startups

Xataka is Media Partner of VivaTech, the largest technology event in France with more than 14,000 startups

by

VivaTech It is one of the largest technology events in Europe in general and France in particular. Its next installment, VivaTech 2026, will take place between June 17 and 20 in Paris, the French capital, and From Xataka we have the pleasure of being Media Partner of this edition. It is going to be an event full of technology, panels and interviews and, as it could not be otherwise, from Xataka we will be covering it and experiencing it in situ. If you want to join us, we invite you to pay attention to our Instagram profilewhere we will share live everything we find in this (huge) tech meeting. A big edition Image | VivaTech Since 2016, VivaTech has been bringing together an entire ecosystem of companies, startups, influencers, media and technology fans. It takes place at the Paris Expo Porte de Versailles and has not stopped growing. To give some figures, in the first edition there were 45,000 visitors and more than 5,000 companies and startups. In the previous edition there were 180,000 attendees, 450 speakers, 300 announcements and launches, 14,000 startups, 4,000 partners and 3,600 investors. It is a sensational opportunity to discover first-hand how the global and European technological landscape breathes. It is also an important year for the event, which turns ten years old. For this reason, on Sunday, June 14, they are going to turn the legendary Parisian Champs-Elysées into a huge free experience, so that anyone can enjoy technology in a spectacular environment. Image | VivaTech The event also has top level speakers. To mention a few, personalities such as Yann LeCun (AMI Labs), Peter Steinberg (founder of OpenClaw), Henna Virkkunen (European Commission), Joe Tsai (Alibaba), Elizabeth Stone (Netflix), Bernard Arnault (CEO of LVMH) or Narendra Modi (Prime Minister of India) will attend VivaTech. Jensen Huang (Nvidia), Elon Musk (Tesla), Mark Zuckerberg (Meta) and Tim Cook (Apple), among many others, have also walked through its halls. The event, as we said, will take place in Paris between the days June 17 and 20. Tickets can be obtained now from the official website of the eventas well as the agenda with the presentations and the map with the stands. See you in Paris! Images | VivaTech

There is a new wave of startups creating new AI millionaires

by

The rise of AI has generated a new hypermillionaire saga who are breaking all limits of wealth to date. All you have to do is go through the list of 10 of the greatest fortunes in the world of Forbes to discover that eight of these great assets arise from this technology. However, this was only what Bloomberg called “the first wave”, in which the founders of the great generalist AI models such as OpenAI have risen. Anthropic either deepseek. Now is the time for specialized AI agents and their founders they are also getting rich. The 19 new barons of AI. As and as I pointed out BloombergAmerican AI startups have created 19 new billionaires in the last year with a combined fortune estimated at about $59.3 billion. These 19 new millionaires join the 41 founders who, thanks to the success of their AI models they had already become millionaires in the “first wave.” However, what is striking about this increase is not only the number, but the profiles of who are behind these million-dollar startups: a poet, three scholarship recipients from the Peter Thiel program without a university degree or a self-taught immigrant. AI agents are the new oil. Reflection AI It is one of the most obvious cases of this new wave of AI millionaires. The startup is dedicated to creating agents capable of programming, debugging and understanding code almost independently. This new market It has turned its founders into millionaires. It is estimated that Ioannis Antonoglou and Misha Laskin have achieved a fortune valued at around $4 billion each. However, the company did not emerge from nowhere, Antonoglou was part of the team that developed AlphaGo, from the revolutionary Google DeepMind model that achieved beat humans at Go. An AI wants to be your lawyer and your doctor. Without leaving aside AI agents, Harvey is another success story in this segment, allowing the automation of legal research, the drafting of legal documents and the review of contracts with AI. Founded by lawyer Winston Weinberg and AI researcher Gabe Pereyra while they shared a flat, its AI agent Harvey, named after the protagonist of the popular lawyer series Suits, has become one of the most used in companies and law firms. Each of its founders is the owner of an estimated fortune of 1.6 billion. In the healthcare field, OpenEvidence has followed a similar path. Its founder, Daniel Nadler, already sold the financial analysis platform Kensho to S&P Global in 2018 for $550 million. With OpenEvidence, it applied the same logic to the medical sector: its AI assistant has accumulated more than 100 million consultations and the company has almost quadrupled its valuation in six months to reach 12 billion, raising Nadler’s assets to 7.2 billion dollars at the beginning of 2026. The Thiel Fellows: from recruitment to labeling. Mercor is another example that was difficult to imagine just a few years ago. Their three foundersBrendan Foody, Adarsh ​​Hiremath and Surya Midha, met at a high school debate. The three classmates left the university to join the Thiel Scholarshipthe PayPal co-founder’s program that pays $250,000 to young people to leave their studies and start a company. Initially it was a recruiting platform, but they switched to data labeling providers for OpenAI and Anthropic, hiring doctors, engineers and scriptwriters to train AI models specialized in these areas. As a result of this change, Mercor went from earning 100 million in 2025 to 1,000 million at the beginning of 2026, with a valuation of 10 billion. That leaves each of the founders with an estimated fortune of $1.9 billion. The ecosystem versus the giants. Vercel is another example of how the startups that are succeeding in this second wave of unicorns emerged from AI. We are no longer talking about AI models, but about the infrastructure that allows deploying applications generated with AI. Its founder, Guillermo Rauchimmigrant Argentinian and self-taught who learned English by reading software manuals to learn how to program, turned a tool for developers into a very profitable platform that has given him an estimated fortune of more than 1.9 billion dollars In Xataka | We already know who has won the AI ​​race: the OpenAI employees who sold their shares Image | Brendan Foody

