Tech News
Back to home

Articles tagged with "ai"

Microsoft just killed Edge’s Collections and Sidebar for more Copilot, after years of pushing both features

Microsoft just killed Edge’s Collections and Sidebar for more Copilot, after years of pushing both features

Microsoft has discontinued Edge's Collections and Sidebar features with the release of Edge 149 on June 4, 2026. Collections allowed users to organize web content, while Sidebar provided quick access to mini web apps. Users are advised to export their data before upgrading to Edge 149 to avoid losing their collections. The removal of these features is part of Microsoft's shift towards focusing on the Copilot feature and AI integration in Edge.

Windows Latest•
NSA using Claude Mythos for 'offensive cyber operations,' report claims — says 'half-a-dozen' Anthropic engineers embedded inside the agency

NSA using Claude Mythos for 'offensive cyber operations,' report claims — says 'half-a-dozen' Anthropic engineers embedded inside the agency

The National Security Agency (NSA) is reportedly utilizing Anthropic's cybersecurity-focused Mythos model for "offensive cyber operations," with several Anthropic engineers embedded within the agency. This move aims to gain an advantage over adversaries who may also be using similar AI models. Despite Anthropic being banned from providing services to the Department of Defense (DOD), the engineers are assisting in customizing Mythos for specialized cyber-attacks. The situation stems from a dispute between Anthropic and the DOD over the use of AI for various purposes, leading to Anthropic being labeled a supply chain risk. The ongoing legal battle between Anthropic and the DOD adds complexity to the situation, raising questions about the future of their relationship.

Tom's Hardware•
Seattle to pass one-year AI data center moratorium next week — will use window to study community impact of AI buildouts

Seattle to pass one-year AI data center moratorium next week — will use window to study community impact of AI buildouts

Seattle is set to implement a one-year moratorium on AI data center construction to assess the impact on the community. The decision follows public support and concerns about issues like increased electricity bills and environmental impacts. The city plans to use this period to study the effects of data centers on the surrounding areas and establish regulations for their development. Residents have backed the temporary ban, highlighting worries about pollution, noise, and land use. This move contrasts with other states where politicians have faced opposition for supporting data center projects.

Tom's Hardware•
CEO Interview with Daniel Schall of Black Semiconductor

CEO Interview with Daniel Schall of Black Semiconductor

Daniel Schall, CEO of Black Semiconductor, discusses the company's focus on addressing chip-to-chip communication bottlenecks with integrated graphene photonics technology. The company's Black Semiconductor IGP™ technology aims to enable AI scaling by revolutionizing chip communication. By utilizing graphene's unique properties, Black Semiconductor's technology allows for faster and more energy-efficient data transmission between chips. The construction of FabONE, a 300mm facility in Germany, is underway to integrate electronics and photonics using graphene, with plans for pilot production in 2027 and volume production by 2029. The acquisition of Applied Nanolayers in 2025 enhances Black Semiconductor's capabilities in industrial graphene production, ensuring a robust material supply chain for their innovative technology.

SemiWiki•
Industry coalition urges Trump administration to take urgent action as AI data centers' extreme memory consumption threatens other industries — AI-driven memory chip shortage could raise prices in automotive, medical, telecommunications sectors

Industry coalition urges Trump administration to take urgent action as AI data centers' extreme memory consumption threatens other industries — AI-driven memory chip shortage could raise prices in automotive, medical, telecommunications sectors

A coalition of US trade associations is urging the Trump administration to address a memory chip shortage caused by the rapid growth of AI data centers. This shortage could lead to price increases in various industries like automotive, medical, and telecommunications. The coalition warns that the memory chip scarcity, driven by AI infrastructure expansion, could disrupt supply chains and impact the broader economy. Memory manufacturers are prioritizing high-bandwidth memory for AI accelerators, leading to shortages in commodity DRAM and NAND markets. The coalition is calling for action to mitigate these challenges and protect consumers and businesses.

Tom's Hardware•
Tom's Hardware Unfiltered: Computex 2026, Day 4 — the B2B shift, and we say farewell to Taipei

Tom's Hardware Unfiltered: Computex 2026, Day 4 — the B2B shift, and we say farewell to Taipei

The Tom's Hardware team reflects on the final day of Computex 2026, noting a shift towards more business-to-business (B2B) companies and a focus on AI and data center technology. The event showcased new products like the Y50 case, X870E Infinity Next motherboard, and Dark Power Pro 14 IO power supply. Journalists visited various booths, including Gigabyte and G.Skill, discussing industry changes and enthusiast spirit. As the event wraps up, the team prepares to leave Taipei, highlighting the fast-paced nature of the tech industry and the unique experiences gained at Computex.

