Category: Artificial intelligence

  • Nvidia Dismisses China’s H20 GPU Security Issues Amid US Export Rules

    Nvidia Dismisses China’s H20 GPU Security Issues Amid US Export Rules

    Key Takeaways

    1. Nvidia emphasizes the importance of cybersecurity and denies claims that its products allow remote access or control.
    2. China’s concerns arise from a draft U.S. law requiring disclosure of advanced chip locations to prevent exports to embargoed countries.
    3. The H20 chip, designed for the Chinese market, lacks a hardware tracking module found in restricted components.
    4. Experts have mixed opinions on China’s approach to Nvidia, with some seeing hardware as leverage and others viewing pressure as mostly symbolic.
    5. Despite regulatory challenges, demand for Nvidia products in China remains strong, with ongoing imports and investments in domestic alternatives.

    Nvidia has stated that “cybersecurity is very important” and denied claims that any of its products allow remote access or control. This statement came after the Cyberspace Administration of China (CAC) called the company to talk about possible risks to user data related to the H20 artificial-intelligence GPU.

    Beijing’s Response

    China’s worries are partly in reaction to a draft U.S. law that would require advanced chips sold internationally to disclose their location. This law aims to stop these chips from being sent to countries under embargo. This situation comes shortly after the U.S. lifted an April ban on H20 exports, which had already been adjusted to meet the 2023 performance limits.

    H20 Specifications

    The H20 chip is a simplified version of the H100 and does not include a hardware tracking module, unlike fully restricted components. Reports from the industry suggest that this chip was specifically designed for the Chinese market after tighter U.S. controls were put in place.

    Varying Opinions on China’s Strategy

    Experts have different views on how aggressively China will pursue this issue. Tilly Zhang from Gavekal Dragonomics believes that the government now views Nvidia hardware as leverage due to the rise of stronger domestic alternatives. On the other hand, Charlie Chai from 86Research thinks that the pressure will mainly be symbolic, since many Chinese developers are still heavily reliant on Nvidia’s CUDA software.

    Despite facing regulatory challenges—including an ongoing antitrust probe—demand for Nvidia accelerators in China remains strong. Reuters has reported a recent order for about 300,000 H20 units from TSMC. Other U.S. suppliers like Micron have also gone through similar security assessments, highlighting Beijing’s strategy of using these investigations while local semiconductor capabilities develop.

    Future Outlook

    Currently, the CAC has not provided specific counter-measures. Without a strong large-scale alternative, analysts predict that China will continue to import Nvidia GPUs, but with increased scrutiny, while also boosting investments in domestically produced accelerators from companies like Huawei, Biren, and Cambricon.

    Source:
    Link

     

    Tags: ,
  • “New Algorithm Enables Cheaper, Accurate, Lightweight AI Models”

    “New Algorithm Enables Cheaper, Accurate, Lightweight AI Models”

    Key Takeaways

    1. A new approach effectively manages symmetric data in machine learning, improving efficiency in calculations and data requirements.
    2. Symmetries are important as they convey essential information about data, and incorporating them into machine learning is crucial.
    3. The research introduces a novel algorithm that combines algebra and geometry principles to honor symmetry in learning.
    4. This method requires fewer data samples for training, potentially enhancing model precision and flexibility.
    5. The findings could lead to more robust and resource-efficient AI models, with applications in materials discovery, astronomy, and climate pattern analysis.

    A team of scientists has tackled a key issue in machine learning by developing the first approach that effectively manages symmetric data while ensuring efficiency in both calculation and data requirements. The primary difficulty lies in AI’s tendency to misinterpret symmetry; for instance, it may view a rotated molecule as a brand new entity rather than recognizing it as the same structure.

    The Significance of Symmetry

    Symmetries carry essential information that nature conveys about the data, and it is crucial to incorporate them into our machine-learning frameworks. “We’ve now shown that it is possible to do machine-learning with symmetric data in an efficient way,” stated Behrooz Tahmasebi, an MIT graduate student and one of the main authors.

    A New Approach to Algorithms

    Some existing models, such as Graph Neural Networks, are capable of addressing symmetry, but the reasons behind their effectiveness remain unclear. The MIT researchers adopted a novel strategy by developing a new algorithm that merges mathematical principles from algebra and geometry. This allows for a system that can efficiently learn while honoring symmetry.