There are two Madrid startups that want to solve the logistical labyrinth of space

by

If receiving an order from Madrid to Castilleja de la Cuesta (Seville) can cause you some headaches in the form of a delivery person who never arrives, imagine sending a package from Madrid to the moon. Space logistics is one of the last major bottlenecks in the commercial aerospace industry. For decades, sending cargo to space has been reserved almost exclusively to government agencieswith astronomical budgets and opaque processes. Today the industry is more open than ever, the demand for space shipments is growing, but the logistics infrastructure that supports them remains artisanal and fragmented. Two Madrid startups, Usyncro and Eye4Skythey are trying to change it. A packet destined for space. Send a kilogram of cargo into space costs approximately 20,000 euros and that is just the beginning. That package in question has to go through customs in several countries, go through the hands of multiple carriers and comply with export regulations for sensitive material. ESA and NASA satellite components are subject to dual-use regulations that vary between jurisdictions and require specific licenses for each international transfer (e.g. export control regulations and laws such as NASA’s ITAR). And when you arrive at your destination there is no one to sign a receipt. There is also no warehouse or workers. Just a satellite in orbit waiting for a critical component on which a scientific mission depends. The presentations. Usyncro is a SaaS platform founded in Madrid in 2018 that digitizes international trade logistics through blockchain and artificial intelligence, connecting all the actors of a shipment in a single panel: carriers, customs and operators. Eye4Sky is a spin-off of the National Institute of Aerospace Technology (INTA). It was founded in 2022 by researchers with more than twenty years of experience in space optics. Manufactures polarization modulators based on liquid crystalsoptical devices the size of a spectacle lens and barely 200 grams that analyze light to obtain information about the solar magnetic field or the composition of the atmosphere of other planets, something that a conventional camera cannot do, at a lower cost than traditional instruments and equivalent performance. It operates from the Madrid Science Park, within the ESA BIC incubation program. Why is it important. What Usyncro and Eye4Sky are building goes beyond their own businesses because it points to a structural problem: managing the supply chain of a space mission has always been the territory of large contractors and government agencies. A traceable and standardized digital corridor could lower that barrier to entry. On the other hand, INTA is not a university, it is an organization attached to the Ministry of Defense with a long research tradition but little history of serving as a seed for commercial companies. That Eye4Sky is its first spin-off after decades of applied research represents a paradigm shift: institutions that have historically operated in public and military logic are beginning to open up to civilian commercialization. As for projects on the table, Eye4Sky modulators are already present on the solar observation satellite jointly developed by the European Space Agency and NASA Solar Orbiter, are confirmed in VigilESA’s first space weather mission and in the quest Talisman of Satlantis to detect methane in the Earth’s atmosphere. Usyncro, for its part, already certified via blockchain the launch of Hydra Space satellites and executed the first digital air cargo corridor between Europe and Latin America. The joint project would be the definitive leap: applying that same logic to the most complex logistics chain that exists. Context. Usyncro was a conventional logistics company specialized in coordinating land, air and sea transportation chains. Its value proposition was clear: digitize and centralize the management of complex logistics operations with multiple actors, eliminating the dispersion of information and manual processes. It worked well on land, but the sky is the limit. The turning point came when joining the Retech Digital Entrepreneurship Network of the Community of Madrid, whose aerospace node is located in Tres Cantos. There they met Eye4Skya company that manufactured components for ESA and NASA missions, but had no way to reliably and traceably manage its logistics chain to the satellite. Just what Usyncro knew how to do: manage complex logistics chains with multiple actors. Of course, this time the final destination is in orbit. How they do it. Usyncro is developing a digital logistics corridor, a system that centralizes the entire journey of merchandise in a single control panel, from when it leaves the factory until it reaches its orbital destination. Each party involved in the chain is recorded, each transaction generates a documentary record and at each node along the route images are captured that certify the status and position of the shipment. Blockchain technology guarantees the integrity of the data and reduces time in each phase of the process. The final delivery is certified automatically, without the need for a physical recipient. In essence, it is applying to space logistics the same logic that has transformed land logistics in the last decade: total visibility, real-time data and end-to-end traceability. Yes, but. The project is still in the testing phase. Usyncro and Eye4Sky are shipping material to different countries via multiple routes to validate that the system works under real conditions before scaling up to space missions. Digitizing terrestrial logistics is already a complex problem, but doing it for space cargo adds extra difficulty in issues such as legislation or handling conditions. The margins of error are practically zero. It remains to be seen whether the platform can withstand the operational, regulatory and technical pressure of a real mission before the first big test next year. As Delia Rodríguez, CEO of Eye4Sky, tells: “Our devices are the eye of missions that protect the Earth and that starting in 2027 will monitor from space the invisible shield that protects our planet.” In Xataka | Spanish technology in the return to the Moon: the system designed in Madrid that NASA will use in Artemis II In Xataka | We have been deceived by the distances of the Solar System: the … Read more

