companies

25 technology companies come out in defense of open AI

by

Yesterday, July 24, 25 American technology companies and organizations (the discordant note is given by the French Mistral) as powerful as Nvidia, Microsoft or Meta published an open letter addressed to the White House that they later shared on their networks. Jensen Huang either Satya NadellaCEOs respectively of Nvidia and Microsoft. All of these entities have a common goal: asking the US government not to restrict open-source AI models. The document warns that these restrictions could “slow down competition or take innovation outside the country” What do they ask for?. That the US government encourages and protects the open AI ecosystem instead of hindering it and they do so with clear requests: No prohibitions or premature restrictions on open models. Guarantee access to computing capacity for academic research and startups, not just large companies. Investment in shared training resources such as databases, evaluation and security tools or frameworks. Legislate based on real and proven harms, not hypothetical risks. That the regulations focus on specific misuse, instead of assuming that what is closed is safer by default. Why is it important. Because this debate about open or closed models of artificial intelligence will determine who controls AI in the next decade, that is, whether it will be concentrated in a small oligopoly of large companies or in an ecosystem made up of thousands of organizations. This will directly affect the costs of AI adoption for SMEs, the competitive ability of the United States against China or how the benefits of the pie are shared. Security is mentioned in the letter and is an essential topic: depending exclusively on closed models is not a guarantee of security, since they are subject to failures or attacks that may go unnoticed outside the company. The transparency of open models allows a broader community to identify and fix bugs. Context. The moment chosen to publish this letter comes shortly after the Chinese startup Moonshot AI launched Kimi K3, a true missile to the waterline of the main American models (attending to the evidence) that brings with it a reality: China can win the AI ​​battle and do so with an open model. The United States’ response is to kill flies with cannon fire: it considers restricting access to Chinese AI models for reasons of national security and intellectual property. Furthermore, Kimi K3 arrived with an accusation under his arm: the director of technology policy Michael Kratsios accused Moonshot AI having “distilled” Anthropic’s Fable model to develop Kimi K3. In detail. The 25 signatories are Nvidia, Microsoft, Meta, Palantir, Dell Technologies, IBM, Hugging Face, Mozilla, Mistral, Andreessen Horowitz, Y Combinator, American Innovators Network, Arcee AI, Arena, Black Forest Labs, Box, CrowdStrike, Emergence Capital, The Linux Foundation, Mariana Minerals, Perplexity, Reflection, Replit, ServiceNow and Telnyx. Just as important as who’s on the list is who’s not: OpenAI and Anthropic, both valued at nearly $1 trillion and considering IPOs, don’t appear. There is no Google either. Because this debate is not just about national security, but also about two ways of understanding business and who feels like they are winning. A section worth focusing on is illegal distillation, a technique that allows knowledge to be transferred from a large and complex model to a smaller and lighter one to achieve similar performance with fewer resources. The signatories acknowledge their concerns, but ask politicians not to confuse legitimate techniques with the theft of intellectual property. Yes, but. Undoubtedly, opening the models allows for a better distribution of technological power and democratizes access to AI, but it is worth remembering that this list of signatories is not neutral: there are component manufacturing companies such as Nvidia or capital funds that have direct interests in the open models prospering because their business depends on the infrastructure and tools to deploy them. In Xataka | AI companies are buying tons of old books because they are free of AI Slop. Then they destroy them In Xataka | More and more US companies are using AI models like Kimi K3. The United States is not happy at all Cover | Gemini

AI companies are buying tons of old books because they are free of AI Slop. Then they destroy them

