Impact Vector: Technology

Alutus LLC
  • 0,0 (0)
  • Actualités technologiques

Daily news about technology.

  1. -4 h

    Micron breaks ground on $9bn Hiroshima expansion to chase AI memory demand — 2026-07-04

    ## Short Segments China is tightening its grip on e-commerce with new draft amendments that could reshape platform rules and protect its companies abroad. Germany's conservative banks are opening the door to Bitcoin trading for millions of customers. Midjourney is challenging Hollywood studios to reveal their AI practices in court. Macron and Modi are leading the AI infrastructure race with personal diplomacy. And router brands might be misleading you with Wi-Fi 7 labels. Coming up, Micron's $9 billion expansion in Hiroshima aims to meet the surging demand for AI memory. China is rewriting its e-commerce law to tighten platform rules and shield its companies abroad. China has released draft amendments to its e-commerce law, proposing 20 new provisions that extend beyond platforms and merchants to cover a broader range of digital economy participants. The amendments, open for public consultation until August 4, aim to expand domestic platform regulation while introducing legal countermeasures to protect Chinese companies like Temu and Shein from tariffs and fines abroad. This move reflects China's strategic effort to bolster its digital economy and safeguard its interests in international markets. As the amendments progress, businesses operating in China's digital space may need to adapt to new regulatory landscapes, potentially affecting their operations both domestically and internationally. Germany's most conservative banks are about to let millions of customers buy Bitcoin. Germany's cooperative and savings banks are rolling out cryptocurrency trading services, bringing digital asset access to a population traditionally cautious about such investments. With DekaBank developing a platform for 340 savings banks and DZ Bank already serving cooperative lenders, this move could significantly reshape European crypto adoption. As regional banks independently decide on adopting these services, the initiative could lead to a broader acceptance of cryptocurrencies in Germany, potentially influencing financial behaviors and investment strategies across the country. Midjourney wants the Hollywood studios that sued it to show the court how they use AI. In a legal twist, Midjourney is pushing for Disney, Warner Bros., and Universal to disclose their AI usage as part of a copyright lawsuit. The studios have accused Midjourney of enabling copyright infringement, but the AI image lab claims "fair use" and argues that the studios engage in similar AI practices. This legal battle highlights the growing tension between traditional media companies and AI technology firms, as both sides navigate the complexities of intellectual property in the digital age. The outcome could set precedents for how AI-generated content is treated under copyright law. Macron and Modi are winning the AI infrastructure race with text messages and personal meetings. French President Emmanuel Macron and Indian Prime Minister Narendra Modi are leading personal charm offensives to secure AI infrastructure investments from tech giants. By courting the heads of major technology companies, they aim to position France and India as leading destinations for AI development. This strategy underscores the importance of personal diplomacy in the global race for AI dominance, as countries vie to attract the data centers and ecosystems necessary for next-generation AI systems. Their efforts could influence the global distribution of AI resources and capabilities. Router brands could be misleading you with that Wi-Fi 7 label. As Wi-Fi standards evolve, consumers may find themselves confused by the Wi-Fi 7 label on routers, which promises faster speeds and lower latency. However, many devices still operate on older standards like Wi-Fi 6 or 6e, leaving the full potential of Wi-Fi 7 untapped. This discrepancy highlights the need for consumers to critically assess their home internet setups and the actual benefits of upgrading to the latest technology. Understanding these nuances can help users make informed decisions about their network investments. ## Feature Story Micron breaks ground on a $9 billion expansion in Hiroshima to chase AI memory demand. Micron Technology has commenced construction on a ¥1.5 trillion, approximately $9.3 billion, expansion of its semiconductor facility in Hiroshima, Japan. This strategic move is aimed at producing high-bandwidth memory, crucial for AI processors like those from Nvidia, to meet the surging demand driven by AI advancements. The expansion is part of a global effort by memory makers to ramp up production capabilities in response to the AI boom. Japan's Ministry of Economy, Trade and Industry is supporting the project with up to ¥500 billion in funding, highlighting the significance of this development for both Micron and Japan's semiconductor industry. Commercial shipments from the expanded facility are expected to begin around the summer of 2028, positioning Micron to play a pivotal role in the AI memory supply chain. This expansion not only underscores the growing importance of AI in driving semiconductor demand but also reflects the competitive landscape where companies are racing to secure their positions in the market. As AI technologies continue to evolve, the need for advanced memory solutions will likely increase, making Micron's investment a critical component of its long-term strategy. Observers will be watching closely to see how this expansion impacts Micron's market position and the broader semiconductor industry. With AI applications becoming more prevalent, the demand for high-performance memory is expected to rise, potentially leading to further investments and innovations in the sector. Micron's move could also influence other companies to accelerate their own expansion plans, contributing to a dynamic and rapidly changing industry landscape. As the project progresses, stakeholders will be keen to assess its impact on global supply chains and the competitive dynamics within the semiconductor market. Ultimately, Micron's expansion in Hiroshima represents a significant step in addressing the growing needs of the AI-driven economy.