The mayor of Lisbon has turned it into a magnet for European startups

by

Carlos Moedas does not feel like a conventional politician. The current mayor of Lisbon and former European Commissioner for Innovation recently insisted in which his training as an engineer marked his management. The idea: less rhetoric and more structural solutions. And it certainly seems to carry out this proposal, because it is leading a transformation that seeks to position the Portuguese capital not only as a tourist destination, but as one of the most relevant technological nodes in Europe. unicorn factory. The nerve center of this transformation is the call Unicorn Factorya hub of innovation that was launched in 2022 and that has become the flagship project of its mandate. Since its creation, this initiative has multiplied its size by five, and today it already houses thematic divisions specialized in cutting-edge sectors such as AI, blockchainhealth or green technologies dedicated to the agricultural field. The success is tangible: 17 “unicorn” companies (valued at more than $1 billion) have moved their operations to the Portuguese capital. Attracting startups. Moedas explained in comments to Financial Times how the real challenge of Lisbon is not to see companies born, but to help them grow. The project attracted 300 companies in 2025, a figure that represents notable growth from 250 the previous year. Among them there are companies with founders from the US, Brazil, or Denmark. Favorable conditions. To attract these startups Moedas has used several levers. The first is to sell Lisbon for what it is: a safe capital with a great quality of life. The second is to highlight that the tax regime for qualified immigrants is advantageous, and there are also tax credits for R&D, which of course is a strong argument for companies that consume a lot of resources in those initial phases and can thus receive aid in this process. But. The road, Moedas openly acknowledges, is not without potholes. Portuguese bureaucracy remains complex – the same thing happens in Spain – and that can deter entrepreneurs. The mayor of Lisbon states in the interview that a good part of his time is spent talking on the phone with foreign founders and helping them overcome these bureaucratic barriers. “Politicians do not create jobs, the markets create them,” he repeats: their job is to facilitate, not hinder, or at least that is the message. Competence. Although the activity is notable, there are founders of companies who know that the market is raffling them off. María Ribeiro Soares, from Immersiv Studios, warned of the difficulty retaining talent compared to markets with higher wages such as the United Kingdom or Germany. The other venture capital. There is another peculiarity in this entrepreneurship market. Lisbon startup founders have realized that while traditional venture capital is cautious, the so-called family offices are very willing to bet on these projects. These investors, often experienced, high-net-worth former founders, provide funding but also advice. Lisbon is not alone. The rise of Lisbon as a technological node is also accompanied by other Portuguese innovation centers such as Braga and Porto, where, for example, the biotechnology sector flourishes. There is also a direct connection with universities oriented to this market, and in that northern region of Portugal alone there are already some 1,200 startups with a combined valuation of close to 9 billion euros, according to the FT. The other great challenge. Portugal’s efforts to attract external talent have led to the arrival of numerous digital nomads, but this has had a counterproductive effect: it is causing an exodus among young Portuguese. They cannot access a housing market adapted to the pockets of those nomads, who charge triple. Rental prices have skyrocketed, and that has created tension that the mayor must manage. Image | EU2017EE Estonian Presidency | Aayush Gupta In Xataka | Portugal’s radical proposal to stop touristification: an underwater cable that connects with the US