by

The impact of AI is no longer only in digital terms, but it affects frameworks as seemingly little linked to the virtual environment as second-hand books. Booksellers around the world are detecting the purchase of batches of books focused on very specific points. By crossing the data, the intentions are guessed, and again the AI ​​and its voracious need to be fed with data in an almost gargantuan way is behind it. Buyer robots. Marçal Font runs the Fènix bookstore, in Badalona, ​​and has been receiving purchases for weeks from a Canadian company that suspects it is automated: consecutive orders from the same buyer separated by just a minute. About twenty Spanish second-hand bookstores have sold to the same company since the end of April, some with orders of more than a thousand copies, almost always out-of-print non-fiction: they talk about a monograph on the castellers of Granollers from the seventies, a technical winemaking manual, conference minutes from half a century ago, diaries from the Civil War… All over the world. The same pattern is repeated in Germany, the United States, New Zealand and Australia: a Silicon Valley company has placed an order for more than 3,000 books with a Spanish bookstore. What distinguishes this type of buyers from ordinary customers is the indifference about price and the absence of thematic coherence. another example: An American bookseller went from selling twenty books in a good week to hundreds, with purchases that ignored the market value of the copy and an eye for subtlety, jumping from one topic to another with no apparent relationship. On the forums of Alibris, another book-selling marketplace, a platform official attributed the uptick to the arrival of new wholesale buyers who are hoarding commercial books. Two companies, one destination. The buyer identified in Spain is ZoomBooks, a Canadian second-hand book buying and selling company that sends orders to an Illinois logistics operator, PrepFort, in charge of scanning and cataloging the shipments. When asked, ZoomBooks denies collaborating directly with Anthropic, the company behind AI assistant Claude, and defines itself as a second-hand bookstore dedicated to recycling. Regarding the final destination of the books, and about the publications that its own website published about how to feed algorithms with used copies and destroy them later, the company responds that it does not comment on its commercial agreements because they are subject to confidentiality. Photo by Ugur Akdemir on Unsplash In parallel, another type of intermediary operates, more explicit about its function: ISBNdb, a bibliographic metadata database that now offers AI laboratories the purchase of between a thousand and a million books on request. Your own website puts it crudely: printed books before 2022 are the reserve of human text that no internet tracking can any longer guarantee, protected against contamination by texts generated by AI itself that now flood the network. Forced destruction. This commercial offer is supported by a specific ruling. Federal Judge William Alsup ruled in June 2025, in the case Bartz v. Anthropicthat training language models with legally purchased books constitutes a transformative use protected by the doctrine of fair use. He also considered it legal to digitize those copiesbut only because the printed copy disappeared in the process. That is, the digital copy takes the place of the paper book. Put another way: if the physical book survived the scan, the company that bought it would keep two copies having paid for one, and the argument of fair use would weaken. The shredder is a mandatory step in the legal process. Jason Leung on Unsplash Secret project. Months before that ruling, Anthropic was already carrying out a large-scale purchasing operation known internally as Project Panama. An internal company document, uncovered by The Washington Post based on declassified judicial documents, thus defined the action: “Scan Destructively all the books in the world.” The same document asked that the project remain hidden. ISBNdb has turned that same caution into a sales argument: it offers its clients, AI companies, a confidentiality agreement with each order and bluntly admits the image problem caused by the mass destruction of books. No headline about destroyed books generates sympathy, or as they themselves say, “destroying books gives a bad image“. What worries booksellers. Spanish second-hand booksellers have alerted the Ministry of Culture of what is happening. Miguel Ángel Ortega, president of the UNILIBER antiquarian booksellers association, recalls that his union fulfills functions of preservation and conservation of bibliographic heritage in addition to buying and selling, and it is contradictory for him to sell copies only for them to end up destroyed. Font puts it without half measures: “we are facing a form of literary plunder.” What is already protected by legal deposit and well cataloged runs less risk. But there are also “fanzines, neighborhood newsletters, local publications, documentation of social movements…” according to Font. Xavier Vinaixa, a researcher who helped uncover the case, and analyst Antonio Ortiz agree that the industry increasingly needs human text not contaminated by AI to avoid the collapse of the model, although Ortiz clarifies that the scarcity of data weighs less today than it did two years ago because training relies increasingly on reinforcement learning on already qualified data. Researcher Patrícia Ventura talks about how knowledge has to remain a public good, not a private reserve of a handful of companies. Image | Prateek Katyal in Unsplash

It took 15 years for computers to improve the productivity of companies. AI may take longer for one reason: sabotage