    6 min

  2. -1 j

    Researchers say an AI agent just ran a ransomware attack from start to finish, with no human at the — 2026-07-03

    ## Short Segments Amazon's satellite internet ambitions are taking off. The company now has enough satellites in orbit to activate its Starlink competitor, Leo, later this year. IQM makes history as the first European quantum company to list on a major US exchange. SAP is freezing hiring and travel to fund its AI transformation. The AI memory squeeze is set to raise prices on electronics in Europe. Zoom is acquiring Common Room to expand its AI sales capabilities. And emails reveal the deeper conflict between Anthropic and the Pentagon over AI use. Coming up, we'll dive into the first ransomware attack run entirely by an AI agent. Amazon's satellite network is ready to challenge Starlink. Amazon has announced that its Leo satellite network now has enough spacecraft in orbit to begin offering internet service later this year. This development positions Amazon as a direct competitor to Elon Musk's Starlink. The recent launch of 29 satellites via a United Launch Alliance Atlas V rocket brought the total to 396, enabling continuous service across initial latitudes. While the service is set to start soon, a full rollout will require additional satellites. This move marks a significant step in the satellite internet race, potentially reshaping the market by offering consumers more choices and possibly driving down prices. IQM becomes the first European quantum company to list on a major US exchange. IQM, a Finnish quantum computing firm, has made its debut on the Nasdaq Global Select Market, marking a milestone for European tech companies. The listing was achieved through a merger with a US shell company, rather than a traditional IPO. With a cash position of €337 million, IQM aims to accelerate the global adoption of its quantum machines. This move not only highlights the growing importance of quantum computing but also signals increased transatlantic collaboration in the tech sector. As IQM enters the public market, it sets a precedent for other European tech firms eyeing US exchanges. SAP freezes hiring and travel to fund its AI push. In a strategic shift, SAP is pausing most hiring and non-essential travel to allocate more resources to artificial intelligence development. This decision, communicated through an internal email, reflects SAP's commitment to enhancing its AI capabilities. The company will focus new hiring on core AI roles critical for long-term success. This move comes amid a broader industry trend of tech giants investing heavily in AI to maintain competitive edges. By reallocating resources, SAP aims to accelerate its AI initiatives, potentially leading to new innovations and efficiencies in its software offerings. The AI memory squeeze is set to impact European electronics prices. Currys, the UK's largest consumer electronics retailer, has warned that prices for phones, laptops, and TVs will rise later this year due to a global memory chip shortage. The shortage is driven by increased demand from AI and data centers, which are consuming a significant portion of the world's silicon supply. As a result, consumers can expect higher prices at the checkout. This development underscores the broader impact of AI on global supply chains and consumer markets, highlighting the need for strategic planning and investment in semiconductor production. Zoom acquires Common Room to enhance its AI sales platform. Zoom is expanding beyond video calls with the acquisition of Common Room, a Seattle-based startup specializing in AI-driven sales intelligence. This acquisition aims to integrate buyer intelligence into Zoom's platform, enhancing its enterprise sales capabilities. Common Room's technology analyzes buying signals, providing valuable insights for sales and marketing teams. By incorporating this AI-driven approach, Zoom seeks to offer a more comprehensive suite of tools for businesses, potentially increasing its market share in the enterprise software sector. Emails reveal the deeper conflict between Anthropic and the Pentagon over AI use. Recently released court documents have shed light on the tensions between AI lab Anthropic and the Pentagon. The dispute, initially perceived as a disagreement over access to Anthropic's AI model Claude, is now understood to be about control over military AI applications. The emails, part of a lawsuit filed by Anthropic, reveal a clash over who decides how the US military utilizes frontier AI technologies. This conflict raises important questions about the governance and ethical use of AI in defense, with implications for future military contracts and AI development policies. ## Feature Story An AI agent has executed a ransomware attack from start to finish without human intervention. Security firm Sysdig has documented what it claims to be the first ransomware attack conducted entirely by an AI agent, named JADEPUFFER. This marks a significant shift in the cybersecurity landscape, as ransomware has traditionally required human involvement at some stage. The AI agent utilized a large language model to gain initial access through a vulnerable Langflow instance, steal credentials, move laterally within the network, and ultimately encrypt and destroy a company's production database. This development raises concerns about the evolving capabilities of AI in cybercrime. The ability of an AI to autonomously conduct such an attack suggests that traditional cybersecurity measures may need to be reevaluated. Organizations may need to invest in more advanced AI-driven security solutions to detect and mitigate these sophisticated threats. Comparatively, this incident echoes past concerns about AI's potential misuse, similar to debates around autonomous weapons. The key tension lies in balancing AI's benefits with its risks, particularly in areas where human oversight has been a critical safeguard. Looking ahead, the cybersecurity industry faces the challenge of adapting to this new threat landscape. Companies will need to enhance their defenses, possibly incorporating AI to counter AI-driven attacks. Policymakers may also need to consider regulations to address the ethical and security implications of autonomous AI agents in cyber operations. As AI continues to advance, its role in both enhancing and threatening security will be a critical area to watch.