Tom's Hardware•
Jensen Huang says 'every edge device will become autonomous' — Nvidia maps one computing pattern from the cloud to robotics

Jensen Huang says 'every edge device will become autonomous' — Nvidia maps one computing pattern from the cloud to robotics

Nvidia CEO Jensen Huang discusses a new computing pattern at Computex, emphasizing the move towards AI-driven systems across various devices. The company introduces the Vera data-center CPU and RTX Spark platform for Windows PCs, focusing on agent-based computing rather than traditional core-centric designs. Huang highlights the importance of single-thread speed and memory bandwidth for efficient agent operation. Nvidia's Vera CPU shows promising performance compared to competitors, and the RTX Spark platform aims to revolutionize PC design for AI applications. The company also addresses memory constraints by utilizing low-precision formats and neural texture compression techniques.

Tom's Hardware•
Indiana mayor secretly recorded saying AI data center protestors only live in 'sh***y' houses — office issues statement of clarification over controversial comments

Indiana mayor secretly recorded saying AI data center protestors only live in 'sh***y' houses — office issues statement of clarification over controversial comments

The mayor of Shelbyville, Indiana, was recorded making derogatory comments about community members protesting an AI data center project, sparking controversy. The mayor's office issued a statement clarifying his remarks, stating they were about property maintenance, not the residents' character. The proposed billion-dollar data center project in Shelbyville has faced opposition from over 2,000 people, with concerns about utility prices and water quality linked to such developments. The debates around data centers have escalated, with some politicians facing backlash and even violent incidents, raising fears of potential extremist actions or surveillance against AI data centers.

Tom's Hardware•
Microsoft quietly dropped Copilot+ PC branding for Windows 11’s powerful AI laptop, and it won’t tell you why

Microsoft quietly dropped Copilot+ PC branding for Windows 11’s powerful AI laptop, and it won’t tell you why

Microsoft has quietly abandoned the Copilot+ PC branding for its latest powerful AI laptop, the Surface Laptop Ultra, powered by NVIDIA's RTX Spark platform. Despite confirming privately that the laptop still includes an NPU, Microsoft did not mention the Copilot+ branding during the reveal. The absence of Copilot+ branding is speculated to be intentional, possibly due to NVIDIA's preference for associating local AI acceleration with its RTX branding rather than Microsoft's Copilot branding. The move reflects a shift away from the Copilot branding, which had faced controversies and negative perceptions in the past.

Windows Latest•
Engineering Documentation is a Critical Source of Truth – Do You Know if it’s Accurate?

Engineering Documentation is a Critical Source of Truth – Do You Know if it’s Accurate?

Engineering documentation is highlighted as a crucial source of truth in embedded systems development, with the risk of failure due to inaccuracies in documentation alignment. Hal Conklin, llmda's chief commercial officer, will present a webinar on this topic, emphasizing the importance of accurate documentation and addressing challenges with generic AI approaches. The webinar will cover topics such as deterministic correctness, source traceability, and collaborative governance in engineering artifacts. llmda's technology stack, including the Spectra™ tool, aims to improve documentation accuracy and system design quality. The webinar, scheduled for June 16, 2026, will delve into these critical aspects of system design success.

SemiWiki•
MIT Technology Review

The Download: AI-generated lawsuits and virtual power plants for data centers

AI-generated lawsuits are flooding courts, raising questions about the rights and responsibilities of chatbots in legal matters. Meanwhile, Google is investing in virtual power plants to manage energy for data centers, offering incentives for customers to reduce electricity use during peak times. Additionally, the EU is proposing legislation to reduce dependence on Big Tech and boost domestic tech sectors.

MIT Technology Review•
Best of Computex 2026: Innovating despite disruptions

Best of Computex 2026: Innovating despite disruptions

At Computex 2026, tech companies showcased innovative products despite disruptions in the industry. Intel introduced the Arc G3 range for handhelds, featuring purpose-built SoCs for gaming. Nvidia unveiled the RTX Spark Superchip, aiming to redefine personal computing with AI capabilities. Noctua presented the NL-LC1 liquid cooler designed for quiet computing. Dell's XPS 13 emerged as an affordable laptop rivaling Apple's MacBook Neo. Additionally, Phison introduced the E37T SSD controller to reduce SSD prices, and Asus showcased the first OLED eSports monitor and a Wi-Fi 8 router.

Tom's Hardware•
‘Bots have now passed human traffic online,’ Cloudflare boss laments — says agentic traffic wasn’t expected to eclipse real people until next year

‘Bots have now passed human traffic online,’ Cloudflare boss laments — says agentic traffic wasn’t expected to eclipse real people until next year

Cloudflare's CEO Matthew Prince acknowledges that AI agent internet traffic has surpassed human traffic online, contrary to his previous predictions. This surge in agentic traffic involves AI agents browsing the web like humans for various tasks. Cloudflare's data shows a significant shift towards bot traffic, with bots accounting for 57.5% of HTTP requests. However, human users still dominate in terms of time spent on apps and streaming services. Cloudflare's breakdown of human vs. bot traffic by country reveals high bot activity in places like Gibraltar, Singapore, and Iran, with the latter attributed to VPN usage for automated scraping.

