traffic fights and intimate moments captured in garages individuals. Neither case is directly related to ADDW nor did it occur in the European Union, but both perfectly show what can happen when an automaker accumulates sensitive data under vague rules. The road safety argument. On June 16, the Swedish data protection authority (IMY) ruled that the company Securitas Sverige had violated the GDPR by installing cameras with artificial intelligence that continuously analyzed the behavior of its drivers. ”The security argument alone is not enough to legally justify this type of behavioral surveillance,” they shared in Tech Times. Although this case is not specifically about ADDW, but rather about corporate fleet cameras, it also sets an uncomfortable precedent just when that same argument, that of road safety, is the legal basis of the entire European regulation. The other system, the black box. The EDR (Event Data Recorder), mandatory since July 2024 and of which we already detailed more in depthis only activated in the event of a collision, saves just a few seconds of technical data (speed, braking, steering wheel angle) and cannot be transmitted wirelessly, since it is necessary to physically connect to the OBD port of the vehicle to extract it. Its own design, designed to investigate accidents, is so limited that the road safety think tank ETSC came to complain that, by not being able to record location or time, the device is almost useless for investigators studying the causes of accidents. In other words, one falls short due to excess caution and the ADDW falls short in guarantees. And now what. The European Commission has already announced that it will continue developing the ADDW until 2027, incorporating the detection of cognitive and not just visual distractions, which implies even more sophisticated sensors pointing inside the cabin. For our part, we can review the manufacturer’s actual privacy policy (not the commercial website) to check how long the eye-tracking data is retained, if it is shared with insurers or third parties, and if the system ever leaves the vehicle itself. The law requires that the data not leave the car. Now, that it is fulfilled, it is a matter of trust. Cover image | Vitaly Gariev In Xataka | Waymo has not confirmed that it will arrive in Spain. But your documentation says things

let data centers pollute without limit

by

The summer of 2026 started strong with the first extreme heat wave. We have talked about the heat in Spain and throughout Europebut in the United States they have not been spared either of the consequences. With the air conditioners working at full capacity In homes, the electrical grid is strained and the Secretary of Energy made a decision that directly affects data centers. Activate your generators. The heat wave reached its peak last week and mainly hit the US East Coast. Given this situation, the secretary of energy Chris Wright asked data centers and other large customers to begin using their backup generators, which are normally idle. The objective of this measure is to relieve pressure on the electrical grid, which has experienced an increase in demand caused by high temperatures that have touched 40 degrees in many states. Why is it important. The order demonstrates how data centers are impacting the US power supply, especially in the mid-Atlantic area, where it is concentrated the largest number of data centers in the world. According to Data Center Map dataIn the United States there are more than 4,300 data centers and in Virginia alone there are 637. In other affected states there is also a high concentration: 133 in New York, 133 in Pennsylvania, 93 in North Carolina, 67 in New Jersey, 53 in Maryland and 42 in Massachusetts. Air quality. The biggest loser with this measure is the environment. We are talking about diesel and natural gas generators that emit pollutants such as carbon monoxide, nitrogen oxides and generate suspended particles. The use of these types of generators is limited by their effects on air quality, e.g. In Texas only its use is authorized provided that “the operating time does not exceed 10% of the annual operation of the main power source.” According to a Washington Post analysisin the state of Virginia there are at least 10,500 diesel generators installed. It is estimated that just using them for one hour a week has a health impact equivalent to that of five large gas plants combined and is associated with thousands of asthma episodes each year. Regulatory exception. The Department of Energy already had these limitations and In their directive they authorize connect these generators “without prejudice to air quality limitations.” It is a temporary decision designed to protect the stability of the network and the thermal comfort of millions of homes, but it does not come for free. The order sets an uncomfortable precedent: if environmental regulations are turned off every time there is extreme stress on the grid, the exception may end up becoming the new normal. Without batteries. The manager that offers supply in the area is PJM and last week it surpassed the demand record that occurred two decades ago. As they say in the New York TimesPJM does not have an extensive battery system that allows them to store energy to be used in situations like this. In other states such as California or Texas they do have this type of systems, which allows them to face heat waves or frosts without having to resort to measures of this type. Image | Generac In Xataka | Giant data centers have a space and power problem. A company wants to cut them up and distribute them around the houses