    This method, which is proven to be efficient, needs fewer data samples for training, which can enhance a model’s precision and flexibility. The researchers believe their findings may pave the way for more robust and resource-efficient AI models applicable in various fields, “from discovering new materials to identifying astronomical anomalies and deciphering complex climate patterns.” Their research was recently showcased at the International Conference on Machine Learning.

    Source:
    Link

     

    Tags:
  • Samsung HBM3E Demand Surges, Pressuring Prices and Margins

    Samsung HBM3E Demand Surges, Pressuring Prices and Margins

    Key Takeaways

    1. Samsung is increasing HBM3E production faster than market demand, which may lower prices temporarily.
    2. Rising contract prices for standard DRAM will narrow profit margins between HBM3E and regular memory.
    3. Competitors like SK Hynix and Micron are producing 12-stack HBM3E, risking inventory buildup before demand rises.
    4. Samsung’s semiconductor division profits have dropped 94% year-on-year, prompting cost cuts to regain Nvidia’s business.
    5. Samsung is considering price cuts to compete, with future market dominance depending on cost and yield rather than just bandwidth.

    Samsung informed investors that the production of the fifth-generation HBM3E is increasing quicker than the demand in the market, which the company believes “will affect market prices for the time being.”

    Profitability Concerns

    Management pointed out that the rising contract prices for regular DRAM are going to narrow the previously large profit margin between HBM3E and standard memory in the second half of the year. This will limit the potential for increased margins, even with a rise in volumes.

    Customer Shift and Competitive Landscape

    This alert comes at a time when clients like Nvidia and AMD are moving towards the 12-stack HBM3E for their next-gen AI accelerators. Other competitors, including SK Hynix and Micron, are already producing this denser version in large amounts, which increases the risk of inventory accumulation before the anticipated demand surge happens.

    Internally, Samsung’s semiconductor division is facing challenges: its quarterly operating profit has plummeted by 94 percent year-on-year to 400 billion Korean won ($287 million) due to export controls and inventory adjustments affecting the results. To stop this decline, the company is cutting HBM3E production costs in hopes of regaining Nvidia’s business that has mostly shifted to SK Hynix.

    Growth in Memory Revenue

    Memory revenue has seen an 11 percent increase compared to Q1 as HBM3E shipments have risen. Samsung intends to boost production of 128 GB DDR5, 24 GB GDDR7, and 8-gen V-NAND by the end of the year. A deal worth $16.5 billion to manufacture Tesla’s next-gen AI6 chips in Texas should also help stabilize foundry usage, although new 15 percent US tariffs on Korean products cast a shadow over the demand forecast.

    Price Concessions and Market Dynamics

    Sources within the industry mention that Samsung has proposed price cuts. Meanwhile, Nvidia is confirming its 12-layer stacks, indicating that market dominance in the next cycle may depend more on cost and yield rather than just bandwidth. If Samsung can achieve high-yield and lower-cost production ahead of its competitors, the company might regain some market share in the profitable AI-memory sector—however, any error could deepen the oversupply issue that is currently pressuring HBM3E prices.

    Source:
    Link

    Tags: , ,
  • Alpon X5: High-Performance AI Edge PC with Raspberry Pi Technology

    Alpon X5: High-Performance AI Edge PC with Raspberry Pi Technology

    Key Takeaways

    1. Alpon X5 is a new edge AI system seeking funding on Kickstarter, focusing on processing data close to the device.
    2. The system offers enhanced AI performance with 25 TOPS of processing power, particularly in image processing tasks.
    3. It operates on a Raspberry Pi CM5 and features user-friendly software for easier implementation.
    4. Connectivity options include a cellular modem, WiFi 5, Ethernet, USB 3.0 ports, HDMI, and GPIO for external sensors.
    5. Early bird pricing for backers is set at $549, with potential risks associated with crowdfunding projects.

    Alpon X5 is a newly launched edge AI system that is currently seeking funding through Kickstarter. The “edge” concept refers to processing data nearer to the device or application rather than in a central location. For instance, this might occur on a small computer system situated near an industrial device like a CNC machine. This allows the Alpon X5 to manage specific processes while also sending data to a central monitoring system or the cloud.

    Enhanced AI Performance

    Those who use the Alpon X5 are expected to experience better AI acceleration, particularly in areas like image processing and understanding what images show. An example could be spotting fires using a surveillance camera. The system is equipped with a DeepX DX-M1 M.2 module that claims to deliver 25 TOPS of AI processing power while drawing only 3.5 watts. It operates on a Raspberry Pi CM5, and utilizing the appropriate software is reported to be straightforward.