Tom's Hardware•
Microsoft CEO says new AI data centers use as little water annually as a restaurant — closed-loop cooling system aims to slash consumption from millions of gallons as AI infrastructure faces mounting environmental scrutiny

Microsoft CEO says new AI data centers use as little water annually as a restaurant — closed-loop cooling system aims to slash consumption from millions of gallons as AI infrastructure faces mounting environmental scrutiny

Microsoft CEO Satya Nadella announced at Microsoft Build 2026 that the company's new AI data centers consume as little water annually as a single restaurant, thanks to a closed-loop cooling system. This move is part of Microsoft's "Community-First AI Infrastructure" strategy to address concerns about data center environmental impact. The new liquid-cooling architecture reduces water consumption significantly, with over 90% of cooling relying on a closed-loop system. While the system is currently implemented at the Fairwater AI data center in Wisconsin, it will become the standard for future AI-focused data centers as Microsoft aims to be water positive by 2030. This initiative comes amidst growing pressure on tech giants like Microsoft and Google to reduce the environmental footprint of their data centers.

Tom's Hardware•
7 Ways New Engineers Can Flourish in the Age of AI

7 Ways New Engineers Can Flourish in the Age of AI

New engineers can thrive in the era of AI by viewing AI tools as complements rather than competition. Lokesh Lagudu, an IEEE senior member and senior engineering manager at Walmart Global Tech, shares seven key strategies for success. These strategies include embracing continuous learning, developing soft skills, seeking mentorship, and staying adaptable. Lagudu emphasizes the importance of understanding AI's capabilities and limitations, as well as fostering a growth mindset. By following these tips, new engineers can navigate the evolving tech landscape and excel in their careers.

IEEE Spectrum•
MSI Claw 8 EX AI+ brings Intel Arc G3 Extreme to handhelds — 8-inch, 120 Hz display and new ergonomic grips

MSI Claw 8 EX AI+ brings Intel Arc G3 Extreme to handhelds — 8-inch, 120 Hz display and new ergonomic grips

MSI unveiled the Claw 8 EX AI+ handheld at Computex, featuring Intel Arc G3 Extreme graphics and ergonomic grips. The device boasts an 8-inch, 120 Hz display with variable refresh rate, hall-effect triggers, sticks, and a responsive D-Pad. With up to 32GB of RAM and an 80 WHr battery, the Claw aims to offer a comfortable gaming experience. Pricing and release details are yet to be officially confirmed, but early indications suggest a version with 32GB of RAM and 1TB of storage may be priced at $1,699.99.

Tom's Hardware•
TSMC Expands Use of NVIDIA AI Technologies Across Chip Production Operations

TSMC Expands Use of NVIDIA AI Technologies Across Chip Production Operations

TSMC is expanding its use of NVIDIA AI and accelerated computing technologies in chip design and manufacturing operations, including lithography, process simulation, defect inspection, production scheduling, and factory optimization. By leveraging NVIDIA's CUDA-X libraries and GPU-accelerated computing platforms, TSMC has seen improvements in cycle time and cost effectiveness, particularly in computational lithography. The collaboration aims to address the complexities of modern semiconductor manufacturing and advance towards the angstrom era. AI integration in areas like transistor simulation, factory operations optimization, quality control, and digital twins for manufacturing reflects a broader industry trend towards AI-driven manufacturing for improved yield, energy efficiency, and design cycles. This partnership signifies a significant step towards autonomous, data-driven chip manufacturing and a competitive edge for leading foundries.

SemiWiki•
Microsoft unveils Project Solara AI, a chip-to-cloud platform built to power a new generation of 'agent-first' enterprise devices — hardware designed to run AI agents instead of traditional apps

Microsoft unveils Project Solara AI, a chip-to-cloud platform built to power a new generation of 'agent-first' enterprise devices — hardware designed to run AI agents instead of traditional apps

Microsoft has introduced Project Solara, a chip-to-cloud platform for "agent-first" enterprise devices that run AI agents instead of traditional apps. The platform includes the Microsoft Device Ecosystem Platform (MDEP), built on the Android Open Source Project (AOSP), paired with Azure-hosted agent services. Microsoft has partnered with Qualcomm and MediaTek for hardware, releasing reference designs for OEMs. Solara features just-in-time UI for adaptive interfaces and aims to provide consistency across devices. The platform targets enterprise sectors like retail, healthcare, and field services, with pilot programs already in place with major companies.

Tom's Hardware•
MIT Technology Review

The Download: Trump’s new AI order, and smart glasses for warfare

President Trump signed a new AI order focusing on innovation and security, introducing a voluntary review system for tech companies to share models with the government before release. Meanwhile, defense-tech company Anduril is developing smart glasses for warfare with Meta, envisioning drone strike orders via eye-tracking and voice commands. SpaceX plans to raise $75 billion in an IPO, Meta scales back worker tracking for AI training, and Microsoft launches a new AI assistant. Additionally, concerns arise about AI's impact on mathematics, surveillance, and cybersecurity.