While most oppose AI data centers, there is one group enthusiastic about them: merchandise thieves

by

In the United States, at least a third of all data centers of the world. With investment reaching insane levelseach time they are built larger and more powerful infrastructures Unfortunately for the neighbors, who report an increase in pollution, increases in the electricity bill and problems with water supply. While all this is happening, there is a small sector of the population that has seen a business opportunity: thieves. What’s happening. They tell it in Business Insider. Last week in Illinois, police recovered two truckloads of stolen data center equipment. First they found a trailer containing $300,000 in coils of copper cable at a truck yard in Chicago. Later, they discovered that the same person who left the trailer had stored another one a week before and here came the bombshell: the second trailer contained material worth 1 million dollars. It is not the only theft of this type that has happened recently. At the beginning of the month they had The Canadian Press about $5 million worth of data center materials were disappearing while “in transit.” The containers left the port and were listed as having been picked up by the supposedly legitimate carriers, but never reached their destination. Why is it important. The data center boom is creating a supply chain of very valuable materials, especially DRAM chips and memories, but also metals such as copper. Organized crime is no stranger to this and has begun to attack these infrastructures that are so present in the US. Furthermore, they are using the very technology they are stealing from to achieve successful operations. Theft of merchandise. They say in Futurism that the US Department of Homeland Security estimates the theft of merchandise in losses of 35,000 million dollars a year. These gangs often attack retail trade, intercepting goods before they reach stores such as sports equipment from Nike. More sophisticated. Speaking to the Canadian Press, the head of the venture company Verisk CargoNet states that “The bad guys are very good at marketing (…) Now it is much more strategic, much more specific… They know what is in fashion and what sells well.” In the case of the 5 million theft that we mentioned before, the thieves used generative AI and other tools to impersonate a real transportation company and manipulate systems and communications. For days, it seemed like everything was going well and by the time they realized the deception it was too late. Image | Xataka with Magnific In Xataka | The iPhone is one of the most coveted pieces by street thieves. Apple wants to make it more difficult for you

The new star signing of AI is neither an engineer nor a data scientist, but he masters stoicism: the philosophers

by

A little less than a decade ago, studying philosophy implied a question: “And what are you going to live on?” Like many other races in the Humanities branchthe philosophy registered a low job insertion rate. According to published data in 2023 by EuropaPress, The unemployment rate for Philosophy graduates was 20 times higher than that of Electronic Engineering graduates. However, in the midst of the rise of AI, the companies that are training and evolving it have realized something: they do not need engineers to program, what they need is to hire philosophers who define how a model should think and how an AI that talks to millions of people every day should behave. Demand has reached a level where their salaries rival those of any senior engineer. From Socrates to defining AI. Studying philosophy was a risky bet due to its few professional opportunities (mainly teaching) and precarious salaries. But something has changed in the sector that was least expected: that of cutting-edge technology that was developing the AGI. According to collected In an article published in AtlanticIn 2013, only 1% of the offers published on PhilJobs, the academic job portal, mentioned artificial intelligence in the description of their offers. In 2025, that figure It was already close to 16%. And a good part of those positions were junior positions. That is, they join technology companies even with profiles with little experience. Why an AI company needs a philosopher. The reason for this change is that companies have made AI capable of processing data emulating the operation of a human neural network, but its interactions are with humansso their responses and decisions must be in tune with the ethical and moral values of humanity. Philosophers have been studying precisely that for centuries. Anthropic is perhaps the clearest case. Your philosopher Amanda Askell He leads the team that shapes the character of the model, and in January 2026 he published what the company itself calls the Claude constitutiona document of more than twenty thousand words that establishes the values ​​that the system must follow. As the company itself explains, this text is used directly in the training of the model. Askell counted to the magazine time that his way of approaching this work is as if he were dealing with a highly gifted child: “you have to be honest, because a smart child immediately detects when someone is lying to him.” Google opened the season to sign thinkers. Anthropic is not the only one that has incorporated philosophers into its staff. Google DeepMind took a similar step in April 2026. As and how I collected the university newspaper Varsitythe company announced the incorporation of Henry Shevlin, a philosopher of mind from the University of Cambridge, for a position that the company itself baptized, literally, as “Philosopher.” His work at DeepMind was to focus on issues of artificial consciousness, human-AI relations, and preparation for AGI. OpenAI has also taken note. OpenAI has followed a similar path, although less transparent in the details. Sam Altman has come to affirm that ChatGPT’s current responses “are the result of a consultation with ‘hundreds of experts’”, specifying that they were philosophers who have reflected on the ethics of technology and systems. In fact, even universities have jumped on the bandwagon and the American Philosophical Association (APA) has been delivering since 2024 two annual awards of 10,000 dollars for philosophical investigations on AI. Profession of the future, but a future that is too fast. Not everything is good news for the union. Daniel Fogal, professor of bioethics at New York University, told The Atlantic that this boom has a real distorting effect on the discipline. According to Fogal, there are philosophers who deep down do not want to dedicate themselves to AI, but feel that they have no choice if they want to enter the job market. The risk, he warns, is that a lot of mediocre work is published just to fit into a passing fad. Good philosophy takes time, Fogal summarizes, and rarely emerges as a direct response to the market. AI companies, on the other hand, launch new models every few months. The philosopher may be the star signing of the sector, but he will continue to be the person least comfortable with the rush. And perhaps for that very reason it is the one that is most needed in this development. In Xataka | We thought that leaving university and starting to work “on your own” was impossible: the key is knowing what to study Image | Unsplash (Sarah Sheedy)

