Mobile Tech & Gadgets

Microsoft has just made the greatest investment in its history. And not in Openai, but in an unknown Dutch company

by

Nebius Group, an unknown company based in Amsterdam, has signed a surprising multiannual agreement worth $ 19.4 billion with Microsoft. It is in fact the largest investment ever made by the firm of Redmond, and the question, of course, is why. What is Nebius Group. The company was founded in 1989 as Yandex NV, Yandex’s legal matrix, the well -known search engine that was a rival of Google in Russia. After the invasion of Ukraine by Russia, Nebius Group Yandex sold to a group of Russian investorschanged his name to the current one and focused on a key segment: that of artificial intelligence. Data centers to power. Specifically, in the field of servers and data centers. Since then Nebius Group has been dedicated to providing cloud infrastructure for companies that develop and run AI models. Your rivals They are companies such as Coreweave, Crusoe or Lambda Labs, which are a step below the “hyperscators”, Aws, Azure, Google Cloud or Oracle. The agreement. In it Document registered in the SEC The US indicates that Nebius will yield the computing capacity of the GPUS of its data centers in varisa phases this year and the one that comes, and that the total value of the contract will be 17.4 billion until 2031, with an option for Microsoft to extend those services worth 2,000 million additional dollars. Microsoft cloud reinforcement. The agreement will allow Microsoft to access the computing resources available to Nebius in its Vineland Data Center (New Jersey, USA). It is a movement clearly for solve the shortage of resources that is coming when managing AI workloads: more and more users make use of this type of technology and Microsoft current data centers have a limited capacity. More than OpenAi. The operation is even greater than the one that Redmond’s firm He did in Openai Estimated at $ 13,000. This alliance has allowed Microsoft to have exclusive access to OpenAi’s models and reuse them in the form of its Copilot platform. Meanwhile, Openai has been able to use the Microsoft infrastructure to train and serve those same models to the general public. Nebius rises to the beast in the stock market. The agreement has triggered the value of Nebius’s shares, which had already folded their value in what we had been, but after the news They have grown 60%. The effect is contagious, because one of its srival, Coreweave, has also risen 5% in the stock market without having made any announcement: it has simply become possible candidate for Microsoft or any other large company to invest in its services soon. European taste centers. Although it has roots in Russia, Nebius seems to want to leave that past behind to settle definitively in the European Union. The company current account with five data centers: three operations (Helsinki, New Jersey and Kansas City) and two in development (Keflavik, in Iceland, and Paris). The focus on the installation of data centers in European territory is clear, as these last two projects in full development demonstrate. Another great “European” unicorn. After the creation of its Data Center in Paris – which will presume to have N200 N200 chips – Nebius announced its intention to invest more than 1,000 million dollars in mid -2025 in its AI infrastructure in Europe. Its new data centers in Paris and Iceland demonstrate that vocation, and the company is managing to capitalize on that commitment to AI. It is undoubtedly one of the last protagonists of the European technological scene, which little by little begins to raise alternatives. Freepik did it in Spainthey have done it Mistral and ASML with its unique agreement This week, and now Nebius does. Image | Nebius In Xataka | The ASML-Mistral alliance reveals the European plan B: if we cannot manufacture chips, we will at least control how they are manufactured

Alibaba has presented its largest AI model, with a billion parameters. The question is whether at this point that means something