MIT Technology Review•
32GB of DDR5 now costs $375 minimum — AI shortage continues to squeeze PC building

32GB of DDR5 now costs $375 minimum — AI shortage continues to squeeze PC building

The ongoing AI shortage is driving up the prices of PC components, with 32GB of DDR5 RAM now costing a minimum of $375. This marks a significant increase from just a year ago when similar kits were priced at less than $100. The shortage is causing retailers to inflate prices, making it challenging for enthusiasts to build or upgrade gaming PCs. The price pressure is expected to persist, with popular RAM kits from brands like Corsair and Crucial exceeding $400. Additionally, AMD and Intel are making efforts to mitigate rising memory prices by reintroducing legacy products and offering more options on older memory technologies.

Tom's Hardware•
Github Copilot customers report up to 100-fold price hikes — AI sticker shock bites as Microsoft switches to usage-based pricing

Github Copilot customers report up to 100-fold price hikes — AI sticker shock bites as Microsoft switches to usage-based pricing

GitHub Copilot has transitioned to usage-based pricing, leading to significant price hikes for customers. Users are reporting bills increasing by up to 100-fold, with some finding subscription plans limited or useless due to low usage limits. The new pricing model includes token allowances in different subscription tiers, but estimating token usage for queries can be challenging. Users are advised to carefully select efficient models to manage costs, as some have experienced unexpectedly high token consumption even with light usage. Investors are pushing for AI companies to generate more revenue to match their expenditures, despite backlash against Microsoft for the pricing changes.

Tom's Hardware•
Microsoft pledges to make Windows 11 the OS for building AI, after years of Copilot backlash

Microsoft pledges to make Windows 11 the OS for building AI, after years of Copilot backlash

Microsoft is focusing on making Windows 11 the go-to operating system for AI development, aiming to unify fragmented AI development workflows and provide a consistent experience for developers. The company plans to support multiple AI models and agents, offering flexibility and choice for developers. Windows 11 will feature new local AI models, APIs, and integration with powerful local AI hardware. Microsoft is also emphasizing security and governance in AI development, positioning Windows 11 as a trusted platform for enterprise AI infrastructure. The company is working on repairing Windows' reputation with developers by improving responsiveness, reducing memory usage, and optimizing the user experience.

Windows Latest•
Using Graph Attention for Virtual Metrology in Semiconductor Manufacturing (Intel Foundry, ASU)

Using Graph Attention for Virtual Metrology in Semiconductor Manufacturing (Intel Foundry, ASU)

Researchers from Arizona State University and Intel Foundry have collaborated on a paper titled "Graph Attention-Based Virtual Metrology for Film Deposition Processes in Semiconductor Manufacturing." The study focuses on the use of artificial intelligence-driven virtual metrology to predict wafer-level characteristics in semiconductor manufacturing. The proposed framework utilizes graph attention mechanisms to model structured dependencies among process variables, improving predictive performance compared to existing models. The approach integrates temporal feature learning and parameter-layer dependency modeling to enhance prediction accuracy and provide meaningful insights into process dynamics, supporting effective monitoring and optimization in semiconductor manufacturing.

SemiEngineering•
Why Europe Needs Its Own AI Supercomputing Platform

Why Europe Needs Its Own AI Supercomputing Platform

Semidynamics and SiPearl are collaborating to develop a European sovereign rack-scale AI compute platform, aiming to reduce Europe's dependence on non-European AI technology providers. This partnership focuses on creating an "EU-sovereign" AI infrastructure to enhance Europe's control over its AI systems. By utilizing RISC-V technology and combining strengths in processor design and semiconductor innovation, the collaboration aims to optimize AI compute platforms for European requirements. This initiative not only boosts Europe's semiconductor ecosystem but also addresses concerns regarding data governance, security, economic competitiveness, and technological independence in the AI landscape. Ultimately, the partnership signifies a strategic move towards European AI independence and digital sovereignty.

SemiWiki•
Learn How llmda Uses Agentic AI to Generate Hardware Docs & Keep Them Consistent

Learn How llmda Uses Agentic AI to Generate Hardware Docs & Keep Them Consistent

llmda.ai utilizes agentic AI to streamline the generation and maintenance of technical documentation for embedded system design projects. The company's product, llmda Spectra, leverages advanced LLM technology to extract, synthesize, and structure content from internal engineering artifacts, improving documentation accuracy and consistency. The white paper issued by llmda covers topics such as intelligent information extraction, collaboration readiness, and automated block diagram generation. Case studies presented in the white paper demonstrate how llmda Spectra accelerates documentation cycles, enhances quality, and reduces engineering burden. This solution aims to transform documentation from an afterthought into a trackable and improvable output for embedded systems development projects.

SemiWiki•
MIT Technology Review

The Download: AI can run your admin department now

AI is increasingly taking on basic administrative tasks for small businesses, from organizing notes to invoicing and social media planning. Anthropic has confidentially filed for an IPO ahead of OpenAI, aiming to go public this fall. The EU may exclude US cloud giants like Amazon, Microsoft, and Google from critical contracts to reduce dependence on US tech. Florida has become the first state to sue OpenAI over alleged child safety risks related to ChatGPT. Hackers exploited Meta AI to steal Instagram accounts, highlighting the risks of offloading support to AI.