by

We keep hearing that AI is going to replace employees in their jobs and that this will mean an apocalypse in the labor market. While it is true that a paradigm shift is coming, even the most ominous voices they have toned down their speech by slowing down the real impact about employment. Four decades ago something similar happened with the personal computer. The offices were filled with computers that did the work of various accountants, but the companies’ balance sheets they didn’t notice its benefits nothing for years. Just like collect Fortunea Goldman Sachs economist has just reviewed that history with a magnifying glass and her conclusion is that AI could take even longer than computers to begin to work. demonstrate its benefits. The J-curve of PCs. Elsie Peng, economist at Goldman Sachs, has analyzed What happened after the arrival of the PC in 1981. The data suggests that in the first four years of the implementation of computers, productivity fell a little. Then, that productivity stagnated another four. Only in the eighth year did the first signs of growth in productivity attributable to the use of these new technologies begin to be noticed. Its peak came twelve years after its release. The trend that the implementation of computers in companies registered is known as J curve and, before computers, it was already experienced by other disruptive technologies in the labor market, such as the steam engine and the arrival of electricity. If AI is “the new PC”, that curve would place the first signs of impact around 2030, and its peak performance would not arrive until 2034. It’s not AI, it’s teleworking. From Goldman Sachs They admit that much of the impact of current AI is still not noticeable in official figures. This happens despite the huge investment figures that companies are making in AI systems. In the eighties with the PC it happened exactly the same: Chips and hardware were expensive and applications that brought real value, like the Internet, took years to have a large enough critical mass of users to have an impact on real productivity. In fact, according to the investigations conducted by Stanford expert Nick Bloom, the recent increase in productivity that has been recorded would not respond to the arrival of AI, but to the massive implementation of teleworking after the Covid-19 pandemic. You don’t invest in technology, you invest in processes. According to the data it collects Fortune According to the Goldman Satch report, for every dollar spent on hardware by companies in the 1990s, another dollar and seventy cents was needed for something that is not always taken into account: redesigning how people worked. In other words, it was not enough to fill office tables with computers, but all processes had to be redesigned and staff trained to make the technology have a visible impact. This expense in reorganizing did not take off until a decade after the arrival of the PC to the offices. The companies that changed their way of working won, not those that bought the hardware first. Investment in data centers for AI is growing faster than then. But the expense of reorganizing work with AI it goes slower than in the nineties with PCs. A survey by the Federal Reserve Bank of Atlanta estimated at about 280 billion dollars in intangible spending linked to AI in 2026. When the employee says no. However, the Goldman Satch analyst highlights a big difference between the implementation scenario of computers and that of AI: computers did not have workers against them. This scenario may still delay that peak productivity expected for 2034. Work reorganization is not done on its own, but depends on the employees. If a large part of the workforce resists, the implementation becomes complicated. A study of Writer and Workplace Intelligence to 2,400 workers found that 29% of employees sabotage in some way your company’s AI strategy. Among young people, the figure rises to 44%. Other survey of WalkMe points in the same direction. 54% of workers consciously avoided using AI tools at least once in the last 30 days. They preferred to do the work themselves. Harvard researchers they have named it. They call it self-disruptive technology. Employees do not reject AI because it fails, but because they feel that threatens your job. The study revealed that at least 30% of generative AI projects will end up abandoned due to this employee rejection. In Xataka | We thought that AI was going to take our position. The reality is that it is making us work more and rest less Image | Unsplash (Flipsnack)

Pedrerol leaves Atresmedia for Mediaset after 13 years, and his departure is the perfect reflection of the current war between private companies