by

The Chinese giant Alibaba has announced a new language model, the largest they have announced to date. It is called Qwen-3-Max and presumes that it has more than 1 billion parameters. The biggest. It is the last model within the series Qwen3 which was launched in May of this year and, as its name ‘Max’ indicates, it is the largest to date. Its size is given by the parameters, 1 billion to be exact, while the previous models of its series reached a maximum of 235,000 million. According to South China Morning Post (Which owner Alibaba), his model stands out in understanding of language, reasoning and text generation. Benchmarks. The results of the benchmarks place QWen3-Max ahead of competitors such as Claude Opus 4, Deepseek v3.1 and Kimi K2. If Gemini 2.5 Pro or GPT-5 does not appear, it is because they are models of reasoning and have only compared rapid response models. As they point out in Dev.toboth Gemini 2.5 Pro and GPT-5 obtain higher scores in mathematics and code, so reasoning models continue to have advantage in those areas. Qwen3-max-preview can already be tested free of charge. Benchmarks shared by Alibaba. Parameters. The parameters are all the internal variables that a model learns during training. In other words, it is the knowledge that the model has obtained from the data with which it has trained and allows it to interpret our requests and generate their answers. In theory, the more parameters, the model will have more and better capabilities. It also implies that it needs more computational power both to train and to execute the model. More does not mean better. The speech of the parameters remembers that of the megapixels with the first cameras. A 100 megapixel sensor will take larger photos than a 10 sensor, but there are other crucial factors that affect image quality such as sensor size or lens luminosity. Quality data. More parameters can be translated into more learning capacity and more resolution of complex tasks, as long as quality training data has been used. It is obvious: a language model that has been trained with redundant, incorrect or biased data will learn and continue to reproduce those errors in their operation. There are more. In 2022, the laboratory Deepmind from Google, discovered that many models were oversized in parameters but underlined in data. To demonstrate it they created the Chinchilla model with “only” 70,000 million parameters, but four times more data. The result was that it beat Gopher, a model with four times more parameters. Architecture. The architecture of the model is another decisive factor in order to achieve an efficient model; A standard architecture is not the same that forces the model to use its entire neuronal network, than one like Mixture of experts which consists of many smaller networks. It would be something like having an expert committee each with a specialty. In this way, the model can choose your expert for each query and not have to use the entire network. For example, with this technique, Mistral manages to use only a fraction of his parameters And so it is faster and cheap to execute. Image | Markus Winkler, via Pexels In Xataka | The ASML-Mistral alliance reveals the European plan B: if we cannot manufacture chips, we will at least control how they are manufactured

that Russia is getting closer

by

1,300 kilometers from the north pole is the town more northern inhabited of the world. We refer to the Norwegian people of Svalbard, an extreme region, a gigantic ice cream, much above the Arctic Circle where 2,500 people officially have Forbidden to die. And yet, what worries them most has nothing to do with their climate or geographical place. What worries them most is that Russia is getting closer. Forbidden to die. It We have counted. In Svalbard one of the most peculiar rules on the planet governs: it is Forbidden to die In Longyearbyen, its capital, because the permafrost prevents the decomposition of the corpses and preserves viruses and bacteria for decades, which was already checked when bodies buried in 1918 still contained traces of the Spanish flu virus. Since 1950the cemeteries stopped accepting deceased and any seriously ill or elderly person must move to the continent to spend their last days there, while the few still visible burials are vestiges of previous times. The rule reflects both the hardness of living in the Arctic Like fragility of a settlement where life, and even death, is conditioned by ice eternal. The new Arctic border. I told the weekend Bloomberg. On the top of the island of Spitsbergen, in the Svalbard archipelago, Raise Svalsatthe largest earthly satellite station in the world, with 170 domes that connect every few seconds with satellites in polar orbit and provide critical data for weather, navigation, climate research and maritime security. Its location, 78 degrees north and with access to airports and ice -free ports, it gives this Norwegian enclave a unique strategic relevance in an Arctic every time More disputed. Norway sees it as a Sovereignty symbol And it has expanded with energy and communications projects, collaborating with agencies such as NASA, ESA or Oneweb. However, the dependence of this technological node and its potential dual use reinforce the Russia tensionwhich accuses Oslo of rape Svalbard’s treaty When there are submarine cables and demonstrate too much emphasis on territorial control. Svalbard Satellite Station Russia and symbolic persistence. Just a few kilometers from Longyearbyenthe Norwegian capital of the archipelago, resists BARENTSBURGa Russian mining settlement that, with just 400 inhabitants, on Soviet flags and maintains a bust of Lenin as a reminder of its history. Managed by the state company Trust Arktikugolthe people have renewed facilities, open a gym, promoted tourism and signed agreements with Russian universities, although their mine lacks real economic viability. In practice, it represents A strategic symbol For Moscow in the Arctic. After the invasion of Ukraine, the relations between the Norwegian and Russian communities cool: They canceled visits Tourist to Barentsburg, joint sports activities were suspended and many Ukrainian miners returned home. What previously survived the Cold War today has broken under the pressure of war and sanctions. BARENTSBURG Svalbard and uncertainty. The Norwegian government insists in that “Svalbard is as Norwegian as Oslo”, reinforcing his control through visits from authorities, indirect military presence and international projection. However, everyday reality is that of a community of 2,500 people of more than 50 nationalities, Tourism dependent And of expensive imports in a hostile environment where permafrost melts, basic services are limited and the last Norwegian coal mine is about to close. The transition from an industrial past to a post -mimine economy leaves open questions: What will replace mining, how much tourism can be allowed and how to sustain a diverse population without causing internal tensions. The Life in Svalbardmarked by polar bears, mortal avalanches and the lack of maternity or geriatric centers, remember that there everyone lives in time borrowed. “ A global rivalry. The War in UkraineTrump’s gestures claiming Greenland and the Russian militarization of the Kola Peninsula have turned the north into an epicenter of Geopolitical rivalry. With the accelerated thaw opening routes such as the passage of the northeast and multiplying the strategic value of resources and communications, Svalbard emerges as a microcosm of the worldwide struggle: A place where science, defense and trade are intertwined with political symbols and sovereignty shocks. For Longyearbyen residents, the transition has been palpable: from a mining people cohesive to a community exposed to global currents that generate diffuse identity and tensions with their Russian neighbors. As expressed A veteran resident, even in the coldest times of the cold war, local cooperation persisted. Now, that coexistence also seems lost, leaving Svalbard in the awkward first line of a new era of Arctic Competition. Between tourism, science and climate. Plus: the mines It raises an economic vacuum that only partly They fill tourismincreasingly massive with the arrival of cruises, and scientific projects such as World Seed Bankwhich barely offer employment. Meanwhile, Arctic heating, four times faster That the global average melts permafrost and compromises infrastructure, makes energy more expensive and multiplies natural risks. The dilemma of how much tourism growth accept, how to attract permanent families and what role in global research defines the next Svalbard stage. And in the midst of these tensions, the inhabitants try to maintain their identity and resilience, aware that they live in a territory where nature and Geopolitics They weigh as much as the will of their own residents. Image | Sprok, Bernt Rostad In Xataka | Eight Rompehielos have turned Russia into the power of the Arctic. Your secret: Nuclear force to operate all year In Xataka | We already know why Greenland is so important: the key is called Giuk and gives access to China and Russia east of the US