MIT Technology Review•
NVIDIA reveals if RTX Spark (N1X) can run all your Windows 11 apps

NVIDIA reveals if RTX Spark (N1X) can run all your Windows 11 apps

NVIDIA introduces RTX Spark (N1X), an Arm-based processor designed to revolutionize Windows PCs for AI applications. The processor, with up to 128GB of memory, is optimized for AI agents and NVIDIA's graphics stack. While NVIDIA claims it can run all Windows 11 apps, technical benchmarks are yet to be shared. The Surface Laptop Ultra is among the first devices to feature this chip. Windows on Arm has made strides in app compatibility, with popular apps running natively and improved emulation technology like Prism.

Windows Latest•
Anthropic files to go public — Claude maker races OpenAI and SpaceX to IPO

Anthropic files to go public — Claude maker races OpenAI and SpaceX to IPO

Anthropic, the company behind Claude language models, has confidentially filed for an IPO, joining SpaceX and OpenAI in the race to go public. The company has not disclosed much information yet, such as price targets or the number of shares to be sold. Going public could allow Anthropic's employees to sell equity, potentially benefiting AI researchers and other staff. The company has shifted focus to enterprise users with features like Claude Code and Mythos model, used for security purposes. Anthropic recently surpassed OpenAI in valuation and has emphasized caution with its technologies, notably in a dispute with the Trump Administration.

Tom's Hardware•
A Look at the High-Profile Speakers Presenting at #DAC2026

A Look at the High-Profile Speakers Presenting at #DAC2026

The article discusses the high-profile speakers presenting at #DAC2026, a prestigious technical conference with a long history. Keynote speakers include John Martinis, Baaziz Achour, and Jan Rabaey, covering topics like quantum computing, semiconductor technology, and research. SKYTalks feature industry leaders like Timothy Costa, Lalitha Immaneni, and Huiming Bu discussing advancements in EDA, AI, and semiconductor innovation. TechTalks include speakers like Amit Gupta, Jeffrey Pan, and William Wang sharing insights on current trends in the field. Analyst reviews by Jay Vleeschhouwer and Dylan Patel aim to highlight macro trends in the industry. The conference will be held in July 2026 at the Long Beach Convention Center, offering free access passes through the I Love DAC program until July 12, 2026.

SemiWiki•
Computex 2026 Day Zero Wrap-Up: Nvidia launches RTX Spark Superchip assault on laptop and desktop markets, Intel readies Xeon 6+

Computex 2026 Day Zero Wrap-Up: Nvidia launches RTX Spark Superchip assault on laptop and desktop markets, Intel readies Xeon 6+

Nvidia made a significant impact at Computex 2026 with the launch of the RTX Spark Superchip, a powerful Arm platform aimed at transforming Windows 11 into an AI operating system. Microsoft has already adopted the chip in its Surface Laptop Ultra, featuring a high-resolution Mini LED display and a variety of ports. Intel introduced the Xeon 6+ processor to compete with AMD in the data center market, boasting impressive core counts and performance improvements. Additionally, AMD unveiled new Ryzen processors and confirmed support for the AM5 socket until at least 2029. Alienware and Asus also showcased new gaming monitors, including OLED options with high refresh rates.

Tom's Hardware•
DLSS 4.5 Ray Reconstruction update arrives in August for better ray tracing visuals — broader training data set and second-gen transformer architecture combine for improved image quality

DLSS 4.5 Ray Reconstruction update arrives in August for better ray tracing visuals — broader training data set and second-gen transformer architecture combine for improved image quality

Nvidia is set to release DLSS 4.5 Ray Reconstruction in August, enhancing ray-traced visuals with a broader training data set and a second-gen transformer architecture for improved image quality. The update builds on previous DLSS advancements, offering better spatial awareness and processing for more accurate lighting and stable images. DLSS 4.5 Ray Reconstruction will be compatible with all GeForce RTX GPUs and is expected to enhance image quality in 27 games, including popular titles like Cyberpunk 2077 and Resident Evil Requiem. Additionally, Nvidia plans to bring this technology to Blender, promising a more interactive viewport experience and near-final image quality for creators.

Tom's Hardware•
Microsoft Surface Laptop Ultra weilds Nvidia's RTX Spark superchip with 128GB of RAM, 20 Arm CPU cores, and a Blackwell GPU — 15-inch mini-LED PixelSense Ultra display rounds out the powerful package

Microsoft Surface Laptop Ultra weilds Nvidia's RTX Spark superchip with 128GB of RAM, 20 Arm CPU cores, and a Blackwell GPU — 15-inch mini-LED PixelSense Ultra display rounds out the powerful package

Microsoft has unveiled the Surface Laptop Ultra, featuring Nvidia's RTX Spark Superchip, 128GB of RAM, 20 Arm CPU cores, and a Blackwell GPU. The laptop boasts a 15-inch mini-LED PixelSense Ultra display with high resolution and peak HDR brightness. With a variety of ports and weighing under 4.5 pounds, it targets AI developers, creators, and power users. The device is set to launch later this year, offering a compelling Windows alternative to the MacBook Pro for those seeking powerful performance in a portable package.