Nvidia’s solution to reduce water consumption in data centers to zero aims to be genius: use hot water

by

Nvidia has announced a system of liquid cooling very special. Above all, because the concept of “refrigeration” is a bit confused here. The company has managed to reinvent this type of systems, and according to those responsible “the challenge of water consumption for data centers it is practically solved.” That’s saying a lot… or isn’t it? Jacuzzi cooling. The own Nvidia engineers They begin the description of their system by talking about how in jacuzzis the water is usually between 38 and 40 ºC, a temperature that makes most people last about 15 minutes in them. And the funny thing is that Nvidia’s new AI servers can use liquid cooling with water that is even hotter: up to 45ºC (113ºF). That is precisely the key to making the entire system efficient. Goodbye to fans. Nvidia’s new Rubin architecture claims to be the first in the world with an end-to-end liquid cooling system. At Nvidia they seal the server boards and eliminate noisy fans that shoot noise levels above 85 decibels. Elementary physics. The concept behind this idea is counterintuitive, but also brilliant. The chips generate so much heat that a liquid composed of 75% water and 25% propylene glycol entering at 45ºC is capable of absorbing the thermal load dissipated by these chips. The fluid absorbs that heat and ends up leaving the circuit at around 55ºC without the performance of the processor degrading at all. Good for 45ºC. The key is that Nvidia starts from that starting temperature of the water, which is 45º. In a liquid cooling circuit in PCs, the liquid is usually between 25 and 30 ºC. Here Nvidia manages to ensure that with this initial temperature the thermal difference with the outside air is high enough for the system to work passively in most temperate climates. The heat is expelled by gigantic external radiators. Water is only needed on the first fill. Even more interesting is the fact that this water circuit only needs to be filled once for the entire useful life of the plant, at least in theory. This eliminates traditional systems that use evaporative cooling towers. These systems consume enormous amounts of water, and according to Nvidia this makes it possible to cut water consumption of data centers by almost 100%. But. Although Nvidia’s idea is promising, the company only talks about what happens within the four walls of data centers. The impact internally in the data center is extraordinary, but what about outside? The problem, they explain on TechCrunchis that data centers they need a lot of energyand both the generation of that energy and the creation of the chips themselves used in data centers can fold either triple to the consumption of the data center itself. The gas and coal bill. Although renewable energies are increasingly covering a greater part of the needs in these data centersthe use of coal and natural gas will continue to be very notable in these facilities. According to the International Energy Association (IEA), these two sources will continue to represent 40% of the total used in AI data centers until at least 2030. And the generation of both types of energy requires vast quantities of water: natural gas plants they use 1.17 liters of water for every kWh of electricity they generate. Coal ones are even worse, requiring 2.2 liters per kWh. Good news, but not so good. The system devised by Nvidia can solve the problem within data centers and is even promising when it comes to reduce noise levels generated by these facilities. However, there are still equally important challenges to ensure that water consumption in other phases of this cycle is not so colossal. It is a good step, without a doubt, but the room for improvement is still notable in this area. In Xataka | Jensen Huang has taken a look at the idea of ​​putting data centers in space and has come to one conclusion: let’s not freak out