Carrefour is liquidating this 65 -inch neoqled samsung

by

Renewing our TV can be a difficult decision to make due to the large number of models that exist in the market. Carrefour is one of those stores where we can find numerous offers, such as this one that we present to you. Now you can take this Smart TV SAMSUNG TQ65QN74FAT by 899 euros. In addition, you receive a 15% coupon for upcoming purchases and exclusive gift content. TV Samsung TQ65qn74fat 65 “ The price could vary. We obtain commission for these links A perfect TV to be able to ride your own home cinema This TV of the Korean firm mounts a 65 -inch neoqled panel and offers 4K resolution. It is compatible with HDR10+has a filmmaker mode and presents a 144 Hz soda rateso it is also ideal for gaming. In the audio, it comes with Two 2.0 channel speakers which offer a 20 W RMS power and offer surround sound. Its design is Airsimlim type, so it has hardly any thickness and its base is Aer Center. Works under it Tizen operating system and brings the standard bixby assistant, although it is also Compatible with Alexa and Google Assistantin addition to Apple AirPlay. It also highlights its Multi-View mode and in the connectivity section, it has four HDMI, two USB-A ports, Audio Digital Health, Wifi 5, Ethernet, Bluetooth 53 and HDMI EARC. Some accessories that may interest you for this Samsung Smart TV Amazon Fire TV Stick 4K (last generation) The price could vary. We obtain commission for these links Samsung Sound bar HW -C430/ZF – Wireless subwoofer included The price could vary. We obtain commission for these links Some of the links of this article are affiliated and can report a benefit to Xataka. In case of non -availability, offers may vary. Images | Webedia and Samsung In Xataka | Mega-guide to ride a home cinema: projector, screen, sound system and more In Xataka | Best home film projectors. Which to buy and five recommended models from 299 to 18,000 euros