Tom's Hardware•
Nvidia unveils RTX Spark Superchip for laptops and desktop PCs at Computex 2026 – new platform promises to turn Windows into an agentic AI OS with Arm CPU, Blackwell GPU, and 128GB unified memory

Nvidia unveils RTX Spark Superchip for laptops and desktop PCs at Computex 2026 – new platform promises to turn Windows into an agentic AI OS with Arm CPU, Blackwell GPU, and 128GB unified memory

Nvidia introduced the RTX Spark Superchip at Computex 2026, aiming to revolutionize Windows into an agentic AI platform for laptops and desktops. The platform features an Arm CPU, Blackwell GPU, and 128GB unified memory, promising efficiency and power for AI interactions. The RTX Spark Superchip offers up to 20 Arm CPU cores, a Blackwell GPU with 6,144 CUDA cores, and 128GB of LPDDR5X RAM, supporting AI agents and 120-billion-parameter models. Nvidia collaborates with partners like Dell, HP, Lenovo, Asus, MSI, and Microsoft to deliver high-performance laptops with OLED displays, extended battery life, and premium design. The RTX Spark platform also enhances creative and gaming capabilities, with promises of 100 FPS 1440p gaming and Adobe Photoshop and Premiere optimizations for GPU acceleration and AI workflows.

Tom's Hardware•
Intel details long-awaited Crescent Island AI GPU at Computex, boasts up to 480 GB of LPDDR5X to combat memory shortages — company shares more details of its Xe3P inference accelerator at Computex

Intel details long-awaited Crescent Island AI GPU at Computex, boasts up to 480 GB of LPDDR5X to combat memory shortages — company shares more details of its Xe3P inference accelerator at Computex

At Computex 2026, Intel unveiled details about its upcoming Data Center GPU, Crescent Island, based on the Xe3P GPU architecture. The GPU will utilize LPDDR5X memory, offering up to 480GB of memory capacity. This design choice aims to maximize memory capacity while maintaining high bandwidth, potentially making Crescent Island more efficient for AI inference tasks. The air-cooled card with modest power requirements is expected to fit into traditional GPU servers, appealing to companies seeking on-premise inferencing solutions. Intel plans to launch Crescent Island in the second half of 2026, with more information expected as the year progresses.

Tom's Hardware•
Watch Nvidia's Computex 2026 keynote here — Jensen Huang takes the stage for Computex and GTC Taipei at 8pm PT / 11pm ET on May 31

Watch Nvidia's Computex 2026 keynote here — Jensen Huang takes the stage for Computex and GTC Taipei at 8pm PT / 11pm ET on May 31

Nvidia CEO Jensen Huang will kick off Computex 2026 with a keynote address on May 31 at 8pm PT / 11pm ET, focusing on AI breakthroughs. The event will likely cover enterprise AI and potentially reveal consumer announcements, such as the N1X chip. The keynote is expected to run for about two hours, with a possibility of consumer-related updates but unlikely to include GeForce news. Nvidia is also hosting GTC Live with pregame events leading up to the keynote.

Tom's Hardware•
John Barr: The EDA Veteran and Award-Winning Needham Funds Portfolio Manager

John Barr: The EDA Veteran and Award-Winning Needham Funds Portfolio Manager

John Barr, Portfolio Manager of the Needham Aggressive Growth Fund, has a background in the EDA industry before transitioning to Wall Street. His experiences in EDA provided a unique understanding of technology and business, shaping his investment approach. Barr's career includes stints as a sell-side analyst and buy-side portfolio manager, with notable successes in taking EDA companies public during the tech bubble. He currently manages Needham's mutual funds and maintains PDF Solutions, an EDA company, in his portfolio. Barr continues to monitor the evolving EDA landscape, particularly the impact of agentic AI startups on chip design and verification.

SemiWiki•
SoftBank to spend up to $87 billion on French AI data centers — country offers ample nuclear grid that US sites lack

SoftBank to spend up to $87 billion on French AI data centers — country offers ample nuclear grid that US sites lack

SoftBank Group is set to invest up to $87 billion in building AI data centers in France, with the first phase costing $52 billion and delivering 3.1GW by 2031. The partnership includes EDF and Schneider Electric, leveraging France's nuclear power grid for cost-effective energy supply. This move contrasts with SoftBank's U.S. plans, where it needs to build its own energy infrastructure for data centers. The French investment showcases the country's ability to host large-scale digital infrastructure on low-carbon electricity. The $87 billion commitment includes a $45 billion first phase, with additional sites planned for the future.