In 1999, a farmer donated his land for a children’s playground. Now the city council has sold them for millions for a data center

by

There are times when a good deed turns against you. Or against an entire town. In 1999, a Texan farmer named Charles Bland decided it was time to give something back to his town, a small town called Taylor. It occurred to him to donate about 35 hectares of his land to the city of 15,000 inhabitants with one condition: The city council had to use those lands to create a public park. To formalize the transaction, the city paid a symbolic amount of $10 to the farmer. 27 years later, the accounts come out in favor of the city council: it has sold part of the property not to create the promised park, but for a data center from the Blueprint company. The operation? one of 10 million dollars. And the neighbors were quick to light the torches. An offer that could not be refused This story is one of the most bizarre that we have seen in this matter of relationship between cities and data centers. The middle 404 has echoed the situation and the anger that is being aroused by both Bland’s initial objective and the fate of the land and, above all, the vague explanations of the city council is logical. The farmer initially donated his land to the Texas Parks Foundation, a nonprofit organization, with the goal of holding it in trust for future use exclusively as a playground. This is what appears in the documents and in the land deed, but there is a problem: in the following years, the property changed hands. The Parks Foundation gave it to a different nonprofit called the Williamson County Parks Foundation in 2003, but a month later, it gave it to the city itself. In 2008, the land changed hands again after a $15,000 sale to the Taylor Economic Development Corporation (TEDC). According to your page government, it is a non-profit corporation funded by the City of Taylor and, although it is a separate entity, it is the city council that appoints the five members of a Board that change terms every three years. Is this important? Well yes, since as 404 points out, it was this TEDC that, last year, sold part of the plot to Blueprinta company that plans to build a data center next to the town for those aforementioned 10 million euros. For a non-profit, the business has been successful. When they found out about both the sale and the plans to build a data center, the neighbors took to the streets to protest. Not only was the intention of that farmer who gave his land to the city with the sole interest of children having a place to play being violated, but an installation was going to be built that heats the air and that increase the electricity bill for the neighbors. Because of this, the TEDC has published a document in which he explains what is going to be done and what the city is going to get out of all this. They state that Blueprint’s data center will be used “for a wide variety of purposes including data storage, website hosting and artificial intelligence processing” in a facility that will have a total investment of 1,000 million dollars both to construct the main buildings as well as an electricity substation, backup generators and a closed-loop heat cooling system. It’s… too close to the city In a very brief way, The document addresses the complaints and concerns of neighbors referring to a report that ensures that the facilities will comply with everything necessary in terms of emissions and impact on air quality, noise pollution, light pollution, fire risk and water consumption. Always, of course, stating that the risk and consumption is minimal and that everything will go in the best possible way. Yes, the project will be connected to the electrical network for power supply. They expose that there is no way to go back and that, basically, that is what it is. Doesn’t anyone think about the children? The City Council does There will be no park, but it will be a great benefit for the city. As we see in Straight to the Palatethe facility will not occupy the 35 hectares donated by the family, but 21 hectares. The other 14 continue to belong to the city and the intention is to build a barrier to limit the impact that this data center may have on the neighbors. In addition, and as detailed in the aforementioned document, “the City expects to receive up to 30 million in total additional income that can be used to reduce property taxes and invest in streets, sidewalks, parks and other services”. With more conditionals on the list, they say that “the school district projects up to $20 million that could be used to improve facilities, increase teacher salaries and provide a better education for our students.” The truth is that the justification could well come from an episode of ‘The Simpsons’, but if we pay attention to that phrase that points out that, no matter what happens, the decision is irrevocable, the options for the neighbors do not seem very hopeful. The United States is currently experiencing two speeds in this artificial intelligence infrastructure. On the one hand, Big Tech raising monumental data centers. On the other hand, the people and political parties (both Republican and Democrat) that are rising in some states against these facilities. Studies are already appearing that leave aside tabout water consumption (minor one that was talked about a few months ago) as energy consumption (this one is tremendous) to point to the havoc on the environment what these “giant computers” cause. But hey, in the end Taylor won’t have a new park, but he will have a few million that could be used to make a park next to a giant radiator. In Xataka | AI has caused the collapse of even a non-AI industry: gas turbines