The Saudi strategy is not only economical, it is also political

by

Fifty years after the Arabic embargo of 1973, which fired prices and changed the world economy forever, OPEC+ moves again. This time, it is still in the opposite direction. Instead of restricting the offer to make crude oil, the group led by Saudi Arabia and Russia has decided to continue with the open tap. The announcement that stirred the market. The decision came after a video call between eight key countries of the OPEC+. According to the press releasethe group will add 137,000 barrels per day in October, first step to dismantle 1.66 million b/d that were still frozen until 2026. The context matters: in April, the block had already surprised returning 2.2 million B/D a year earlier than expected. This acceleration resulted in a 12% drop in crude oil prices so far from 2025, According to Bloomberg. The market immediately reacted. As Reuters has collectedBrent rebounded 1.95% up to $ 66.78 and WTI 1.94% up to 63.07. Analysts such as Ole Hansen (Saxo Bank) have interpreted the reaction as a classic “sells with the rumor, buys with the fact”: the increase was lower than anticipated, which relieved fears of immediate oversupply. But why does the open tap continues? As Bloomberg has pointed outthe poster has abandoned the role of “price defender” and has pivated towards the recovery of market share. In addition, Saudi Arabia demands compensation to countries that have overcover, such as Kazakhstan, Iraq or United Arab Emirates. The gradual increase allows us to emphasize quotas and reveal who really has the capacity to pump more and who does not. A deeper goal. The movement also has a political reading. According to BloombergMohammed Bin Salman will visit Washington in November, and the increases send a sign of goodwill to President Donald Trump, which has been demanding lower prices for months as a measure to contain inflation. Reuters He recalled That Trump has even hinted at a second phase of sanctions to Russia, which reinforces the logic of lowering oil. The play also exposes an American contradiction. As we have detailed in Xatakalower prices relieve inflation and give political air to Trump, but at the same time suffocate fracking, cornerstone of the energy independence that he claims. Many shale companies need quotes of $ 60–65 to be profitable. If the Brent falls below that threshold, Trump’s “victory” over cheap gasoline could become a blow to one of the strategic sectors of his own country. RUsia quiet in the OPEC+. In theory, Moscow should oppose: you need high prices to finance its war in Ukraine. In practice, accept the Saudi plan. In fact, Russia He already asked for a pause in Julysupported by Algeria and Oman, but was ignored. The Russian lifeguard is in Asia. As the BBC explainedPutin met in Tianjin with Xi Jinping and Narendra Modi at the Shanghai cooperation organization. There he reinforced his links with the two largest buyers of his crude. On the one hand, China imported more than 100 million tons of Russian oil in 2024, almost 20% of its energy purchases. On the other, India multiplied its purchases to 140,000 million dollars from 2022, after western sanctions. These clients, attracted by Russian discounts, are lifeguard that allows Moscow to tolerate lower prices in the OPEC+. For Modi, in addition, challenging Washington’s pressures reports internal political benefits, According to the BBC. Why don’t prices sink? Despite the increase in supply, crude prices have remained surprisingly stable. Several factors help explain this resilience. First, the increases have been more nominal than real: effective production is below what announced, According to The New York Times. To this is added the threat of new sanctions against Russia, which maintains a risk premium in the market. In addition, barriles return the “safety network” of idle capacity, which paradoxically limits the bearish pressure by leaving less margin against disruptions, Bloomberg warns. Finally, the Saudi “boldness” has reinforced the confidence of the operators: after the initial drop of 2025, the Brent stabilized around 66–67 dollars. The Saudi paradox. In parallel, Saudi Arabia is reducing its internal oil consumption to release barrels to the international market. As we have pointed out in Xatakathe kingdom displays solar and storage projects that replace crude oil in electricity generation. The logic is simple: each solar megavatio is equivalent to an extra barrel to export. This strategy strengthens its role in OPEC+, but generates fiscal tensions that have already forced megaprojects such as Neom, which is an early sign that the offensive to gain fee can hit the heart of the Saudi reformist agenda. Horizon 2026. Beyond the present, the look is set in the future. According to ReutersGoldman Sachs projects a slight surplus in 2026 for supply improvements in the Americas and the weakening of Russian crude. Its forecast places the Brent in 56 dollars and the WTI on 52 on average that year. The Saudi strategy seeks precisely to reach that scenario with greater share and margin to cut if necessary. According to Financial Timesthe real impact of October increase could be much more modest than it seems. Although the OPEC+ announced 137,000 additional daily barrels, analysts estimate that the effective figure will be around 60,000 b/d, since most countries involved pump almost at full capacity. Only Saudi Arabia and, to a lesser extent, United Arab Emirates have real maneuvering margin. The decision, however, has an internal function: Riad takes advantage of the return of barrels to measure the production capacity of each member with a view to renegotiating quotas in the future. The true test will arrive in the fourth quarter of 2025, when the market must absorb a greater flow of crude oil in full season of weaker demand. Chronicle of an announced break. In 1973, the OPEC paralyzed the West with an embargo that triggered prices. Half a century later, the same poster changes course: it floods the market to gain share, satisfies Donald Trump, discipline to the shale and seeks to reaffirm internal leadership. The movement is not … Read more