by

We have seen this before: Josep Pedrerol shows up at the door of a chain at just the worst moment for the one hosting him to lose him. Well it’s happened again. Only this time the chain where it lands is already waging an open judicial war with which it is saying goodbye. The hostilities embodied in lawsuits and (future) trials between Atresmedia and Mediaset find in Pedrerol’s escape almost a symbol of the tensions that plague Spanish television. What has happened? Atresmedia and Josep Pedrerol They announced last Tuesday that end their professional relationship. There is talk of mutual agreement and a very specific expiration date: the Catalan presenter’s contract expires on July 19, and ‘El Chiringuito’ says goodbye a day later with a special dedicated to the final of the 2026 World Cup. Pedrerol summarized the balance by saying that “they have been thirteen fantastic years in which we have worked with absolute freedom.” On behalf of the chain, the general director of Atresmedia Audiovisual, José Antonio Antón, thanked in a statement the professionalism of the team and the achievements achieved. The outing is not limited to the evening program. Also closing is ‘Jugones’, the desktop news program that Pedrerol has presented on laSexta since 2013 and which, according to the statement itself, has remained a “reference for sports news at that time.” This is how Pedrerol says goodbye. The story has an almost identical precedent. Pedrerol directed ‘Punto Pelota’ on Intereconomía when the chain, in the midst of a default crisis, He terminated his contract on December 4, 2013.. By then he had already presented “Jugones” on laSexta, since September of that same year, which allowed him to land on Atresmedia shortly after: the heir to “Punto Pelota” debuted as “El Chiringuito de Jugones” on January 6, 2014 on the now defunct channel Nitro, before going through laSexta and Neox and settling on Mega since 2015. This time there are no defaults or termination, but a contract that has not been renewed after months of stalled talks. The piggyback format. For a presenter to change channels is normal on Spanish television, but it is not so normal for him to carry an entire format on his back, with equipment and production company included. The precedent most cited is that of Pablo Motos, who moved ‘El Hormiguero’ from Cuatro to Antena 3 in 2011 maintaining program and collaborators. Pedrerol starts from a similar position: ‘El Chiringuito’ is produced by Radio Sport Plus, its own production company, and according to Digital Journalist Citing sources in the sector, Mediaset is also negotiating to incorporate a good part of its usual team to transfer the format almost entirely, although under a different name. Mediaset is in a hurry. Telecinco closed the 2025-2026 season with an 8.9% average share, its worst historical record and the fifth consecutive season with a decline, while Atresmedia led the group with a 25.6% share. sharethe greatest advantage it has ever taken from Mediaset. Coinciding with the World Cup semi-finals on La 1, Telecinco signed the worst day in 36 years of history. A format with a low production cost and an already formed community of followers fits with what Mediaset needs before La Liga starts on August 15. It is still ironic, of course, that the same sport that is giving Telecinco so many headaches could serve as a rescue raft in the near future. War on. Mediaset and Atresmedia have been in litigation for years over ‘Pasapalabra’ and its final test, the Rosco: the Supreme Court forced Antena 3 to withdraw the Rosco in May, Atresmedia replaced it with a new test called AlaZ on June 19 (which has continued to triumph in hearings), and the Dutch production company MC&F and Mediaset They have announced that they will sue her considering that the new test is too similar to the original. Taking away one of its most profitable sports brands from Atresmedia would have something symbolic. No confirmation at the moment. Neither of the two networks has confirmed the signing, and there are still many pending details: whether the new program will land on Telecinco or Cuatro, with how many historical collaborators it will do so and what it will be called (a possible clue: on April 13, Radio Sport Plus record before the Spanish Patent and Trademark Office the name “Chiringuito Gol TV”). Again, August 15 and the start of La Liga can be a key date so that, around it, we know more things. In Xataka | Netflix users love to watch the first season of a series. Then they love to stop seeing it completely

It serves companies both to cut back and to expand

by

Asha Sharma, new CEO of Xbox, announced this week the dismissal of 1,600 people from its division. It is the first tranche of a plan that foresees more cuts, up to 3,200, this year. Three days later, the US Federal Reserve announced his appointment to a group that advises on “employment and productivity in the age of AI.” Black moon emoji. She is accompanied by Marc Andreesse and a Stanford economist who works with Anthropic. The irony: whoever has just decided what work is left over in their company will now have an opinion on what work is left over, in general. The numbers that Sharma presented in your statement are incontestable: Xbox has lost 64 cents for every dollar invested in small, independent studios, with margins three to ten times worse than any comparable business. But the statement says, almost in the same sentence, that these positions are not taken away by AI, and that the company reorients people and investment towards its AI priorities. The staff is told one thing. To the investor, the opposite. And here is the almond tree trick: AI has become the perfect alibi. It serves to justify both the more aggressive expansion of one company and the more honest surrender of another.. It no longer describes a technology. It is an absolution of universal validity. The pattern is repeated everywhere, always with the same verb: reorient, never replace. Amazon: 16,000 layoffs in its second round in three months, on top of another 14,000 in October, while it puts 200,000 million into AI infrastructure this year. Goal: 10% of the workforce out while spending on data centers skyrockets. Google: has quietly emptied part of Cloud (including the unit that sells cybersecurity as an argument for trusting its cloud) claiming that “we must reinvest in growth areas, such as AI.” Cloudflare: 1,100 out “preparing for the agentic era.” The dismissal itself no longer says much. The clue is where they get the money from: they do not cut where the AI ​​already does the work, but where the business has a worse multiple, less future story. The AI ​​does not execute the layoff, but decides which division survives the scissors. And it is not even the same movement in all cases. SAP has frozen hiring to finance its “significant bet on AI” while Its stock has plummeted 49% in one yearits CEO has said that he doesn’t know if in two or three years anyone in his company will still be programming. Intel has done just the opposite: it admits that it is no longer among the top 10 in the sector, that it is late against NVIDIA, and it fires 20% to retreat to on-device AI, away from data centers. You bet everything. The other gives up. They both call it the same: “AI strategy”. It will be or it will be. But the label does not describe what is going to happen, but rather what needs to be said today so that no one keeps asking. STMicroelectronics announced 2,800 departures within a plan that started in 2024, just before “IA” was the joker universal and yet the press release found space to mention it. The restructuring would have come the same. The label is new. What these companies buy with their layoffs is not, yet, the productivity that AI promises. It is credit against a market that In June it punished Microsoft with its worst month since the dotcom bubblefor not seeming committed enough, for having only thrown one ball into the matter and not both. Layoff is the entry toll to continue telling the story that the technology is going to work. And there is a place where this story meets flesh and blood: in Bethesda, HR ordered the removal of a small memorial that the colleagues themselves had left with photos of those fired. It did not fit into the environment that the company wanted to project.. So the next time someone tells you that they fire “because of AI”, or that they don’t fire “because of AI”, the question we should ask ourselves is who decides what counts as a healthy business, when the diagnosis is signed by whoever benefits from it. In Xataka | GPT-5.6 is probably the best AI model in the world. And precisely for that reason, the majority does not need it. Featured image | Xataka