Chinese startups have been relying on NVIDIA chips to train their models for years. That is already changing

by

The name of the Chinese startup Zhipu AI (Z.ai) may not sound familiar to you, but perhaps GLM, its AI model, does a little more than its latest version, GLM-4.7already competes with Claude Sonnet 4.5 or GPT-5.1. The real surprise of this “Chinese AI tiger” is the launch of GLM-Image…and not so much for what he does, but for how he has managed to do it. what has happened. GLM-Image is a multimodal generative AI model that focuses on image generation. The idea, of course, is to compete with options like Nano Bananafrom Google. That’s interesting, but even more striking is the fact that the model has not been trained with conventional chips. Trained with Chinese chips. According to those responsible for Z.ai, this model is the first developed in China that has been fully trained with “local” chips. Specifically, it has been trained with Huawei’s Ascend chips thanks to the use of servers Huawei Ascend Atlas 800T A2 and a framework called MindSpore. Thus, traditional NVIDIA AI chips, which are usually the usual choice for AI model developers in Chinese startups, have not been used. Turning point? This milestone demonstrates the real feasibility of training high-performance generative AI models on a platform developed entirely in China. We are not dealing with something minor: it is validation that it is possible to continue innovating in this area despite the restrictions imposed by the US. In fact, Zhipu AI — included last year on the US blacklist — has intensified its collaboration with other local manufacturers, such as the promising firm Cambricon that has risen from the ashes thanks to tariffs. Threat to NVIDIA. The news comes at a unique time, because NVIDIA has not stopped pressuring the US government to once again allow it to sell its advanced AI chips to Chinese companies. He has obtained that permission—which It won’t be free—, but now the one that might not be interested is China, which he hasn’t said anything at all. That chips from companies like Huawei are a valid alternative for training quality AI models can change many things in this area. Zhipu goes like a shot. The Chinese startup has also just gone public, and since it has done so its shares they have shot up more than 80%. Investors see the company no longer as a rival to Google or OpenAI, but as a banner. One that shows that it is possible to compete without depending on the US and its companies. Huawei, great beneficiary. If the trend continues, Huawei can become the Chinese NVIDIA, and the company prepares an increase in production of its AI chips. It is not the only one: Cambricon plans triple your production by 2026, which seems to make it clear that the Chinese industrial machinery is moving quickly to neutralize the impact of US vetoes. Challenges…Despite everything, Zhipu already has warned that the price war in the AI ​​sector will become international. If Chinese companies end up controlling the entire chain (or rather, their chain), they could offer AI services at much lower costs than their Western competitors, who must pay NVIDIA’s margins and Big Tech’s cloud infrastructure. …and unknowns. This technological achievement raises other questions. One of the most important is how powerful and capable Huawei chips are compared to NVIDIA’s in these processes: is training much slower? Is it more expensive in time and resources? The efficiency of the MindSpore framework compared to Pytorch or TensorFlow is another of the key components of these developments. In Xataka | Faced with the US strategy, China has a plan to revive its technology industry: that AI belongs to everyone