There is a renewable energy source that works 24 hours a day, and Japan has opted for it: osmotics

by

Omnipresent wind and solar energies have the same problem: intermittency. The wind does not always blow and the sun does not always shine. This dependence on weather conditions It forces us to look for storage solutions OA depend on a 24/7 available source to have a constant supply. But what if that other source was also renewable energy? Japan bets on osmotics. In August, the Japanese city of Fukuoka launched the First osmotic energy center in the country. It is just the second installation of this type in the world and represents more than a pilot test for a technology that promises to revolutionize the energy mix. “I hope it extends not only in Japan, but throughout the world,” said Akihiko Tanioka, an expert in the field of the Tokyo Science Institute. The plant will generate about 880,000 kilowatts-Hora a year to feed part of the desalination plant that supplies fresh water to the city. Although it may seem like a modest figure, equivalent to the consumption of about 220 Japanese homes, its true value is that it works continuously. It is not affected by the weather or time of day, and does not emit carbon dioxide. The power to mix fresh and salted water. Osmotics is a new generation renewable energy based on A natural phenomenon that we all study at the institute: osmosis. When two solutions with different salt concentration are separated by a semipermeable membrane (which lets the water pass, but not the salt), the water of the less concentrated solution flows naturally towards the most concentrated to try to balance the mixture. Fukuoka’s plant places fresh water (treated wastewater) on one side of the membrane, and salt water (sea water) to the other. Fresh water crosses the membrane to the salt water side, increasing volume and pressure on that side. The conression is used to move a turbine that, connected to a generator, produces electricity. Therefore, this renewable source is also known as saline gradient energy or “blue energy”. From the initial promises to the first problems. This technology is not exactly new. Already in 2017, in Xataka we echoed its enormous potential: The studies suggested that it could generate up to 40% of the world energy demand if it would be used in the mouths of all rivers. The challenge was the efficiency of the membranes and the cost of the facilities. The first systems, such as delayed pressure osmosis, suffered bioincructing problems: bacteria blocked membranes, drastically reducing their effectiveness. Other systems, such as “inverse electrodialysis”, were more durable but generated very little energy. The Fukuoka plant, like the first in the world inaugurated in Denmark in 2023 by the Saltpower company, demonstrates that progress in membrane technology is allowing to overcome these obstacles. Nanotechnological membranes. In France, the company Sweetch Energy has developed Much more efficient nanometric size membranes, capable of generating between 20 and 30 watts per square meter. In comparison, the most advanced systems to date generate 12.6 watts. The company seeks to install its first real -scale generator, Osmorhône 1, at the mouth of the Rhone River. The potential of this unique river would reach 500 MW, equivalent to mid -nuclear reactor capable of supplying two million people. And it would be just the beginning. All deltas and estuaries in the world release 30,000 energy from each year, a figure similar to world demand of electricity. Salina bachata in Fukuoka. In addition to the friction of the membranes, the osmotics also loses energy in water pumping. But, as he explains The Guardianthe new Fukuoka plant uses the concentrated brine that remains of the desalination process to increase the salinity difference and, thus, the energy potential available. The inauguration of the Japanese plant and the advances of companies such as Sweetch Energy mark a turning point for the osmotics. They are the first steps to stop being a laboratory promise and become an industrial reality. In a clean, permanent energy, which does not depend on the climate and that can be integrated into existing infrastructures such as ports, desalination makers or locks. Japan has made its bet: the mixture of fresh and salted water will be part of its energy mix. Image | Umi-No-Nakamich Desalination Plant (Obayashi) In Xataka | Japan has just made a monumental bet for Perovskita’s solar panels: they are his best chance against China