    Connectivity Options

    The device includes a cellular modem that works with an eSIM card and is intended to support 4G mobile networks. In addition, it supports WiFi 5 and Ethernet connections. Other notable features are two USB 3.0 ports, an HDMI port, and a camera. External sensors and actuators can also be linked through GPIO.

    For backers of the crowdfunding initiative, the Alpon X5 is available at an early bird price of $549. As is the case with any crowdfunding effort, participants should be aware of the potential risks, including the possibility that the product may not reach the market.

    Source:
    Link

    Alpon X5: High-Performance AI Edge PC with Raspberry Pi Technology

     

  • Create Your Own Series with Amazon and AI Soon Available

    Create Your Own Series with Amazon and AI Soon Available

    Key Takeaways

    1. Amazon is investing in Fable Simulation, a startup focused on AI for virtual reality experiences.
    2. Fable’s Showrunner platform allows users to create and direct their own series, blending famous characters with original stories.
    3. The platform aims to transform entertainment by enabling audiences to produce new episodes of their favorite shows.
    4. Showrunner is expected to launch for free soon, with a subscription fee of 10 to 20 euros per month afterward.
    5. Ethical concerns exist regarding potential misuse of the AI technology, highlighting the need for content creation restrictions.

    AI is rapidly becoming a significant aspect of our everyday lives, and it’s making its way into various industries. However, a revolution may be on the horizon, as you could soon have the chance to craft your own series with the assistance of this technology and Amazon.

    Amazon’s Bold Move

    While the swift rise of AI might alarm many job sectors, Amazon appears to have a different outlook, aiming to capitalize on large-scale innovation. As reported by Variety, the tech giant has put an undisclosed amount into the startup Fable Simulation. If you’re not aware, this company focuses on using AI for virtual reality experiences.

    Create Your Own Series

    Fable also runs the Showrunner platform, which enables you to take the reins as the director of your own series. It’s interesting to note that Showrunner allows you to produce episodes while narrating the story you wish to tell. This means you can combine a variety of worlds, featuring both famous characters and original heroes or villains.

    Although this initiative might seem entirely unrealistic, the company is striving for even more. To roll out such a feature, significant resources are needed to ensure they deliver top-notch AI that meets users’ expectations perfectly. Edward Saatchi, Fable’s CEO and founder, shared with Variety that he is negotiating with Disney to secure rights for several of their licenses.

    User Engagement

    While this could be seen as a drawback compared to other job-generating productions, Edward Saatchi has his thoughts on the matter: “Hollywood streaming services are about to become two-way entertainment: audiences watching a season of a show and loving it will now be able to make new episodes with a few words and become characters with a photo.” In simpler terms, if you want your favorite series to continue, you can take control of it yourself.

    The platform is expected to be available for free in the upcoming months. After that time, you should be prepared to pay a subscription fee that will range from 10 to 20 euros per month. It’s also noteworthy that the testing phase has already drawn in over 10,000 users, indicating a strong interest from the audience.

    Ethical Concerns

    Nevertheless, beyond the potential for unfair competition in this field, the implementation of free-access AI could spark several issues. Indeed, some users might misuse it, leading to ethical dilemmas. Therefore, it would be prudent for Fable to establish certain restrictions on content creation.

    Source:
    Link

    Tags: ,
  • Germany’s First AI-Generated News Program Launches on TV Channel

    Germany’s First AI-Generated News Program Launches on TV Channel

    Key Takeaways

    1. KI-Welt is a weekly news magazine show on Welt TV, fully created by artificial intelligence, airing every Thursday at 2:45 pm CET.
    2. The program features a virtual avatar that uses a synthetic voice to engage with viewers while AI manages content creation and topic selection.
    3. This initiative is an experiment to explore the integration of AI in journalism and seeks to discover new digital production methods.
    4. Welt TV is the first German broadcaster to launch a fully AI-generated news program, emphasizing human oversight to ensure journalistic standards.
    5. The future of KI-Welt is uncertain, with public reception and technological advancements likely to impact its long-term viability in the media landscape.