We already know how much water Amazon consumes in its data centers. We have good and bad news

by

amazon has made public the data of water consumption in their data centers. It is a remarkable exercise in transparency, especially because its division Amazon Web Services (AWS) It is currently the largest cloud infrastructure on the planet. We have bad news… and also good news. 2.5 billion gallons. The figure that stands out the most in this report is the 2.5 billion gallons (almost 9.5 billion liters) of water consumed by its servers per year. It is approximately 5% of the annual water consumption for the Seattle metropolitan area, so this is a notable figure, but one that must be put in perspective because there have already been previous studies that they misunderstood the situation. Golf courses are worse. The company already stated a few weeks ago that “data centers use significantly less water than golf courses or car washes.” They also gave other examples that consume more water, such as the meat production industry or the textile production industry. The efficiency metric. A metric called water use effectiveness (or WUE) is used to measure the impact of a data center. This magnitude indicates the liters consumed for each kilowatt-hour of energy delivered to the servers. According to Amazon, its WUE is 0.18 liters per kWh, much better than Microsoft’s 0.27 l/kWh and of course the 1.1 l/kWh from Google in some of its facilities. It is clear that not all hyperscalers manage to cool their data centers with the same efficiency. Amazon spends a lot of water in its data centers, yes, but according to current data it consumes significantly less than its rivals. Physics is physics. Data centers generate enormous amounts of waste heat, and Amazon indicates that they use direct evaporative cooling systems. Instead of maintaining huge air conditioning systems, the company uses mechanisms that introduce outside air and pass it through humid panels. This allows water to evaporate, absorb heat and cool those rooms where the servers are. The toll exists, of course: evaporated water is lost to the atmosphere and cannot be immediately reused in local ecosystems, which causes pressure drops in surrounding reserves during heat waves. Geography helps. Amazon’s efficiency advantage also benefits from intelligent geographic location of its data centers, the company says. Many of them are in regions with temperate or cold climates in the northern hemisphere. In these areas, cooling by outside air (free cooling) you can take advantage of more than 80% of the days of the year. In contrast, its competitors have often been forced to create their data centers in very hot desert environments which require injecting pressurized water constantly. Promises on one hand, criticism on the other. To compensate this impact due to your water consumptionAmazon has committed to returning more water than it consumes to local communities by 2030. This policy, dubbed “water positive“groups initiatives ranging from the restoration of watersheds to the creation of wastewater treatment plants. Of course, all this discourse faces strong criticism. For these voices, what Amazon is doing is a facelift, but it does not fix the immediate local shortage of wells from which data centers extract water in the middle of summer. The problem persists. Amazon’s report is welcome because it breaks the silence that usually prevails in Silicon Valley regarding natural resources. Even so, it also shows that the world faces a notable problem if the water (and energy) consumption problems generated by data centers cannot be solved. Taking into account the ambition and multimillion-dollar projects of AI companies In this sense, it would be important for official organizations to be in charge of monitoring and regulating these gigantic consumptions. Image | amazon In Xataka | Data centers do not want to depend on the conventional electrical grid. Solution: build your own plants

The world is running out of data to continue training AI. China has an ace up its sleeve