If a film created by AI will convince us to go to the cinema

by

The AI ​​begins to seize Hollywood. Although there prefer not to say it in voice too high. And who is willing to achieve this is Openai, whose generative model will be used to create an animated film that will be called ‘Critterz’ and will be released in 2026. ‘Critterz’. In OpenAi They have yielded Both his tools and his computer resources to achieve something that had to arrive sooner rather than later: an animation feature created almost entirely with AI. ‘Critterz’ is a film that deals with how some creatures of the forest end up living an adventure when a strange disturbs its village. Premiere in Cannes in 2026. The film is the work of Chad Nelson, creative of OpenAi, who began to outline the characters three years ago while trying to create an animation short with Dall-E. The short, by the way, is available on the Openai channel on YouTube. This expert will work with producers in London (Vertigo Films) and Los Angeles (Native Foreign, specializing in combining AI with traditional production tools) in a film that hopes to be released at the Cannes Festival in May 2026. With everything it is faster … James Richardson, co -founder of Vertigo Films, stood out In The Wall Street Journal As the team is trying to complete the film in nine months, when normally a production of this type usually takes three years. … and cheap. In addition, the ‘Critterz’ budget is 30 million dollars, much less than animation films usually cost. Production teams, yes, plan to hire human bending for the voices of the characters, and also want to hire artists to draw the sketches that then use with OpenAi tools such as GPT-5 and their models of generative image generative. How long and money do Pixar cost? Pixar employees themselves They explained How one of his films takes about four years to complete. The first two focus on specifying both the story and the characters, and the other two to cheer and render the film. The budget of the setback ‘ was Of 175 million dollars, and subsequent productions have also been around 200 million dollars. A delicate project. Although companies such as Disney or Netflix have been experiencing with the use of AI in different phases of production, the film industry walks on swampy terrain. This technology and their threat to the workers caused the strike that He had the entire segment in check and in which they were victorious. How to avoid problems. The script has been written by the scriptwriters of ‘Paddington in Peru‘, and the selection of the voices of the characters is already being carried out. The project is funded by the Matrix of Vertigo, Federation Studios, which has already devised a way to avoid possible problems with industry workers. Thus, a compensation model has been developed that allows the nearly 30 people who are working in ‘Critterz’ to have a participation in the benefits. A risky bet. Although for years the possibility that anyone can make a film – animated or not – with AI tools, this production aims to be the clear demonstration of that option has been profiled. To succeed, however, you will need to convince the audience of two things. The first, which is good enough to pay for her in theaters. The second, more importantly, that human beings accept that machines make films. In Xataka | The greatest fear was that AI took our work. The reality is that they are replacing those who are learning to work

China is no longer made up of moving away from Nvidia. His next step is the heart of the AI ​​with a system that breaks molds