    German television channel Welt TV has introduced KI-Welt, which translates to “AI World,” a news magazine show that is entirely created by artificial intelligence. This weekly program will air every Thursday at 2:45 pm CET and will cover subjects like AI, robotics, and future technologies. Although the AI system manages everything from choosing topics to writing scripts and presenting, human journalists still oversee the process to ensure it meets journalistic standards.

    Virtual Avatar Engagement

    A significant feature of the show is a virtual avatar that interacts with viewers during its approximately 15-minute duration, using a synthetic voice. The AI technology at work independently reviews news sources, produces content, and crafts new scripts.

    Exploring New Digital Methods

    Welt.de reports that this initiative is seen as an experiment by the company, aiming to investigate how far AI can be embedded in an editorial context. The goal is to find out new digital production methods that could be utilized in the future.

    “The new KI-Welt show is completely produced by AI, from topic research to script creation and presentation through a virtual avatar.”

    A First in German Television

    Welt TV is the first German station to incorporate a fully AI-generated news program into its regular lineup. Other broadcasters have only tinkered with parts of the process, like using avatar newsreaders or automated scripts. The company claims that KI-Welt is a unique magazine program and will not replace any current shows. The newsroom at Welt TV will remain unchanged.

    The rise of entirely AI-produced programs brings up fresh concerns regarding transparency, accountability, and media literacy. This is why the presence of human oversight by journalists is stressed as a vital measure. Meanwhile, the project showcases the increasing importance of AI technologies in journalism and might set a precedent for other media outlets.

    The Future of KI-Welt

    It remains to be seen if KI-Welt can secure its place in the long term. At this moment, the show is more of an experiment with an unpredictable future. Public reception and possible technological improvements could greatly influence how such formats are integrated into the media landscape going forward.

    Source:
    Link

  • Meta Invests in Personal Superintelligence and Powering Silicon

    Meta Invests in Personal Superintelligence and Powering Silicon

    Key Takeaways

    1. Meta is shifting focus from metaverse goals to “personal superintelligence” as the next phase in consumer computing.
    2. Zuckerberg emphasizes AI as a personalized extension of human abilities rather than a tool for mass automation.
    3. Meta is significantly increasing its AI budget and has created a new division, Superintelligence Labs, to enhance AI models.
    4. Hardware upgrades, including custom accelerators and a next-gen Meta Training and Inference Accelerator, are being implemented to support AI advancements.
    5. Meta is pivoting from its costly metaverse investments to capitalize on the immediate market potential of AI, aiming to establish a competitive edge in personalized superintelligence.

    Meta is changing its long-term research focus to what CEO Mark Zuckerberg describes as “personal superintelligence.” In a public letter shared before the company’s Q2 2025 earnings call, he highlighted this idea as the next phase in consumer computing, marking a shift away from previous metaverse goals and from the industry’s view of artificial general intelligence primarily as a tool for replacing workers.

    A New Perspective on AI

    Zuckerberg believes that AI ought to function as a personalized extension of human abilities instead of a broad system for mass automation. The letter stresses the importance of individual choice: according to Meta, superintelligence should adjust to personal objectives, everyday situations, and creative desires. This stance implicitly challenges the enterprise-focused approaches taken by firms like OpenAI and Google.

    Building the Infrastructure

    Creating real-time, context-sensitive models on a global scale will need significant infrastructure. As a result, Meta has significantly increased its AI budget by several billion dollars, hired experts from top research institutions, and set up a new division called Superintelligence Labs, led by Alexandr Wang, the former head of Scale AI. This group is tasked with enhancing Llama-class foundation models and investigating new architectures that are designed for low-latency inference.

    Hardware Innovations in Progress

    To back these models, hardware upgrades are also taking place. Inside sources reveal that custom accelerators are now working alongside Nvidia H100 and A100 GPUs in Meta’s data centers. Additionally, a next-gen Meta Training and Inference Accelerator (MTIA) is set to be completed later this year. The focus on proprietary silicon aligns with Google’s TPU strategy and hints at future uses in wearable devices, where energy efficiency is vital.

    Meta’s efforts in the metaverse have already led to losses of over $60 billion in Reality Labs. However, AI is gaining immediate traction in the market, and the company seems ready to adjust its investments. Whether personalized superintelligence will become a key platform for consumers depends on Meta’s capability to grow both its algorithms and custom hardware before competitors establish their own ecosystems.