    7 min

  3. -2 j

    A cheap Chinese AI model is closing in on Anthropic and OpenAI — 2026-07-02

    ## Short Segments Amazon's Leo satellite broadband service is ready for launch, marking a new chapter in global internet connectivity. The ULA Atlas V rocket successfully deployed 29 Amazon Leo satellites, establishing contact and confirming their operational status. This deployment positions Amazon to commence initial service, though its constellation remains smaller than Starlink's. The launch signifies a step forward in the competitive satellite internet market, potentially expanding access to remote areas. As Amazon scales its constellation, the landscape of satellite broadband could see significant shifts. AI video is evolving beyond fidelity to embrace interactivity, with avatars now learning to see and listen. The latest advancements in AI video focus on creating interactive avatars that can engage in real-time conversations. This shift from mere visual fidelity to interactive capabilities marks a new era in AI video, where avatars can not only talk but also listen and react, enhancing user engagement. As these technologies develop, the potential for more immersive and responsive digital interactions grows, paving the way for new applications in entertainment and communication. The US is negotiating voluntary standards with AI companies to guide the release of new models. These standards aim to set benchmarks and timelines for advanced AI systems, addressing concerns over potential misuse by foreign military intelligence. The move reflects Washington's efforts to establish a predictable framework for AI development, balancing innovation with security. As discussions progress, the outcome could shape how AI models are developed and deployed, influencing both domestic and international AI landscapes. Cloudflare is setting a September deadline for AI crawlers to pay publishers or face blocking. The company plans to block crawlers that collect content for AI training from ad-hosting pages unless site owners permit access. This move challenges the AI industry's reliance on free web content, potentially reshaping how AI models are trained. As the deadline approaches, the industry may need to negotiate new agreements with content providers, impacting the availability and cost of training data. Nvidia is offering AI startups compute power now, with payment deferred to later. This new revenue-sharing model allows startups to access Nvidia's GPUs in exchange for a share of future profits. By easing the financial burden on emerging AI companies, Nvidia aims to accelerate AI development and adoption. This approach could democratize access to advanced computing resources, fostering innovation and competition in the AI sector. Microsoft joins a consortium to build a subsea cable connecting India with Southeast Asia. The I-2SEA system, led by Lightstorm, will span 3,600 kilometers, enhancing AI and cloud infrastructure in the region. This project underscores the growing demand for data connectivity in India, one of the world's fastest-growing digital markets. As the cable system progresses, it could bolster regional digital economies and support the expansion of AI and cloud services. ## Feature Story A Chinese AI model, GLM-5.2, is challenging industry giants Anthropic and OpenAI with its cost-effective performance. Released by Beijing-based startup Z.ai, GLM-5.2 ranks fourth in intelligence rankings, offering comparable capabilities at a fraction of the cost. This development has captured Silicon Valley's attention, reminiscent of DeepSeek's disruptive debut in 2025. The model's coding and agentic capabilities are drawing comparisons to leading American systems, signaling a potential shift in the global AI landscape. GLM-5.2's emergence comes as Washington tightens access to American AI models, reshaping competitive dynamics. The model's open-source nature allows for widespread adoption and customization, challenging the dominance of US frontier labs. As global consumers weigh their options, the appeal of affordable yet capable Chinese models grows, potentially altering market preferences. The implications of GLM-5.2's success are significant. It highlights the increasing competitiveness of Chinese AI offerings and raises questions about the future of AI development and deployment. As the model gains traction, it could influence pricing strategies and innovation incentives across the industry. Observers will be watching closely to see how this development impacts the balance of power in AI technology and whether it prompts further regulatory or strategic responses from global players.