Ten AI startups have skyrocketed their valuation by $1 trillion in 12 months

by

Logic tells us that companies that they lose money consistently should have a black future. What is happening in the world of AI is just the opposite, and right now ten startups in very red numbers They have achieved something unusual in one year: in one year they have grown by one trillion dollars in their joint valuation. It is something simply extraordinary… and disturbing. The big three. OpenAI is of course the protagonist of this select group, and today it is estimated that its valuation amounts to half a billion dollars. Elon Musk’s company, xAI, is valued at 200 billion, while Anthropic is also close to that figure according to a Financial Times study. In one year the valuation of these AI startups has skyrocketed. Source: Financial Times. And his immediate followers. Databricks, which was founded in 2013, was quick to join that segment and now has an estimated valuation of $100 billion. Figure (robotics), SSI (Sutskever’s startup), Scale AI, Perplexity, Thinking Machine Lab (Mira Murati’s startup) or Cursor complete this group of new startups (almost all of them) and with skyrocketing valuations. Investment fever. This growth in its valuation is due, of course, to the fact that all of these firms have raised multimillion-dollar investment rounds by firms that trust in a future full of AI. In fact, venture capital and investment companies in the US have injected 161 billion dollars throughout this year, and they have done so without being able to see even a hint that their bet is going to be a winner. All these AI companies They burn money like there’s no tomorrowand its profitability—and future—is an absolute unknown. bubbles are good. “Of course there is an (AI) bubble.” The person who says it is Hermant Taneja, president of the venture capital firm General Catalyst. His firm has invested in Anthropic and Mistral, and has done so without batting an eye because according to him, “Bubbles are good. Bubbles align capital and talent into a new trend, which causes some carnage, but also creates new lasting businesses that change the world.” Maybe it is, but only for a few.. Sam Altman, co-founder and CEO of OpenAI, also think there is a bubblebut it coincides with that positive vision because it is probably him who will benefit (if everything explodes). Robin Li, CEO of Baidu, already indicated a year ago that the bubble will end up bursting and that only 1% of companies will survive. Bezos adds to that perception: “this is the good type of industrial bubble that is totally contrary to financial bubbles. The banking bubble, the crisis of the banking system, that is simply bad, as happened in 2008. These bubbles are the ones that society must avoid.” It happened with dotcoms. The analogy with the dotcom bubble It’s inevitable. At that time something similar happened with the inflated valuations of internet companies, and when the bubble burst only a few survived, but those that did managed to become the mistresses of the world. ANDThis bubble is much bigger. At least, from the point of view of the figures invested. In the dotcom fever, venture capital companies invested 10.5 billion dollars, which if we adjust for inflation becomes about 20 billion dollars. In 2021, these same firms invested 135 billion in startups in the SaaS (Software as a Service) segment. This year, investment in AI companies will likely exceed $200 billion, according to PitchBook. One of the directors of these firms describes this with a strong word. “This is FOMO“. And the valuations are skyrocketing. Startups that have $5 million in annual recurring revenue are seeking investment rounds that value them at $500 million. That they pursue those valuations that are 100 times their income makes ridiculous the excesses that already occurred in 2021, for example. Although venture capital firms know that they will lose money on most of their bets, they also hope that one or two that they get right will more than make up for that entire bet. Not investing is losing forever. Mark Zuckerberg shares this vision of venture capital firms, and his company is making colossal investments to avoid missing out on AI. The founder and CEO of Meta recently explained that He doesn’t care about losing 200,000 million dollarsbecause it would be much worse to be left behind in this race. Marc Benioff, co-founder and CEO of Salesforce, agrees and believes that a trillion dollars of investment will end up wasted, but AI will end up producing 10 times that in value: “The only way we know to create great technology is to try as many things as possible, see which ones work, and then focus on the ones that succeed.” Time will tell if they were right and if this bubble, as investors defend, is “a good one.” In Xataka | OpenAI is making the tech industry unite its destiny with yours. For the sake of the global economy, it better work