Making a robotic hand is “100 times more difficult” than building the entire robot. These Chinese companies are determined to solve it

by

China has achieved let your humanoid robots run, they fight and even that execute choreographies with astonishing precision. That was the easy part. The key for a humanoid to stop being a fairground attraction and become a useful product is not in its legs, balance or its “brain”, but in something much smaller and much more complicated than it seems: the hands. The challenge of the hands. The human hand has 27 bones, 34 muscles and countless nerve endings. Replicating this in metal and circuits is the biggest bottleneck facing modern robotics. sums it up well Guardian Zhou Yong, founder of LinkerBot, one of the most advanced Chinese startups in this field: manufacturing a robotic hand is “a hundred times more difficult” than manufacturing an entire humanoid. “The dexterity required by a hand is ten times greater than that of any other part of the body, but its volume is only one-tenth the volume,” Zhou says. That is, not only is it the most technically complex element, it is also the smallest. The Chinese advantage. Pan Yunzhe, founder of Wuji Technology, studied in the United States and considered setting up his robotic hands company there, but soon saw that it was unviable. He tells The Guardian that “It was practically impossible to manufacture hardware in the United States due to the enormous limitations of the supply chain.” And China has an unmatched supply chain: it is agile, it is sophisticated and, above all, it is cheaper. This gives them a key advantage in hardware and allows companies like Linkerbot to already manufacture 5,000 robotic hands per month, a figure impossible to match anywhere else in the world. The problem is the software. Making the hand is only half the job, you also have to make it move like a human hand and there is still a long way to go here. Nathan Lepora, professor of robotics and AI at the University of Bristol, sums it up: “the challenge of making these hands is already being solved,” but controlling them “is a completely different game… no one knows how to do it yet.” The scarcity of data is one of the big problems and it is that, while LLMs have been trained with the enormous amounts of data on the Internet, there is hardly any data on how a human hand moves and, above all, what it feels like when touching something. To try to overcome this obstacle, Wuji Technology is testing a glove packed with sensors that captures all the movements of the human hand in everyday tasks. Its founder admits that being able to capture “how a person moves and what they touch or feel” is an “extremely complex and unsolved” task. The robotics market. The Chinese robotic hands sector had a turnover of $7.4 billion last year, almost four times more than in 2024. LinkerBot, one of the leading startups, aspires to a valuation of $6 billion. The case of this startup is just one example of the robotics boom in China, where more than a million companies are already registered, 40% more than last year. This translates into brutal dominance: China already manufactures 90% of the world’s humanoid robots and it is the country with the most industrial robotsby much difference with the rest. Yes, but. Robotics is advancing by leaps and bounds, but there are voices calling for calm, and not just any voice, but that of the International Federation of Robotics. In your report published in September last year were clear: “True multipurpose humanoids are still far away.” We are going to continue seeing more and more amazing demonstrations, but from there until they are sold en masse and we can all have one at home that is truly capable, there is a way to go. Image | Xataka with Magnific In Xataka | We still don’t know if humanoid robots will be the next great technological revolution. Yes we know that China will lead it