by

In 2017, the Paper “Attention is all you need”Google changed the technical basis of language generation: the Transformers They allowed to process long sequences in parallel and climb models to sizes that were previously unfeasible. That climbing route has driven architectures such as GPT and Bert and has converted self -how The central piece of generative AI Contemporary. But this new approach was accompanied by growing costs in memory and energy when the context lengthens, a limitation that has motivated research to develop alternatives. Spikingbrain-1.0 aims to break molds. Of the “attention is all you need” to the brain: the new commitment to break limits in the A team from the Chinese Academy of Sciences Automation Institute He has just presented Spikingbrain-1.0. We are talking about a family of spiky models aimed at reducing data and computation necessary for tasks with very long contexts. The experts propose two approaches: Spikingbrain-7B, of linear architecture focused on efficiency, and spikingbrain-76b, which combines linear attention with Mixture of Experts (MOE) mechanisms of greater capacity. The authors detail that much of the development and the tests were carried out in clusters of GPU Metax C550, with libraries and operators specifically designed for that platform. This makes the project not only a promising advance at the software level, but also a demonstration of own hardware capabilities. An especially relevant aspect if China’s effort is taken into account for reducing his dependence on Nvidia, A strategy that we already saw reflected with Depseek 3.1. Spikingbrain-1.0 is directly inspired by how our brain works. Instead of having neurons that are always “burning” by calculating numbers, uses spiky neurons: units that accumulate signals until they exceed a threshold and trigger a peak (spike). Between peak and peak they do nothing, which saves operations and, in theory, energy. The key is that not only does it matter how many peaks there are, but when they occur: the exact moment and the order of these peaks carry information, as in the brain. In order for this design to work with the current ecosystem, the team developed methods that convert traditional self -acting blocks into linear versions, easier to integrate into its spiky system, and created a kind of “virtual time” that simulates temporal processes without stopping the yield in GPU. In addition, the Spikingbrain-76B version includes Mixture of Experts (MOE), a system that “awakens” only to certain submodos when we are needed, which we have also seen in GPT-4O and GPT-5. The authors suggest applications where the context length is decisive: analysis of large legal files, complete medical records, DNA sequencing and massive experimental data sets in high energy physics, among others. That lace appears reasoned in the document: if the architecture maintains efficiency in contexts of millions of tokens, would reduce costs and open possibilities in domains today limited by access to very expensive computer infrastructure. But validation in real environments is pending outside the laboratory. The team The code of 7,000 million parameters has released in Github next to a detailed technical report. It also offers a web interface similar to chatgpt to interact with the modelwhich according to the authors are deployed entirely in national hardware. Access, however, is limited to Chinesewhich complicates its use outside that ecosystem. The proposal is ambitious, but its true scope will depend on the community to reproduce the results and make comparisons in homogeneous environments that evaluate precision, latencies and energy consumption in real conditions. Images | Xataka with Gemini 2.5 | ABODI VESAKARAN In Xataka | Openai believes having discovered why the IAS hallucinates: they don’t know how to say “I don’t know”

China knows that its population is going to sink but already has a long -term plan to solve it. How not, thanks to the AI

by

In 2024 China exceeded 1,400 million inhabitants, but according to United Nations Data That figure will begin to fall remarkably in the coming decades. Predctions talk about 2050 The population There will already be between 100 and 200 million people, and for 2100 in the best Chinese cases it will have only 900 million inhabitants. Among many other things, there will be a huge impact In a key sector: that of the number of workers available. Source: United Nations In Economist They go further and stand out as today 22% of Chinese citizens are over 60 years old, but in 2035 that fee will rise to 32.5%. The Birth rates They are also very low: to maintain the population that rate should be 2.1 children per woman, but in 2024 the rate It was 1.0 children. And as they point out in Chinese triviumthe situation threatens to make China fall into “the average income trap”, slowing the growth of its economy because, simply, it cannot already have so much cheap labor and the level of exports that it had so far. More productivity. The Xi Jinping solution lies in a concept that the Chinese government calls “new quality productive forces” (NQPF). It is a strategy to boost productivity, and for this you can take advantage of technological innovations, improve education or disseminate ideas and improvements throughout all industries. In April XI He already indicated that its 15th “five -year plan” (from 2026 to 2030) will precisely drive this type of measures. And above all, one. Ai plus. That is where the so -called AI Plus directive, a long -term plan that in the next 10 years hopes to get AI becomes fully infiltrated the entire business and consumption fabric in China. The phases They are clear: 2027: reach a penetration rate of more than 70 % of terminals and intelligent agents in six key sectors (science and technology, industry, consumption, social welfare, global governance and cooperation). 2030: That penetration rate must be 90%. 2025: According to the report, “the AI ​​will be adopted as universally as electricity and the Internet, becoming the ‘basic infrastructure’ of society.” In all areas. The plan pretends that AI is the great technological innovation that allows to boost productivity in all types of areas. For example in innovation and scientific discovery, but also in industrial transformation. In the latter case both agriculture and services are short -term objectives for the deployment of AI. And also for consumption. In the consumer sector, AI must stimulate demand with the application of this technology in products such as connected cars or intelligent domestic devices. And of course, the objective is also that AI accompany and help all citizens in education, work or to improve health. Governance and cooperation. The last two major sectors of the initiative are those that affect government efficiency – for example, for urban planning, national segurity or online censorship – and international competitiveness. In the latter case, the plan defends that AI must be considered an international public good and must follow an open source philosophy. Challenges. Some experts believe that the success of the AI ​​Plus program is not assured, and there are also no clear metrics that serve to really measure the contribution of AI to economic growth. Jeffrey Ding, from the George Washington University, explains that “China faces a diffusion deficit because its ability to innovate and be a pioneer in new technologies far exceeds its ability to spread those advances throughout the economy.” Low adoption. Although the country can train its best talents of AI in its best universities, it has certain problems to achieve that same training between “standard” ingenerios, which can create a barrier for the adoption of these technologies. A survey conducted in 2022 confirmed That perception: only 4.5% of the companies owned by the Chinese government confirmed that the efforts to transform them digitally were “fulfilling expectations.” And the AGI, what? Interestingly, in that document there is no mention of the General Artificial Intelligencewhat seems to suggest once again that China prefers to make generative AI infiltrate throughout society instead of going in search of supremacy in AGI. That does not mean that China does not end up looking for this type of achievement, but at the moment its focus is another. A much more pragmatic and that is precisely aimed at mitigating the effects of its future demographic contraction. Image | Global panorama | Brian Matangelo In Xataka | China has found the formula to upload salaries while still being the world’s factory: a silent robots invasion