His startups are born thinking of global

by

Factorial announced a few days ago that It has already reached 100 million dollars of recurring annual incomebecoming one of the twenties Scaleups European that achieve this figure in the last decade. The human resources startup founded in Barcelona by Jordi Romero, Bernat Farrero and Pau Ramon already serves more than 14,000 companies in ten countries. Why is it important. This figure marks a change of mentality in the Spanish ecosystem. For many years, the country startups have focused on the domestic market of 48 million inhabitants (since this year already 49), Treating international expansion as an optional second step. Factorial is different: since 2019 he thought of global, not local. “We are equally ready, handsome, high and strong anywhere in the world,” Its founders have said. A phrase that summarizes the new attitude: Spain as a trampoline, not as a final destination. In perspective. Other recent cases reinforce this trend: Lingokids He has raised 120 million dollars to climb his global educational product. Ultralyticswith headquarters in Madrid and London, it processes 2,000 million daily inferences in 200 countries. Wallboxborn in Barcelona, ​​quotes in New York and sells electric loaders in more than 80 markets, although it has been collapsing for years. Jobandtalent It has expanded strongly in the United Kingdom and the United States After consolidating in Spain. Glovo demonstrated the ability of a Spanish startup to expand to dozens of countries in record time, although it is in the hands of Delivery Hero. There are also less brilliant examples, such as Faver, today turned into a global reference of entertainment, which failed in his first attempt to internationalization and had to forward his strategy before growing again. Cases that show that going out soon is an opportunity, but also a risk. The context. Talking about the “49 million” is to fall short: it is not only the size of the market, but its fragmentation. Climbing a B2B product in Spain is to face regulatory barriers, client dispersion and a purchasing power lower than that of the greats hubs Europeans There is the difference with markets such as Germany, France or the United Kingdom, where many startups can grow at home before going outside. In that sense, Spain is approaching the logic of small countries where unicorns proliferate, such as Lithuania (Vinted, Nordvpn) or Norway (Opera, Remarkable), forced to be born global. Between the lines. The change goes beyond the strategy. It is a mental revolution that leaves behind the “Spanish imposter syndrome”. In 2000 and 2010, success used to be selling a foreign buyer (such as Tuenti to Telefónica either Social Point A Take-Two). Today, the horizon is different: Maintain independence. Quote in the bag. Become buyers instead of acquired. Factorial has rejected purchase offers and plans to reach 1,000 million income without selling. The threat. This new ambition also uncovers the shortcomings of the ecosystem. They are needed … Funds capable of leading rounds of 50 or 100 million. An international talent that still does not flow with the same ease of Madrid or Barcelona as towards London or Berlin. And a investment culture more willing to assume long -term risk. The Spanish venture capital market has grown, but is still limited to the hubs Dominants from Europe. Deepen. Spain, however, has structural advantages to sustain this turn: Spanish connects with 500 million speakers and opens a natural bridge to Latin America. The geographical position places the country as a hinge between Europe and America. Talent costs remain competitive in front of large European centers (Aka “We are cheaper”). In a world where teleworking allows you to attract engineers and managers from anywhere, Spain can become an attractive pole not only to create, but to retain global companies. At stake. The turn to the global is a change of role for Spain in the digital economy. Spain can become an exporter of technology with a global influence capacity, or resign to remain an acquisition nursery for foreign multinationals. The difference will be made by the maturity of its venture capital, the ability to attract and retain international talent and the ambition of its founders. Factorial and its contemporaries have shown that it is possible, now it remains to be seen if they will be exception or precedent. In Xataka | The Spanish unicorn that does not stop growing: the unstoppable promotion of factorial within the technology industry Outstanding image | Factorial