Elon Musk’s two companies merge because Wall Street loves simplicity

by

SpaceX is no longer SpaceX and xAI is no longer xAI. Instead, the company has decided to merge both names, and from now on it will be called SpaceXAI. That new name makes one thing very clear: the company is selling itself to Wall Street as an AI company that also launches rockets, not the other way around. A fusion that was sung. SpaceX bought xAI —and with it, both the Grok AI model and the social network X— in early February. He did so in a 100% stock move that valued SpaceX at $1 trillion and xAI at $250 billion. The name change is above all a marketing “punchline” about that de facto merger. The strategy has as one of its probable arguments a simplification that will undoubtedly be liked on Wall Street: Musk has created many companies that seemed to operate independently, so consolidating them gives that vision of a unified purpose and objective. This is not just about image. After the merger that occurred in February there was a clear reason: the dream of orbital data centers. Musk has been talking for some time about how ground infrastructure can’t meet AI’s global electrical demand, and SpaceX has already asked the FCC for permission to deploy up to a million satellites that work like computing nodes in low orbit. Therefore, having SpaceX and xAI completely merged also by their name simplifies this entire ecosystem. Going public helps. The decision comes shortly after SpaceX debuted on the stock market in June with the largest IPO in history. It raised $75 billion and earned a valuation of $1.77 trillion. The milkmaid’s tale? Before going public, SpaceX spoke of the “Total Addressable Market” (TAM), an estimate of the total size of the business they could access if they captured 100% of the demand and the figure is colossal: 28.5 billion dollarsof which 26.5 billion would correspond to AI, 1.6 billion in connectivity (Starlink) and only 370,000 to the space segment. Part of the animation of the “fusion” between both names showed this aspect. Grok and Cursor as pieces of the future. Grok continues and will continue to operate under the SpaceXAI umbrella, and of course the infrastructure agreements already signed are also maintained. The most important, the one that Anthropic recently signed and for which will pay 1,250 million dollars monthly to SpaceXAI for access to computing in the Colossus data centers. Google will pay 920 million monthly for the same. The other piece of the future is Cursor, the AI ​​agent for programming which is key so that the company can infiltrate companies. And Tesla, what? Since the merger with SpaceX was closed, there has been speculation on Wall Street with a plausible future: that Tesla will be the next to disappear as an independent company. SpaceX’s own president, Gwynne Shotwell, recognized the day of the IPO that there is a clear “convergence” between both companies, although he avoided talking about dates. Both are already collaborating on projects such as the ambitious Terafab, and Tesla maintains an investment of $2 billion in SpaceX which, after the merger with xAI, has already generated a capital gain on paper of around 64% due to the rise in the value of SpaceX shares. A very strong option. This “merger” with Tesla seems certainly likely. Wedbush consulting analyst Dan Ives esteem that there is an 80% probability that the movement will occur, and the Kalshi betting platform handles in these moments a 51% chance of that move arriving before May 2027. Some of the groundwork is already done in practice: both companies share engineers and both face bottlenecks in the form of power supply and cooling for their AI systems. In Xataka | “The idea of ​​making a cell phone makes me want to die,” said Musk. Two years later, it is very deep with its prototype of a mobile phone with AI

“Society will not tolerate that only a few companies do all the learning”