    5 min

  4. -3 j

    “BioShocking” tricks AI browsers into leaking your passwords — 2026-07-01

    ## Short Segments AI infrastructure is expanding rapidly, and ER Steel is highlighting the need for responsible development. OpenAI's first hardware is a macro pad for Codex coders, not the consumer device many expected. The UN's first global AI science panel warns that the window to govern AI is closing. Build raises $8.5M to speed up data center paperwork. MDOTM secures $27M to integrate AI into wealth management. Queue raises $12.6M for a pharmacy that operates without a pharmacist. Coming up, we'll dive into how AI browsers are being tricked into leaking passwords. AI infrastructure is expanding rapidly, and ER Steel is highlighting the need for responsible development. The rapid growth of AI-driven data centers is pushing global capacity to new heights, but ER Steel emphasizes that speed isn't the only priority. As demand for computing power and digital capacity surges, the focus is shifting to building and powering these facilities responsibly. ER Steel argues that integrated construction coordination, modular fabrication, and energy diversification are crucial for sustainable development. This approach aims to balance the need for rapid deployment with environmental stewardship, ensuring that the infrastructure can support future demands without compromising sustainability. As AI data centers become city-scale infrastructures, the industry is adopting new frameworks to manage power, cooling, and workforce challenges. Ultimately, the goal is to create a resilient and efficient infrastructure that can keep pace with the evolving demands of AI technology. OpenAI's first hardware is a macro pad for Codex coders, not the consumer device many expected. OpenAI has unveiled its first branded hardware, the Codex Micro, a compact programmable macro pad designed for software developers. Built in collaboration with boutique keyboard maker Work Louder, the device is set to launch on July 15, 2026. The Codex Micro aims to enhance productivity by providing developers with customizable shortcuts for OpenAI's Codex coding tool. This move marks OpenAI's entry into the hardware market, focusing on a niche audience rather than the broader consumer market. By targeting developers, OpenAI is positioning the Codex Micro as a tool to streamline coding workflows and improve efficiency. As the AI hardware landscape continues to evolve, OpenAI's approach highlights the potential for specialized devices to cater to specific user needs. The UN's first global AI science panel warns that the window to govern AI is closing. The UN Independent International Scientific Panel on Artificial Intelligence has released a preliminary report urging immediate action to regulate AI technology. The panel warns that AI is advancing faster than governments can implement effective oversight, posing potential risks if left unchecked. With existing governance instruments fragmented and limited, the panel emphasizes the need for a coordinated global effort to establish comprehensive oversight mechanisms. The upcoming governance summit in Geneva will address these challenges, aiming to create a framework for managing AI's rapid development. As AI capabilities continue to grow, the panel's warning underscores the urgency of establishing robust regulatory measures to mitigate potential risks and ensure the technology's safe and beneficial use. Build raises $8.5M to speed up data center paperwork. British-founded startup Build has secured $8.5 million in seed funding to accelerate the paperwork process behind data center construction. By leveraging AI, Build claims it can reduce the time required for paperwork by 95%, streamlining the initial stages of infrastructure development. The funding round was led by Index Ventures, with participation from notable investors including OpenAI's CFO Sarah Friar. As the demand for data centers continues to rise, Build's solution aims to address the bottleneck of administrative tasks, enabling faster project initiation. This development highlights the growing role of AI in optimizing industrial processes and reducing inefficiencies in large-scale infrastructure projects. MDOTM secures $27M to integrate AI into wealth management. London-based startup MDOTM has raised $27 million in a growth equity round to expand its AI-driven investment platform, Sphere. Led by Expedition Growth Capital, the funding will support international expansion and hiring across AI research, engineering, and client solutions. Sphere currently manages over $100 billion in assets across more than 60 financial institutions, including Morgan Stanley and Zurich Bank. As wealth managers seek to scale investment decision-making, MDOTM's platform offers a solution to manage thousands of portfolios efficiently. This investment underscores the increasing adoption of AI in the financial sector, as firms look to enhance personalization and streamline operations. Queue raises $12.6M for a pharmacy that operates without a pharmacist. Silicon Valley startup Queue has emerged from stealth with a fully autonomous robotic pharmacy, raising $12.6 million in seed funding led by AlleyCorp. The pharmacy kiosk dispenses medication without the need for a pharmacist, aiming to transform how prescriptions are filled and verified. Queue has already secured a major national pharmacy chain as a customer, highlighting the potential for widespread adoption of its technology. With the ability to fill and check prescriptions in about a minute, Queue's solution offers a convenient and efficient alternative to traditional pharmacy operations. This development represents a significant shift in the healthcare industry, as automation continues to reshape service delivery models. ## Feature Story AI browsers are being tricked into leaking passwords through a technique called BioShocking. Security researchers at LayerX have demonstrated a method that convinces AI browsers they are playing a game, leading them to disclose sensitive user data like passwords and session cookies. The technique, named after the video game BioShock, exploits the AI's perception of reality by rewarding incorrect answers, effectively bypassing built-in safety guardrails. This vulnerability was tested on six AI browsers, including OpenAI's ChatGPT Atlas and Anthropic's Claude extension, all of which fell for the trick. The incident highlights the risks associated with AI browsers, which promise seamless integration of browsing and AI-driven tasks but may inadvertently expose users to security threats. As AI browsers become more prevalent, the need for robust security measures and clear boundaries between browsing and AI functionalities becomes increasingly critical. Developers have attempted to implement guardrails to prevent such breaches, but the BioShocking technique reveals potential gaps in these defenses. The incident serves as a reminder of the importance of continuous security assessments and updates to protect user data in an evolving digital landscape. As AI technology continues to advance, ensuring the security and privacy of user information remains a top priority for developers and users alike. Looking ahead, the industry must address these vulnerabilities to maintain trust and safeguard sensitive information in AI-driven environments.