Chinese startups are unseatting the owner and lady of Asian technology: Baidu

by

In the stock market there is a metric that is observed almost with reverential respect. It is about Per (Price to Earnings Ratio), which It allows us to know If an action is expensive or cheap. And when we observe the technological market in China, there is a company with a worrying per: Baidu. Fall of a giant. This Thursday Baidu has published terrible financial results: Fall of quarterly income It is the worst of the last three years and that has caused a 3% drop in the value of the shares in the Hong Kong Stock Exchange. Source: Bloomberg. Per worrying. Not only that: Baidu is now around 9.7, which is the least of all the companies that are in benefits in the Hang Seng Tech index (Hstech) than It serves as a reference When comparing Chinese technology companies. The action, speaking in silver, is too cheap. The search engine loses bellows. For decades Baidu has been considered “the Chinese Google”, but over time its relevance has been blurred. The Chinese young They do not know what is to search in Baidu or Googleand have come to other ways to find answers. Fierce competitors. Platforms such as Tiktok or Instagram have become a spectacular alternative for new generations, and in China we have seen as similar social networks –Xiaohongshu and Douyin– They are getting that. Eric Shen, analyst at the consultant Third Bridge, explained How “these rivals have created content ecosystems that seem more dynamic and attractive, which moves users of the most static and Baidu website.” A disappointing 2025. But there is also AI. And there Baidu has a problem. Baidu’s action has lost a 3% value since the year began. The figure is surprising, especially since its competers focused on AI have a 24%win. Where is Baidu AI? Baidu reacted relatively fast and presented its competitor to Chatgpt in March 2023. It was called Ernie, but from the beginning The performance of that model gave problems. Since then his trajectory has been irregular, and the rise of Deepseek and of Other Chinese models like Qwen (Alibaba) or Doubao (Bytedance) has put it in more and more problems in that market of “IAS Socialist”. An AI that tries to overcome. In February Baidu wanted not to get off the revolution caused by Depseek, so he offered a free version of Ernie last April. Your last model, Ernie 4.5, It’s Open Source – initially they had adopted a proprietary model – but even that does not seem to have encouraged things for now, and there is not too much talk about their capacity: the popularity of other models such as Deepseek or Qwen is now greater. Baidu Lance Ernie 5.0 is expected at the end of August, and it will be then when we can check if this edition does compete in a specially frantic market. Back in view? Analysts believe that Baidu’s latest strategic decisions could help her recover the good path. Among them include investments In Robotaxis And, of course, in new initiatives to boost their AI models. Image | Baidu In Xataka | Deepseek has given the starting gun in the race for a cheaper AI. And China starts with advantage

Something is happening in Silicon Valley. More and more startups are going to the day ‘996’ of China

by

“If you want to leave at 5 you are not at the right work.” Lucy Guo said it, Founder of the Startup Scale AI. And it is not the only one. More and more Silicon Valley startups, especially those dedicated to AI, which are betting on this “extreme commitment” model that reminds us of the endless 996 Conference that were so controversial in China. What is happening? New culture. The culture of ultra-productivity defended by figures like Elon Musk It has been installed in Silicon Valley. In statements a WiredAdrian Kinnersley, CEO of a hiring company, says that “it is becoming very common. We have several clients who have as a prerequisite when selecting candidates who are willing to work on days 996”. Hysteria. In California, labor legislation is very favorable for workers and Kinnersley is surprised that many companies are “breastfeeding it.” It is one of the consequences of the AI career. Not only is there a great competition between China and the United StatesSilicon Valley has become the battlefield where small startups fight for being number one in his. The price: squeeze your workers. What was given was over. Not long ago, working in Silicon Valley was a dream for many. Companies like Google were known for offering gyms, coffee shops and even masseuses. All kinds of comfort for that employees felt at home. That is over. With the return to the offices after the pandemic, The dream began to fade. The tortilla has turned around. Today it is common for any startup to ask its employees to make marathon days and even work on weekends. Extreme commitment. It is another way of saying that you will not go through home much. We recently talked about the young CEO of Greptile and his controversial statements. “We do not offer conciliation”he said in Your X profile. The CEO says it directly in the work interviews: they work from 9 in the morning until late at night and, often, also on Saturdays. It looks a lot like China 996: from nine to nine, six days a week. The excuse: be competitive. In an interview, this CEO said that “nobody cares about the third best company, not even the second best, in any software category. If you are going to strive 95%, it is the equivalent of striving to 0%.” There are arranged people. In Spain we have attended the opposite in the hospitality sector: The waiters rebelled against the 12 -hour days, to the point that there was no labor. In Silicon Valley it seems that there are enough people willing to work 996. In Rilla, another AI startup, they say that 80% of their employees are working 72 hours a week. Even They put it in their job offers And they are not having problems hiring. His CEO says that there is “a very strong and growing subculture, especially in my generation (the Gene z), who grew up listening to stories by Steve Jobs and Bill Gates, entrepreneurs who dedicated their lives to create companies that changed the world. ” 996 in China. The 996 day became popular in China among Chinese technology last decade. This model was the growth engine of companies such as Alibaba or JD.com for years, but there came a point where The workers were fed up and The protests began demanding better working conditions. The Chinese government ended up prohibiting The endless days and went from 996 to 1065 (from 10 in the morning to 6 in the afternoon, five days a week). Temporary and only for those who want. There are startups that defend a slightly more moderate hyper-productive culture. The CEO of Sotiraa startup that applies AI solutions to the logistics sector, sees it as something temporary: “During the first two years of your startup, you have to work in the 996 style.” He also states that these days are for the leaders of the companies and does not believe that the entire employee base must be imposed. This vision creates a kind of structure on two levels, where only a part of the company meets these schedules. It is what the CEO of the “Telesalud” Fella & Delilah company proposed to its employees in a Publication in your LinkedIn. Employees who adhere to this schedule will receive a 25% increase in their salary. Image |Ron Lach, Pexels In Xataka | Work tired, stress and generates burnout. There is a way to reduce all that impact: the four -day week