by

That things are not very good lately in the tech industry is a reality (depending on which side you are on, of course). The economy around the exacerbated demand for AI data centers It has become so devirtualized that it is no longer surprising that a major technology company has spent tens of billions of dollars on another big deal. And as a consequence of this, component shortage It is making the purchase of technological products by the consumer increasingly more complicated. So yes, you could say that things are not very there. But there’s also some comedy in Microsoft CEO Satya Nadella coming out to point this very thing out. And the company precisely contributes greatly to the situation we are experiencing. In an interview For the Wall Street Journal, Nadella warns that the current AI development model is neither sustainable nor legitimate in the eyes of society. What is this about? Nadella has long warned that AI has to generate real impact to justify the resources it consumes. Already He did it last January at the World Economic Forum in Davos, where he warned that if AI tokens do not improve tangible results in health, education or productivity, “social permission” to continue allocating energy and money to their development would be lost. Recently, in a similar speechhas dared to point out those who, according to him, are concentrating too much power. Concentration. For Nadella, a small group of companies (those that build the most advanced models, such as OpenAI, Anthropic or Google) are accumulating the value generated by AI while, at the same time, stirring up fear. And the conversation in recent years has revolved around topics such as massive job lossesthe existential risks about its use and about how these companies require almost unlimited resources to continue growing. “You can’t say that all white-collar jobs are going to disappear, that this could be a weapon, and at the same time use all the power available to build data centers,” counted the executive to the WSJ. Society is not going to tolerate a few models and a few companies “doing all the learning in the world,” he continued. “Narrative is not enough because now we have to demonstrate with facts,” he insisted to the medium. Who he points to without naming. Nadella does not mention specific companies in the interview, but the context says it all. Dario Amodei, CEO of Anthropic (and Microsoft partner with a multi-million dollar deal signed last year) predicted in 2025 that AI could eliminate half of jobs entry-level before 2029. Sam Altman, CEO of OpenAI (another long-standing Microsoft partner, in which the company has invested billions) has also made similar warnings about employment, although recently he admitted he was wrong in their predictions. Both companies have led to tensions with the United States Government regarding the safety of its models. What Microsoft is doing. Nadella also points out in the interview that Microsoft has launched a series of low-cost models to make access to AI cheaper for its enterprise clients, and has presented Copilot Coworkan autonomous AI agent that allows the user to choose between different models (including the cheapest ones) depending on the task. The WSJ points out In his article, the company is also considering whether to host a version of DeepSeek on its platform, a company that not long ago turned the technology industry upside down with its R1 model (it is also a company accused by OpenAI and Anthropic of having copied their models). The vision it proposes. For Nadella, the future of AI lies in a more distributed model, that is, companies using their own data, with access to a variety of models at different prices, without depending on a handful of suppliers. He defines the companies of the future as “continuous learning systems” that combine human knowledge and AI. In Nadella’s vision, a company’s capital would not only be its assets, but also its ability to process and learn, something he calls “token capital.” And he warns that protecting intellectual property will be key so that companies do not become mere executors of what the big models dictate. Between the lines. Nadella’s position also has a strategic reading. And Microsoft has not managed to develop its own model that competes with the most advanced ones from OpenAI, Anthropic or Google. Furthermore, according to share WSJ, its Copilot users have begun to prefer alternatives, according to data from the analysis firm Recon Analytics. Without its own header model, it is in its best interest for the market to move towards variety and price competition, and not towards consolidation around the most powerful models (which are, precisely, those of its partners). Cover image | Microsoft and M Rezaie In Xataka | “AI is killing my books”: Tim Ferriss has been selling productivity tips for years that ChatGPT now gives away for free

A job offer without salary is a half offer. These Spanish technology companies are the exception

by

Let me put you in a situation: you are looking for a job and you find an offer that broadly appeals to you, so you apply. Of course, the offer does not say how much you are going to charge, but you give it a chance. Your CV passes the first filter and you end up in some personal interview where they ask you the tricky question of “How much do you expect to earn?” and after dodging the bullet using Bill Gates trickthey end up revealing a salary range that doesn’t fit you. What a waste of time. Unfortunately, it has happened to (almost) everyone. Fewer companies do this transparency exercise than they should, but we have good news: in the absence of the Spanish government transposing the European pay transparency law to remedy it (it is late: the deadline expired on June 7), there is a list of technological employment companies that operate in Spain that do say how much they pay in their job offers. The pasta list. The Getmanfred platform created and maintains This public list on GitHub which Borja Pérez, Rebeca Méndez and Raúl Cotrina are in charge of reviewing and updating by hand by looking at the employment portals of each respective company: if the salary appears, they add to this list. There are more than fifty corporations, from Spanish startups like TaxDown or Landbot to large international companies like GitLab or DuckDuckGo, including Newtral, Mercadona Tech or PcComponentes. Not all of them are there, but they are all that are and if yours does not appear, you can always write to them to remedy it. Why it is important. Because information is power: when you don’t know how much a company pays, you’re at a disadvantage from the start. Furthermore, in general we tend to underestimate the salaries earned in other companies in the sector and this happens even more in low-paying jobs, according to this MIT study. Knowing the conditions from the beginning saves time, energy and disappointment. Salary opacity has a real social cost: in the Spanish state, women earn on average 4,781 euros less per year than men, a gap of 15.74% according to the INE with data from 2023. That salaries are a secret makes it much more difficult to detect and correct these inequalities. Context. Europe has been trying to resolve this by law for years. The Directive 2023/970 of the European Parliament approved in 2023 obliges companies to publish salary information and take corrective measures if their gender gap exceeds 5%. Spain has a previous rule, the Royal Decree 902/2020which required internal records, but the new directive goes much further. The interesting thing about Getmanfred’s project is that it includes the voluntary movement: companies that are ahead of mandatory requirements precisely because transparency (and the conditions) constitute an advantage for attracting talent. In detail. The repository is simple: a table with names and a link to their job portal: there is no automation or data scraping, but rather anyone can contribute to updating the list. In the table, the majority are tech with a remote work culture and Anglo-Saxon origin and just as important as those that are there are those that are not: large Spanish corporations are missing, since there is no trace of banks or traditional consulting firms. Yes, but. That a company makes public the salary of an offer could make one think that it is because it is an incentive, but this is not always the case. Come on, you are going to find some miserable salaries that show a reality: that of low salaries in Spain. On the other hand, there are “transparencies” that are of little use: if the range is “between 30,000 and 70,000 euros”, the information is not very valuable. Likewise, this directory also does not say anything about what happens within these companies: if they pay men and women the same, if the ranges they publish correspond to what they really offer or what criteria there are to move within the published range. In Xataka | If the question is in which country in the world the salary is highest, this graph has the answer: in Spain we could do worse In Xataka | The best paid jobs in Spain in 2026: from 56,000 euros for a doctor to 250,000 for directing private banking Cover | Ron Lach and Atlantic Ambience and Sora Shimazaki