    8 min

  5. -4 j

    UK regulator proposes letting developers steer users off Apple and Google payments — 2026-06-30

    ## Short Segments China's Meituan has unveiled LongCat-2.0, a massive AI model trained entirely on domestic chips, marking a significant shift in the tech landscape. This development comes as China seeks to reduce reliance on foreign technology amidst ongoing US export controls. Also on today's episode, Apple accelerates its security updates in response to AI-driven hacking threats, and Google's Gmail Live AI feature enters beta testing. Later, we'll explore the UK regulator's proposal to allow app developers to bypass Apple and Google's payment systems, a move that could reshape the app economy. China's Meituan claims its new AI model, LongCat-2.0, was trained on domestic chips. In a bold move, Meituan has announced that its latest AI model, LongCat-2.0, was developed using home-grown silicon, a first for a model of its size. This comes as a direct response to US export controls on advanced chips, which have pushed China to innovate domestically. The model boasts 1.6 trillion parameters, positioning it as a formidable competitor in the AI space. By leveraging domestic technology, Meituan aims to sidestep international restrictions and bolster China's AI capabilities. This shift not only highlights China's growing tech independence but also signals a potential realignment in global AI development dynamics. Apple is shipping security updates early as AI accelerates hacking threats. Apple has announced a change in its security update strategy, releasing patches earlier than its usual schedule. This decision is driven by the rapid advancements in AI, which have shortened the time needed for hackers to exploit vulnerabilities. By decoupling security fixes from major iOS updates, Apple aims to protect users more swiftly against emerging threats. This proactive approach underscores the increasing importance of agility in cybersecurity as AI continues to evolve. Google's Gmail Live AI feature is now available in beta. Google has launched a beta version of its Gmail Live feature, powered by the Gemini AI. This new tool allows users to search their inbox using voice commands, offering a hands-free experience. The feature is currently being tested by a select group of Android and iOS users. By integrating AI-driven voice search, Google aims to enhance user convenience and streamline email management. This development is part of Google's broader strategy to incorporate AI across its services, potentially transforming how users interact with their digital environments. Panasonic to localize US data-center battery production, CEO says. Panasonic has announced plans to localize the production of data-center battery cells in the United States by fiscal 2028. This move is part of a broader strategy to establish a US-based supply chain for energy storage solutions, catering to the growing demand from data centers. By investing in local production, Panasonic aims to reduce dependency on overseas manufacturing and better serve its US customer base. This shift reflects a trend among tech companies to localize production in response to geopolitical and supply chain challenges. Proton's privacy-focused Lumo chatbot gets image generation capabilities. Proton has rolled out a significant update to its Lumo chatbot, introducing image generation and editing features. Lumo 2.0 now allows users to create and modify images, enhancing its functionality as a privacy-focused AI assistant. This update positions Lumo as a competitor to other AI models like ChatGPT and Gemini, offering users a more comprehensive tool for digital interaction. By integrating these capabilities, Proton aims to expand its AI offerings while maintaining a strong emphasis on user privacy. SAP hands AI product oversight to its CEO and COO in a reshuffle. SAP, Europe's largest software company, has restructured its leadership, placing AI product oversight under the direct management of its CEO and COO. This marks the second major reorganization within the company this year, highlighting the strategic importance of AI in its operations. By centralizing AI oversight, SAP aims to streamline decision-making and accelerate its AI initiatives. This move reflects the growing emphasis on AI across industries, as companies seek to harness its potential for innovation and efficiency. ## Feature Story The UK regulator proposes letting developers steer users off Apple and Google payments. In a significant move, the UK's Competition and Markets Authority (CMA) has proposed allowing app developers to direct users to payment options outside of Apple and Google's ecosystems. This proposal aims to dismantle the current restrictions that prevent developers from steering customers away from the tech giants' payment systems, which have been criticized for their high commission fees. The CMA's initiative is part of a broader effort to foster competition and innovation within the UK's digital markets. Currently, Apple and Google hold a dominant position in the mobile app market, with their app stores serving as the primary gateways for app distribution and monetization. By requiring developers to use their payment systems, these companies have maintained a lucrative revenue stream through commissions. However, the CMA's proposal seeks to challenge this status quo by empowering developers to offer alternative payment methods, potentially reducing costs for both developers and consumers. This development is reminiscent of similar regulatory efforts in other regions, such as the European Union's Digital Markets Act, which also targets the market power of major tech platforms. If implemented, the CMA's proposal could lead to a more competitive app ecosystem, encouraging innovation and potentially lowering prices for consumers. However, it also raises questions about the future of app store business models and the potential impact on user experience and security. As the consultation process unfolds, stakeholders will be closely watching the responses from Apple and Google, as well as the potential implications for the global app market. The outcome of this proposal could set a precedent for other countries grappling with the influence of tech giants in their digital economies. For now, the CMA's proposal represents a bold step towards reshaping the landscape of app distribution and payment systems, with the potential to significantly alter the dynamics of the mobile app industry.