by

The models of artificial intelligence (AI) have a problem that more powerful chips cannot solve: they are running out of data. Epoch AI, a nonprofit research organization specializing in scaling AI models, warns with 80% certainty that the high-quality text available on the Internet will be exhausted sometime between 2026 and 2032. The reason is very simple: AI laboratories have been extracting everything the web has to offer for many years, and current models already train on data sets that approach the theoretical limit of the information available. When that gold mine empties, data volume scaling will stop working. And if this scenario occurs, AI development will most likely slow down. We still do not know what strategy US companies are developing to solve this problem, but we already know what is China preparing. His biggest rival. In fact, Xi Jinping’s government has decided that this shortage is an opportunity. This week the China National Data Administration published a draft outlining its action plan with a clear objective: to build an ecosystem of validated data by 2028 that will fuel the next generation of AI models. China’s bet is already on the table The document prepared by the National Data Administration identifies which specific sectors are priority objectives for information generation and certification. Some of them are scientific research, manufacturing, agriculture, energy, transportation, finance, healthcare, education and e-commerce. However, his plan does not stop at traditional sectors. China has a structural advantage that no Western laboratory can easily replicate And it also plans to cover cutting-edge fields with quality data, such as AI applied to robots, autonomous driving, low-altitude aviation or biomanufacturing. These are, precisely, domains whose data is not on the internet because they come from sensors, actuators and physical environments. Achieving them requires having industrial infrastructure, and in this scenario China has a structural advantage that no Western laboratory can easily replicate. However, this is not all. The document prepared by the National Data Administration explicitly encourages the expansion of the supply of text, code, images, audio and video necessary to train systems capable of complex reasoning, agentic behavior and control of intelligent robots. In fact, it’s an almost exact description of what the industry calls next-generation models. They are not just systems capable of answering questions; They will also be able to plan, act and operate in the physical world. The availability of high-quality multimodal data, especially that coming from real industrial environments, is today one of the least discussed and most determining bottlenecks in the AI ​​career. In a scenario where access to cutting-edge chips is restricted by US export controlsdata becomes a competitive advantage. If China can’t win the hardware race, it can try to win the fuel race that that hardware needs to be truly useful. Image | Daoducquan More information | SCMP In Xataka | The condemnation that afflicts China: after decades of manufacturing a competitive desktop processor, it is six years behind

Space data centers seem crazy. They make a lot more sense than it seems

by

“Space, the final frontier” became a classic pop culture phrase thanks to the series Star Trek. Now there are those who complete it with “… data centers”, because that is what Elon Musk certainly wants to achieve, and he has a plan to achieve it. At first glance it seems crazybut it turns out that the idea is not at all crazy. Free cooling, nothing. As explained in a very deep report in Semianalysismany analysts support the idea by defending erroneous premises. The space, for example, does not offer free cooling. Since there is no atmosphere, heat is not dissipated by convection, and huge and expensive thermal radiators are necessary. Solar energy is also interrupted in low orbits (LEO), so satellites must be placed in sun-synchronous orbits, a resource that is beginning to become saturated. The current cost does not compensate. The analysis carried out in this study for the Total Cost of Ownership (TCO) for a currently standard 30.5 kW cluster (with two servers with 16 Nvidia B300 GPUs) does not add up. Deploy this infrastructure In space it is necessary to invest 4.1 million dollars, when doing the same on Earth costs 1.4 million dollars. Space data centers are currently 260% more expensive than on the planet’s surface. Bad business. Space transportation makes everything more expensive. He biggest problem What affects these costs is the costs of transporting the material to space. In that proposed example, of the $3.1 million total cost of space infrastructure, $1.6 million is due to launch. But there is also the problem of the useful life of this data center: on Earth these facilities pay for themselves in 15 years, but in space wear and radiation in orbit reduce the operational life of the particular satellite to only five years, which multiplies those capital expenses dedicated to the project almost by 20. The first bottleneck is the chips. Even solving these problems, the main obstacle is simply semiconductor manufacturing capacity. The demand for TSMC’s N3 wafers and the supply of HBM memories is much higher than the supply even without this idea of ​​​​space data centers. That would add even more demand to an absolutely saturated system. But there is also the (lack of) energy. The reason why Musk wants to promote this idea as soon as possible is that obtaining power supply for terrestrial data centers is increasingly complicated. Thus, getting a connection to the electrical grid in Virgnia (USA) already takes seven years. Companies are creating their own power generation plants to solve this problem. Even so, according to the study, it will become increasingly more expensive to access this supply: they estimate that the cost of “terrestrial energy” will be above 20 million dollars per MW when this decade ends. That’s why Terafab. To solve this first bottleneck, Elon Musk has launched its colossal Terafab project in Austin. It is a huge chip manufacturing factory that will need 10 GW of electrical power to produce one million semiconductor wafers each month. The plan takes into account that 80% of the chips produced are destined precisely for space data centers. Starship changes the equation. But Starship stands in front of all these problems. SpaceX hopes to be able to reduce launch costs significantly in the coming years, going from the current $1,400-1,800 per kilo for the Falcon 9 to just $250 per kg for the Starship. This, together with the improvement in radiator and solar panel technology, will reduce the cost gap with terrestrial infrastructure. Now it is 260% more expensive, but at the beginning of the next decade it will be only 30% more expensive and will achieve economic parity by 2040. But. The accounts could therefore come out in the medium term, but it is necessary to take into account other factors as the so-called long-term computing cost. On Earth, between 3% and 6% of GPUs in data centers fail each year and require manual replacement by a technician. In space that option disappears, so it is necessary to oversize the satellites with 20% chips to provide redundancy and thus absorb potential radiation failures. In Xataka | Aragón is quietly becoming a data center “powerhouse” – now it has taken a crucial step