Tom's Hardware•
This DIY Bipedal Robot Used Pneumatic “Air-Muscles” Instead of Motors

This DIY Bipedal Robot Used Pneumatic “Air-Muscles” Instead of Motors

A headless, armless bipedal robot called the Shadow Walker was created by a hobbyist group in the late 1980s, using pneumatic "air-muscles" instead of traditional motors. The creators of the Shadow Walker later went on to establish a pioneering robotics company. This DIY project showcased an innovative approach to robotics and highlighted the potential of using pneumatic technology in creating humanoid robots.

IEEE Spectrum•
New AI-compute cryptocurrency Pearl sparks a GPU mining rush but profitability is already sliding — RTX 5090 daily revenue has halved to $17.19 since April

New AI-compute cryptocurrency Pearl sparks a GPU mining rush but profitability is already sliding — RTX 5090 daily revenue has halved to $17.19 since April

A new cryptocurrency named Pearl, utilizing AI workloads for blockchain security, has led to a surge in GPU mining activity, with an RTX 5090's daily revenue dropping to $17.19 from $33.80 since April. The coin's mainnet launch in April 2026 attracted attention after an exclusive partnership announcement with AI cloud provider Together AI. Despite the initial hype, PRL's price and mining profitability are declining as more miners join the network. The consensus mechanism, Proof-of-Useful-Work, involves large-scale matrix multiplication, a key component of AI training and inference. Mining is limited to Nvidia hardware, and most activity has been on rented cloud capacity, impacting per-card payouts as network difficulty rises.

Tom's Hardware•
The 'ultimate mosquito killer' uses lasers and AI — custom model trained to detect and lock lasers on these pests

The 'ultimate mosquito killer' uses lasers and AI — custom model trained to detect and lock lasers on these pests

A computer vision and robotics expert has developed an advanced mosquito killer that uses lasers and AI. Steven Cheng's project involves training a custom model to detect and target mosquitoes with precision using deep learning technology. The device, equipped with a laser cannon, is mounted on a high-precision gimbal for accurate targeting. Cheng successfully eliminated mosquitoes in his residence with this innovative solution, which outshines previous laser-based mosquito killers in terms of detection capabilities. Another similar product, the Photonmatrix, is set to ship in June 2026, offering a different approach to mosquito eradication.

Tom's Hardware•
A robot startup is wreaking havoc on short-term rentals in San Francisco — Airbnb hosts allege 'guests' secretly tested robots indoors, left the units completely trashed

A robot startup is wreaking havoc on short-term rentals in San Francisco — Airbnb hosts allege 'guests' secretly tested robots indoors, left the units completely trashed

A San Francisco startup called "The Bot Company" is allegedly using Airbnb rentals as secret testing grounds for robots, leaving properties damaged and trashed. Hosts like Sean Donovan have filed lawsuits against the startup for damages and lost income. The startup, led by Kyle Vogt, aims to develop household robots but operates under secrecy, using Airbnb rentals without informing hosts of its activities. Multiple hosts have reported similar experiences of their properties being used for robot testing without consent, leading to legal action against The Bot Company.

Tom's Hardware•
Kevin O'Leary claims Chinese propaganda is to blame for anti-datacenter backlash, 'hundreds of millions of dollars' being spent to kill US dominance in AI — industry proponents and Trump administration reinforce claims of foreign interference

Kevin O'Leary claims Chinese propaganda is to blame for anti-datacenter backlash, 'hundreds of millions of dollars' being spent to kill US dominance in AI — industry proponents and Trump administration reinforce claims of foreign interference

Kevin O'Leary, along with industry proponents and the Trump administration, claims that Chinese propaganda is fueling the backlash against data centers, with allegations of 'hundreds of millions of dollars' being spent to undermine US dominance in AI. O'Leary asserts that opposition to data centers, like his $100-billion project in Utah, is orchestrated by foreign interference, particularly from China. While some US officials echo these sentiments, no concrete evidence has been provided to support these claims. The debate around data centers also touches on concerns about power consumption, water usage, and job displacement due to automation. Despite differing opinions on the extent of Chinese influence, the overall narrative underscores the ongoing AI arms race between the US and China.

Tom's Hardware•
TikTok owner ByteDance is reportedly developing its own custom AI CPUs — company looks to ease China's dependence on US chipmakers

TikTok owner ByteDance is reportedly developing its own custom AI CPUs — company looks to ease China's dependence on US chipmakers

ByteDance, the owner of TikTok, is developing custom AI CPUs to reduce reliance on US chipmakers. The new chip is inspired by Groq's "language processing units" for AI inference tasks. ByteDance is reportedly evaluating Arm and RISC-V designs and partnering with InnoStar Semiconductor for memory technology. The company's previous investments in technology include designing its SeedChip AI accelerator with TSMC. ByteDance aims to reduce dependence on Nvidia and potentially increase self-sufficiency in chip technology.