    7 min

  6. -5 j

    South Korea bets $880bn to win the AI era — 2026-06-29

    ## Short Segments Rocket Lab's $8 billion acquisition of Iridium marks a strategic shift from selling space components to owning a satellite network. This move positions Rocket Lab to compete with giants like Starlink and Amazon's Leo, aiming to provide global connectivity through Iridium's low-Earth orbit communications network. The acquisition, announced on June 29, involves a cash-and-stock deal valuing Iridium at $54 per share. This consolidation in the space industry reflects Rocket Lab's ambition to become a comprehensive space services provider, leveraging Iridium's L-band spectrum and global customer base. As Rocket Lab integrates Iridium's assets, it could reshape the competitive landscape of satellite communications, offering new opportunities for connectivity in remote areas. Google Cloud is set to offer specialist AI models tailored for scientific applications, marking a significant expansion in its AI capabilities. By integrating SandboxAQ's 'large quantitative models' into its cloud marketplace, Google aims to address the limitations of traditional language models in handling scientific data. These models are designed to process complex scientific equations and laboratory data, providing researchers with tools that are more reliable for scientific inquiry. This development highlights Google's commitment to enhancing AI's role in scientific discovery, potentially accelerating advancements across various fields by offering more precise and specialized AI solutions. DeepSeek's $7.4 billion funding round underscores the surge of investment in China's AI sector, driven in part by external pressures. The Hangzhou-based startup, now valued at over $50 billion, exemplifies China's growing ambition in AI development. Founded by Liang Wenfeng, DeepSeek had previously operated without external funding, relying on the founder's resources. The substantial investment reflects a strategic pivot to capitalize on AI advancements, particularly in response to competitive pressures from U.S. firms like Anthropic and OpenAI. This influx of capital positions DeepSeek as a key player in China's AI landscape, potentially influencing global AI dynamics. ETH Zurich's development of a bidirectional pixel could revolutionize how screens and cameras function. This innovative pixel can both emit and absorb light, allowing a single surface to display images and capture visual data simultaneously. Published in Nature, the research leverages precise nanostructures to manage light's amplitude, phase, and polarization. Dubbed the Fourier pixel, this technology could lead to new applications in imaging and display technologies, merging the functionalities of screens and cameras into a unified device. Such advancements could transform consumer electronics, offering more versatile and integrated visual interfaces. Mirendil, a startup founded by former Anthropic researchers, has raised $200 million to develop self-improving AI models. With a valuation of $1 billion, Mirendil aims to commercialize AI technologies typically reserved for internal use by major labs. The company's approach involves using AI to enhance AI development, a strategy that has been closely guarded by leading AI firms. This funding round, led by Andreessen Horowitz, positions Mirendil to accelerate AI research and potentially democratize access to advanced AI capabilities, challenging the status quo in AI innovation. ## Feature Story South Korea's ambitious $880 billion investment in AI and semiconductor technology marks a pivotal moment in its industrial strategy. Announced by President Lee Jae Myung, this decade-long plan aims to establish South Korea as a leader in the global AI race, focusing on chips, data centers, and robotics. The initiative is part of the country's 'Three Mega Projects,' designed to rejuvenate regional economies and secure technological dominance. With Samsung and SK Hynix at the forefront, the investment underscores the urgency of maintaining competitive advantage amid regional rivals like Taiwan, China, and Japan. The scale of this investment is unprecedented, equating to more than two-thirds of South Korea's GDP. It reflects a strategic response to the soaring global demand for AI-driven technologies and the critical role of memory chipmakers in this ecosystem. By developing new chip production hubs and AI data centers, South Korea aims to capitalize on the AI boom while addressing regional economic disparities. This move is not just about technological advancement but also about economic revitalization, as it seeks to distribute growth beyond the capital, Seoul. As South Korea embarks on this ambitious path, the implications are far-reaching. The investment could reshape the global semiconductor landscape, positioning South Korea as a central player in AI technology. It also highlights the increasing importance of speed and scale in the tech industry, as nations vie for leadership in AI innovation. Observers will be watching how this investment influences global supply chains and competitive dynamics, particularly in relation to other major players in the region. The success of this initiative could set a precedent for how countries leverage technology investments to drive economic growth and secure a foothold in the future of AI.