We believed that Adobe was threatened with death by AI. It turns out that it is one of the few companies that is making money from it

by

In the last year Adobe has lost almost half of its value. Twelve months ago it was trading at $382. Today your actions They are listed at $197.. Wall Street has clearly punished a company that certainly seemed to be threatened with death by new generative AI models. The surprise is that Adobe is not only not doing badly with AI, but it is doing really well. Wall Street says one thing, the data another. In the second fiscal quarter of 2026, the company achieved some historical results with record revenues of $6.62 billion (up 13% year over year). Earnings per share (EPS) have shot up 18% to $5.96, and although the NASDAQ charts say otherwise, Adobe is not only not suffering, but appears to be stronger than ever. A money making machine. Adobe is in fact behaving like a true software empire. In 2025, it managed to have free cash flow of $9.85 billion on revenue of $23.8 billion. Its operating margins are also the envy of the sector: they stand at 37% (47% in terms non-GAAP). Its subscription business is absolutely exceptional, with annual recurring revenue (ARR) already reaching $27.1 billion. Others burn money with AI: ADobe earns it. While most companies do not stop investing money in the hope that the bet will pay off very profitable in the future, Adobe is ensuring that its generative AI tools are already having a positive impact on its results sheet. Income ARR Dependents on “AI First” solutions have tripled in just one year, and the big driver here is Firefly, Adobe’s generative AI platform. She alone has exceeded $250 million in ARR and continues to grow. Companies trust Adobe. We users may love using Gemini or ChatGPT to create AI images, but companies that relied on Adobe products for their creative workflows continue to do so. 75% of Fortune 500 companies already use Firefly, but the company has also trained more than 2,500 custom AI models for large corporate accounts. Users do not stop growing. Another advantage that AI has brought is that the technical barrier to using Photoshop or Premiere is lowered because the integrated AI assistants help users create what they wanted. There are also simplified tools like Adobe Express (70 million active users per month), and the freemium approach has been another reason for the number of users to grow: in just 12 months Adobe has gone from having 700 to 850 million users. The vast majority are still users of the free features, but Adobe it is enough to convert 2% to paying users to ensure that all these efforts are profitable. But the competition is tight. Although things seem to be going very well, Adobe now faces a fragmented ecosystem with a lot of competition. These new rivals also take advantage of AI to “attack” very specific use cases. It happens with Canva, for examplewhich has become a real threat and already has a turnover of 4,000 million dollars. Midjourney, Runway and even the Spanish Magnific are platforms that have fully embraced the AI ​​revolution to offer services to both individuals and companies. Be careful with subscriptions. Adobe is not immune to market fluctuations. The company stands out for maintaining extraordinary profitability in a very specialized niche, but it has taken questionable business decisions recently. The notable price increase of your subscriptions last year and its limits to generative credits have been decisions highly criticized. This has caused users to flee to cheap or even free alternatives such as Affinity or DaVinci Resolve. Uncertainty. There is another striking detail that can critically influence the future of the company. Adobe CEO Shantanu Narayen advertisement who was leaving the company after 18 years at the helm. There is no designated successor and this comes in the middle of that transition to a world full of generative AI solutions. The risk is therefore notable, and whoever takes the reins of the business will have to make very important decisions to not jeopardize the future of an already legendary technology company. In Xataka | Adobe was not born with Photoshop. It started by solving a huge and inconspicuous problem: printing well

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.