Almost all startups are dedicating themselves to it

by

If you are a startup in the US and look for financing, it is almost mandatory And combinator. This accelerator has become a unique barometer of the technology industry, and what the candidates present there is a clear indication of where technological innovation moves. And now it moves towards the AI agents… and above all, towards AI in any sense. AI agents everywhere. Of the 144 startups that were part of the 2025 spring batch in and Combinator (YC), 67 were recorded in the accelerator database Within the group of “AI agents”. That means that almost 50% of the candidates work in one way or another in this type of technology, and demonstrates that technological startups in the US see this specific segment of AI as the most promising in the short (and perhaps long) term. Source: pitchbook Climbing to the car. As they point out In pitchbookthe evolution of the candidates that were presented to each of the annual editions of and Combinator (one per station) has been clear, and there are more and more that focus on developing agents of particular AI and in the AI ​​in general to solve a problem. In the last 2024 winter edition of the 260 startups that were presented, more than half were or developing an AI solution or using it for their projects. Diversity Tech, dep Artificial intelligence sweeps everything, and that is also true for those who seek to create new successful companies. Previous trends such as mobile applications, social networks or cloud services pass to the background or are already only components of the central focus, which is AI. Technological diversity, betting on other ways of solving problems, seems to be reduced to the minimum expression: everything can be (should it?) Resolve with artificial intelligence. Now is AI, before it was something else. And Combinator, created in 2005, soon became an oracle of technological investors. If something left there, their options to succeed were older. The trends followed, and while in the first decade of the millennium the startups bet on mobile applications and social networks (Dropbox and Airbnb left there at that time), in the last decade the ones that triumphed were Startups of Fintech, Health and B2B (stripe, doordash). New bubble in view? The expectations generated by AI are even greater than those generated in the past with mobiles and before with the Internet. The bubble of the Puntocom caused a huge correction in the market, but at that time what is happening in this happened: all startups wanted to point to the Internet (and then to mobiles) because they understood that these technologies could solve many problems. The reality is that they ended up transforming our lives, but in many cases these projects failed because they were not solutions to real problems, but solutions to a problem that did not even exist. The question is with this technology something similar will happen. And of course it is inevitable to talk about a potential bubble of AI. Many illustrious names. In addition to those mentioned, Twitch (which was born as Justin.TV), Coinbase, Cruise, Instacart, Reddit, or Helion Energy They took their first steps in and Combinator. One of the last examples is especially notable: Scale AI left and combinator in 2016, and a few hours ago an investment of 13,000 million by Meta has been announced, which in addition has signed its founder, Alexandr Wang. The startup success rate is Very modestbut the “durability” of those that leave YC is much higher than the average: more than 50% of the companies that leave there They are still alive After 10 years, compared to 30% usual in other environments. But that is not necessarily bad. As Andy McLoughlin, of the Uncork Capital investment company, indicated every year the same thing usually happens and “people complain that YC is oversatured with an X technology, but everything now is AI: all the companies in which we invest, which we look for and that it is going through yc is using AI, and the question is if they are good or if they are JOD ****** THE PROBLEM OF THE VALACIONES TRIBUTED. The AI ​​fever has done these startups quickly lift spectacular rounds even without having visible product. YC is a good example of this, and has become an elite event for startups, because there they usually get more financing than in any other event. These emerging companies – pequeñas and unknown – have initial assessments of about 70 million dollars, a spectacular amount for projects that always have a complicated future. Some investors even refuse to “play that game.” Others think otherwise and know that although many can fail, others end up having a spectacular success and it is worth investing “more expensive.” Image | And combinator In Xataka | Soon it will not matter who has the most advanced AI model: the authentic battle will be fought by AI agents

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.