    Source:
    Link

    Tags: ,
  • Zhaoxin Launches First NPU CPU and 96-Core Server Chip at WAIC 2025

    Zhaoxin Launches First NPU CPU and 96-Core Server Chip at WAIC 2025

    Key Takeaways

    1. Zhaoxin launched two new processors: the KaiXian KX-7000N for client-class AI applications and the Kaisheng KH-50000 for data centers.
    2. The KX-7000N features a built-in neural-processing unit for enhanced on-device AI inference, aimed at future AI PCs (AIPCs) for tasks like language modeling and speech recognition.
    3. The KH-50000 processor significantly increases core count to 96, enhances cache capacity to 384 MB, and supports advanced memory and I/O capabilities for AI training clusters.
    4. Zhaoxin’s offerings include a range of edge-to-cloud solutions, emphasizing domestic production and open-source software for secure and cost-effective technology.
    5. The company’s dual launch reflects China’s push for self-sufficiency in semiconductor technology, aiming to compete with established x86 manufacturers.

    At the World Artificial Intelligence Conference (WAIC) 2025 held in Shanghai, Zhaoxin introduced two proprietary processors aimed at different segments of the fast-growing AI sector. The KaiXian KX-7000N marks the company’s inaugural client-class CPU equipped with a built-in neural-processing unit. On the other hand, the Kaisheng KH-50000 is a data center-grade chip that advanced Zhaoxin’s server capabilities to a remarkable 96 cores.

    Innovations in AI CPUs

    The KX-7000N, which builds upon the KaiXian KX-7000 series, incorporates a heterogeneously integrated NPU meant to enhance on-device inference. Zhaoxin promotes this chip as a core component for future AI PCs, or “AIPCs,” which are designed to handle large language models, speech recognition, and local image generation. During demonstrations at WAIC, the company showcased personal AI assistants, smart office solutions, and real-time media production, all executed offline on prototype systems created in collaboration with partners such as Lenovo KaiTian.

    Advancements in Data Center Chips

    For business applications, the newly launched Kaisheng KH-50000 triples the core count compared to its KH-40000 predecessor and increases the L3 cache to 384 MB—matching the capacity of competing AMD Epyc 9004 series. This processor also features 128 PCIe 5.0 lanes, supports 12-channel DDR5 ECC memory, and includes a proprietary ZPI 5.0 interconnect enabling dual- and quad-socket setups for up to 384 cores per node. Zhaoxin asserts that this platform delivers the compute density and high-speed I/O essential for AI training clusters, high-density servers, and various heterogeneous accelerator deployments.

    A Broader Vision

    In addition to the main product launches, Zhaoxin’s booth showcased an extensive range of edge-to-cloud solutions, such as AI workstations, educational devices, and specialized application stacks that cover areas from document automation to medical imaging. By combining domestically manufactured CPUs with an open-source software framework, the firm claims it can provide secure and cost-efficient alternatives in markets that have typically depended on foreign processors.

    This dual launch highlights China’s larger initiative for self-sufficiency in cutting-edge semiconductor technology. Although specific clock speeds and architectural details are still confidential, Zhaoxin’s strategy now encompasses client, workstation, and server AI tasks—indicating a focused effort to challenge established x86 manufacturers regarding both performance and features.

    Source:
    Link

    Tags:
  • AI Workforce Impact: Jobs Most Likely to Be Replaced Revealed

    AI Workforce Impact: Jobs Most Likely to Be Replaced Revealed

    Key Takeaways

    1. AI Applicability Score: The study calculated an “AI applicability score” to measure how well Microsoft Copilot assists with various job tasks and their success rates.

    2. Core Task Performance: Occupations that involve essential tasks like gathering information, writing, and clarifying ideas receive higher applicability scores due to Copilot’s effective assistance.

    3. Top Job Categories: The leading jobs identified are mainly in sectors like sales, office support, media, and education, which focus on information creation and sharing.

    4. Limitations in Physical Tasks: Jobs requiring physical skills or direct human interaction, such as nursing and manual labor, ranked lower in AI applicability.

    5. Current AI Strengths: Generative AI excels in language processing, data retrieval, and structured guidance, suggesting its immediate impact is strongest in roles centered on information exchange.

    A recent study analyzing 200,000 interactions in the U.S. with Microsoft Copilot has mapped these chats to various work tasks as defined in the U.S. O*NET database. The study then connects these tasks to specific job categories. The researchers calculated an “AI applicability score” which indicates how frequently Copilot assists with the tasks of different occupations, how often these tasks are successfully completed, and how well these tasks are represented in job descriptions.