WHO wants to improve access to drugs such as Ozempic. Now they are on their list of essential medicines

by

Ozempicthe treatment against diabetes that reached the fame converted into medication for weight loss has lived for a convulsive years that could have resulted in a “death for success.” The popularity of the drug created a severe supply problem. Now, the World Health Organization (WHO) has taken a new step to improve access to the drug, especially in less favored countries. Updating the list. The (WHO) has up to date its model list of essential medicines (EML). The new list Includes GLP-1 drugswhich emulate the peptide that gives its name to the group, the family of treatments flagging by Ozempic. According to the international organism, this inclusion aims to improve access to the popular drug, especially in countries with less income. The extension covers other drugs, such as treatments against various types of cancer, against cystic fibrosis, psoriasis, hemophilia and other blood -related disorders. In total, the new list now incorporates 523 essential medicines. The extension It also affects the WHO model list of essential pediatric medicines (EMLC), which now has 374 treatments. The “GLP-1 family”. The incorporation of the GLP-1 drugs includes compounds such as semaglutida (the Ozempic base and Wegovy), the dorara and the liraglutid Zepbound and Mounjaro). These types of compounds act as agonists of the hormone receptors that we know as LPG-1, or Peptide similar to type-1 glucagon (although some also work as agonists of other hormones). This hormone is produced and segregated by our body after meals and meets Two key functions. On the one hand, it informs the pancreas of the intake to reinforce insulin production, hence these compounds help people with diabetes. The second function of this peptide is the one that makes these useful drugs to lose weight. The LPG-1 also transmits information to our brain, information that it interprets as the feeling of satiety. Improve access. According to WHO explains in a press releasethese drugs can help people with type 2 diabetes improve their glycemic control, but they are also able to “reduce the risk of cardiovascular and renal complications, favor weight loss and even reduce the risk of premature death, especially if they have renal or cardiac failure ”, a fact that would be based, at least in part, the decision. As details in the international organism, the price of these drugs implies significant access to them. Its inclusion in the list aims to expand this access, prioritizing patients who can benefit the most while fostering competition through genericians capable of bringing these medications to primary care, “especially in unattended areas.” “A large part of the direct spending of families in noncommunicable diseases is destined for medicines (…). To offer equitable access to essential medicines, an equal response of health systems, a strong political will, multisectoral cooperation and programs focused on people who allow everyone,” detailed in the press release Deusdedit Mumbangizi, Director of Policies and Standards for Medicines and Health Products of the WHO. Expanding in the fight against cancer. The expansion of WHO lists also include several cancer treatments, including several antineoplastic, drugs that seek to avoid the formation of growth that can become cancerous or neoplasms. Among these drugs are inhibitors of the immune control point PD-1/PD-L1, which help our immune system To recognize and attack with cancer cells. In Xataka | China is the great candy of pharmaceutical thinning. And there is a wild race for selling the new ozempic Image | Chemist4u / Yann Forget

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.