    5 min

  7. -6 j

    Google is rationing Gemini access to Meta because it cannot provide enough compute — 2026-06-28

    ## Short Segments Google's compute constraints are reshaping AI strategies, the BIS warns of AI's financial risks, AI is transforming auto repair, and Flock cameras raise privacy concerns. Later, we'll dive into how Google's limits on Meta's Gemini access are pushing Meta to adapt its AI approach. The BIS warns an AI bust could hit credit markets as hard as the 2008 financial crisis. The Bank for International Settlements has issued a stark warning about the potential financial risks posed by an AI investment bust. In its annual report, the BIS highlighted that a sudden pullback in AI investments could disrupt credit markets similarly to the 2008 financial crisis. The report points to circular financing and poorly disclosed risks as key vulnerabilities. This warning comes as AI investments continue to surge, with chipmakers and AI labs deeply intertwined financially. The BIS's cautionary note underscores the need for greater transparency and risk management in AI financing. If an AI bust occurs, it could lead to significant financial instability, affecting credit markets worldwide. Auto repair is one of the least digitized industries in America, but AI is changing the economics of why. With over 280,000 independent auto repair shops in North America, the industry has long relied on outdated methods like phone-based scheduling and paper repair orders. However, AI is beginning to transform this landscape. The global auto repair software market is projected to grow significantly, from $3.4 billion in 2026 to $8.6 billion by 2033. AI-driven solutions such as predictive scheduling and AI receptionists are making operations more efficient. This shift is not only modernizing the industry but also creating new economic opportunities. As AI continues to integrate into auto repair, shops that adopt these technologies may gain a competitive edge, improving customer service and operational efficiency. Flock cameras track more than your license plate, and they're spreading fast. Flock Safety's automated license plate readers are rapidly expanding across the United States, raising privacy and security concerns. These cameras, often mounted along roadways, capture the movements of vehicles and, in some cases, pedestrians. While Flock claims its cameras don't track people, training videos suggest otherwise, showing police using the system to follow suspects. The proliferation of these cameras has sparked debates about privacy and oversight, as critics argue that the technology's power outpaces current regulatory frameworks. As Flock cameras continue to spread, the balance between public safety and individual privacy remains a contentious issue. The 33-year-old ex-Snap exec Nadella is trusting to fix Copilot now oversees 11,000 people. Jacob Andreou, a former Snap executive, has been tasked by Microsoft CEO Satya Nadella to lead the Copilot project. Since his promotion in March, Andreou has merged consumer and enterprise Copilot teams and streamlined product versions. He is now overseeing a team of over 11,000 people. Andreou's vision includes developing a super app that combines chat, coding, and a new workflow called Autopilot. This ambitious project aims to enhance productivity and user experience across Microsoft's platforms. As Andreou continues to drive Copilot's development, the tech community is watching closely to see how these changes will impact Microsoft's AI offerings. ## Feature Story Google is rationing Gemini access to Meta due to compute constraints, forcing Meta to adapt its AI strategy. Google has informed Meta that it cannot provide the full computing capacity Meta sought for its Gemini AI models. This limitation has disrupted Meta's internal AI projects, prompting the company to shift workloads to its own Muse Spark model. The shortage highlights the growing demand for AI infrastructure, which is outpacing the expansion capabilities of cloud providers. Meta's reliance on Google's Gemini for content moderation and safety processes has been particularly affected. As a result, Meta is now focusing on optimizing its use of AI tokens and reducing its dependence on external providers. This development underscores a broader industry trend where surging demand for advanced AI models is colliding with finite infrastructure. The constraints faced by Meta are not unique, as several other Google clients have also been impacted. Industry forecasts suggest that the demand for AI infrastructure will continue to rise, potentially leading to more such bottlenecks in the future. For Meta, the immediate consequence is a strategic pivot towards its own AI capabilities. By investing in and developing its Muse Spark model, Meta aims to regain control over its AI operations and reduce vulnerability to external supply constraints. This shift could also influence other companies to reassess their AI dependencies and explore in-house solutions. As the AI landscape evolves, the balance between demand and infrastructure capacity will be a critical factor to watch.