    High Scores for Core Tasks

    Occupations achieve a high applicability score when Copilot effectively aids in completing essential tasks that constitute a significant portion of that job. For instance, tasks like gathering information, writing text, and clarifying ideas tend to have high completion rates. Consequently, jobs that heavily involve these activities usually rank higher on the list.

    Leading Jobs Based on Data

    According to the data from the study, the top ten jobs identified are primarily related to creating, customizing, or sharing information. These roles are typically found in sectors such as sales, office support, media, and education, among other fields that require substantial knowledge. In contrast, jobs that focus on physical tasks, manual labor, or operating machinery—like nursing assistants, massage therapists, and roofers—are ranked much lower.

    Generative AI today performs exceptionally well with tasks that require language, data retrieval, and organized guidance. However, it falls short in areas that necessitate hands-on abilities or direct interaction with people. Although technology may evolve and expand its capabilities over time, the data currently indicates that its most significant immediate effects will be felt in positions where the exchange of information is the main focus.

    Source:
    Link

    Tags:
  • Helion Starts World’s First Fusion Power Plant with Microsoft Support

    Helion Starts World’s First Fusion Power Plant with Microsoft Support

    Key Takeaways

    1. Helion Energy is building the world’s first commercial fusion power plant in Malaga, Washington, expected to provide electricity to Microsoft data centers by 2028.

    2. Fusion energy merges atoms to generate power, offering a cleaner alternative to traditional nuclear energy with no carbon emissions and minimal long-term waste.

    3. Sam Altman, CEO of OpenAI, has supported Helion since 2014, highlighting a trend among tech leaders to address sustainability and infrastructure alongside software development.

    4. Helion aims to achieve net energy gain by 2028, which would be a significant milestone for commercial fusion energy, supported by Microsoft’s commitment to using this power in its operations.

    5. The success of Helion’s fusion project could provide a scalable and reliable energy source for AI systems, aligning with the growing demand for high-density computing resources and contributing to a sustainable future.

    Fusion startup Helion Energy, which has the support of OpenAI’s CEO Sam Altman, has kicked off construction on a project that might turn into the first commercial fusion power plant in the world. This facility, situated in Malaga, Washington, is anticipated to start delivering electricity to Microsoft data centers by 2028.

    A New Era for Fusion Power

    This partnership, along with a power purchase agreement with Microsoft, serves as a clear indicator that nuclear fusion could transition from theoretical concepts in physics books to actual, functioning infrastructure within this decade.

    In contrast to conventional nuclear energy, which relies on splitting atoms, fusion merges them together, generating energy without carbon emissions, risks of meltdowns, and only minimal long-term waste. This technology has often been seen as something from the future, however, Helion’s Polaris reactor aims for a practical objective: to produce electricity at grid scale, without the need for steam or turbines.

    Cleaner Power for AI

    Should this endeavor prove successful, it could provide cleaner energy for the AI systems that increasingly demand high-density computing resources. Altman has been backing Helion since 2014, long before ChatGPT’s OpenAI became the center of attention. This financial support highlights a growing trend among technology leaders to address infrastructure and sustainability issues, rather than solely focusing on software development.

    As AI workloads require more and more power, innovative solutions like fusion present a means to ensure the longevity of the data economy while also reducing the carbon footprint of activities ranging from training large language models (LLMs) to executing search queries.

    Ambitious Goals Ahead

    Helion’s objective is bold: to achieve net energy gain—where energy output exceeds energy input—by 2028. This would be a historic milestone in a commercial context. Although doubts persist, Microsoft’s engagement indicates a strong level of confidence. The tech giant intends to incorporate the generated power into its data center operations, consistent with its goal of being carbon-negative by 2030.

    If Helion succeeds, it could unveil a new energy source that is scalable, reliable, and specifically designed for the computationally intensive future we are moving towards.

    Sam Altman is not only focusing on AI development. Through Helion, he is making a significant wager on an energy system that could support it.

    In a field characterized by rapid growth and substantial energy consumption, fusion energy presents a unique opportunity: a sustainable long-term solution. Regardless of whether Helion meets its timeline, this development signals a shift toward concrete, physical investments that correspond with the scale of the ongoing digital transformation.

    Source:
    Link

    Tags: ,