France has been determined to rob Spain of its position as a data center power in Europe

by

The French country has hit the table in its ambition to become a technological benchmark in Europe. He agreement reached between Emmanuel Macron and Masayoshi Son (CEO of SoftBank) aims to deploy up to 5 GW of computing capacity for AI data centers in northern France. This movement competes with all the projects that are underway in Spain, one of the countries that until now had attracted the greatest interest from hyperscalers. The problem is that neither France nor Spain will gain much from these initiatives. Nuclear counterattack. France has taken advantage your energy network —with a clear prominence of its nuclear power plants— to attract AI supercomputing projects. The SoftBank project will start in the Hauts-de-France region with an initial phase of 45 billion euros to build data centers in regions such as Dunkirk. In this first phase we want to achieve that the total capacity rise to 3.1 GW in 2031followed by a second phase that could reach 5 GW. Spain, data center paradise. Faced with this French movement, Spain has been closing agreements in that same area for months. It totals more than 22,000 million euros in recently announced projects. Giants like AWS (15.7 billion in Aragon), Microsoft (more than 7,000 million) and Blackstone have chosen our country to create these data centers. The Spanish advantage is its renewable energy productionwhich has attracted that type of investment. The harsh reality: Europe (probably) loses. Although both this announcement and those made in Spain are very striking, the reality for the Old Continent is quite stark. The data centers in Spain are not Spanish, and those in France are not French either. Europe is becoming the powerhouse for foreign multinationals that invest here because it suits them strategically. Energy resources are great for Microsoft, Amazon, Meta or Softbank, but the real benefit of this computing does not remain in Europe. The accounts. There is a clear difference between the strategies of Spain and France. Spanish soil is filled with hyperscalers like AWS or Microsoft that build, operate their own clouds and then control the flow completely. In the case of France, the initiative depends on a Japanese conglomerate allied with sovereign funds from the Middle East. SoftBank operates here more like a real estate developer– Create the data center and then rent it to third parties. Source: FT. Sovereignty, little. Emmanuel Macron and Pedro Sánchez can sell the message that these projects promote this ambition to have sovereign AI. The problem is that these data centers are simply delegations of big technology companies taking advantage of the advantages offered by their European partners. There may be options in the French project for the country to boost its AI companies —Mistral is the clear example—, but the truth is that these movements do little to help this objective of avoiding the independence of foreign technology companies. Rather they make the situation worse. The other European rivals. Europe’s traditional technology markets, grouped under the acronym FLAP-D (Frankfurt, London, Amsterdam, Paris and Dublin) are giving way to projects in other countries like France or Spain. There are also other protagonists in this new map of decentralized infrastructures: the Nordic countries are also interesting for their cold climates, ideal for helping to cool these centers. The real bottleneck. Beyond the billions of euros that are on the table, the big battle in the coming years will be access to hardware components, especially now that the memory crisis has made everything significantly more expensive. Demand far exceeds supply and it does not seem that this imbalance will be resolved soon, so all of these initiatives could suffer delays and changes in their final costs. In Xataka | Mistral does not generate hype, it is a discreet AI, it does not boost the shares of any company, but it already makes more money than Grok

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.