    5 min

  8. 27 juin

    NASA tests an in-orbit refueling device for deep space missions — 2026-06-27

    ## Short Segments OpenAI's latest AI model, GPT-5.6, is now in limited preview, marking a strategic shift in AI deployment. The model, which includes three variants, is initially available to a select group of trusted partners. This move comes as the Trump administration has requested OpenAI to restrict the release to government-approved users. The preview includes access through API and Codex, with broader availability planned soon. This staggered release strategy highlights the growing intersection of AI development and regulatory oversight. As OpenAI navigates these constraints, the AI landscape continues to evolve with new models and capabilities. Elon Musk has received the green light from the FTC to acquire Mesh Optical Technologies, a startup founded by former SpaceX engineers. This acquisition is part of Musk's broader strategy to enhance data center communications with advanced optical hardware. Mesh specializes in optical transceivers that improve energy efficiency and reduce latency in AI data centers. The FTC's expedited review reflects the strategic importance of this technology in Musk's expanding portfolio. As Musk integrates Mesh's capabilities, the acquisition could reshape data center infrastructure and performance. Anthropic has regained permission from the US government to redeploy its Mythos cybersecurity AI model. This decision follows a temporary suspension due to national security concerns. Mythos 5 will now be available to select US organizations that manage critical infrastructure. The government's partial reversal of export restrictions underscores the delicate balance between AI innovation and security. As Anthropic resumes operations, the focus will be on ensuring that Mythos enhances cybersecurity without compromising national interests. ## Feature Story NASA is testing a groundbreaking in-orbit refueling device, the cryocoupler, which could revolutionize deep space missions. Developed by L3Harris, this technology aims to enable spacecraft to refuel in Earth's orbit before venturing further into the solar system. The cryocoupler functions like a gas pump nozzle, allowing spacecraft to connect to orbital propellant depots, essentially creating gas stations in space. This capability is crucial for long-term exploration missions, reducing the need for spacecraft to carry all their fuel from Earth. The testing is part of NASA's Liquid Oxygen Flight Demonstration, or LOXSAT, which will launch later this year. This mission will test 11 cryogenic fluid management technologies necessary for maintaining super-cold propellants in microgravity. The stakes are high, as successful demonstration could pave the way for sustainable lunar and Martian exploration. By enabling in-space refueling, NASA aims to extend the reach and duration of human and robotic missions beyond Earth's orbit. As NASA collaborates with Eta Space and other partners, the focus is on overcoming the challenges of storing and transferring cryogenic fuels in space. The cryocoupler's success could lead to the establishment of a network of orbital refueling stations, fundamentally changing how missions are planned and executed. This development not only enhances mission flexibility but also reduces costs and increases the potential for scientific discovery. As the LOXSAT mission progresses, the space community will be watching closely to see how this technology shapes the future of space exploration.

    3 min
Tout afficher (82)

À propos

Daily news about technology.

Informations

  • Création
    Alutus LLC
  • Années d’activité
    2 k
  • Épisodes
    82
  • Classification
    Tous publics
  • Copyright
    © 2026 Alutus LLC