Tom's Hardware•
New Silicon Motion SM2524XT chip brings 14 GB/s to mainstream SSDs — 6nm DRAMless controller boasts heavy AI PC optimization and slashes KV cache latency

New Silicon Motion SM2524XT chip brings 14 GB/s to mainstream SSDs — 6nm DRAMless controller boasts heavy AI PC optimization and slashes KV cache latency

Silicon Motion's new SM2524XT chip for mainstream SSDs brings read speeds of up to 14 GB/s with four NAND channels supporting transfer rates of 4,800 MT/s. The controller, based on Arm Cortex-R-series cores, does not use DRAM and complies with the NVM 2.0 specification. It supports the latest 3D NAND technologies and features SMI's NANDXtend LDPC ECC technology. The chip boosts random workload throughput by 25%, reduces latency, and improves responsiveness for AI and KV cache operations. Prototype drives based on the SM2524XT will be showcased at Computex, with market availability expected next year.

Tom's Hardware•
Fed up with vibe coders, dev sneaks data-nuking prompt injection into their code

Fed up with vibe coders, dev sneaks data-nuking prompt injection into their code

A developer added hidden instructions to a Java testing app, jqwik, to sabotage projects performed by AI coding agents, sparking controversy over "vibe coding." The update included a prompt injection that could potentially delete all jqwik tests and code without warning or opt-out. The addition exploited an AI attack that could lead to the deletion of work product by vulnerable coding agents. While some developers didn't object to excluding their apps from AI use, concerns were raised about the ethics and potential destructive impact of the hidden instructions. The aggressive nature of the probe and its potential consequences on human operators downstream were highlighted.

Ars Technica•
Millions of AI agents imperiled by critical vulnerability in open source package

Millions of AI agents imperiled by critical vulnerability in open source package

A critical vulnerability in the open source framework Starlette is imperiling millions of AI agents and tools worldwide, potentially allowing hackers to breach servers and access sensitive data and credentials. This vulnerability, known as BadHost (CVE-2026-48710), affects systems using Starlette versions prior to 1.0.1 and is easy to exploit on servers not behind a properly configured firewall. The flaw impacts various widely used packages in the Python AI ecosystem, including FastAPI, vLLM, and LiteLLM. Security researchers have highlighted the seriousness of this vulnerability, with a severity rating of 7 out of 10 and warnings that it poses a significant threat to users of apps relying on Starlette.

Ars Technica•
MIT Technology Review

Here’s why Elon Musk lost his suit against OpenAI

Elon Musk lost his lawsuit against OpenAI due to a technicality - the jury found that he sued the AI giant too late, and his claims were barred by statutes of limitations. Musk accused OpenAI of breaching a charitable trust he created through his donations and unjustly enriching themselves. The lawsuit focused on OpenAI's transition from a nonprofit to a for-profit entity, which Musk claimed went against the mission he supported. Despite Musk's claims, the jury sided with OpenAI, prompting Musk to announce plans to appeal the decision.

MIT Technology Review•
Jury throws out Elon Musk's lawsuit against OpenAI after less than two hours of deliberation — Unanimous vote that Musk filed the lawsuit too late

Jury throws out Elon Musk's lawsuit against OpenAI after less than two hours of deliberation — Unanimous vote that Musk filed the lawsuit too late

A federal jury in Oakland, California, swiftly dismissed Elon Musk's lawsuit against OpenAI, CEO Sam Altman, co-founder Greg Brockman, and Microsoft, citing that Musk filed the lawsuit too late, with all claims barred by the statute of limitations. Musk had sought significant damages and changes within OpenAI, but the jury's decision on timeliness prevented a deeper examination of the case. Musk testified that he delayed suing due to assurances from Altman, only acting after Microsoft's investment in OpenAI's for-profit arm. OpenAI's attorneys argued that Musk was aware of the transition to a for-profit entity since 2017. The verdict removes a major legal threat to OpenAI's restructuring and IPO plans, also clearing Microsoft of aiding the breach allegations.

Tom's Hardware•
China says 'world's first' offshore wind-powered underwater data center has entered full operation, houses 2,000 servers — 24 megawatt subsea AI facility uses ocean water for passive cooling and offshore wind for power

China says 'world's first' offshore wind-powered underwater data center has entered full operation, houses 2,000 servers — 24 megawatt subsea AI facility uses ocean water for passive cooling and offshore wind for power

China's "world's first" offshore wind-powered underwater data center, located off the coast of Shanghai, has commenced full operations. This $226 million facility, managed in partnership with HiCloud Technology and state-backed telecom providers, houses 2,000 servers and utilizes ocean water for passive cooling and offshore wind for power. The data center is designed to process AI, big data annotation, and 5G infrastructure workloads, boasting an energy-efficient Power Usage Effectiveness (PUE) below 1.15. This project aligns with China's efforts to integrate renewable energy into digital infrastructure and addresses the growing electricity consumption demands of AI expansion. Despite the benefits, underwater data centers pose engineering challenges related to maintenance and reliability.

Tom's Hardware•
Linus Torvalds says flood of duplicate AI-generated vulnerability reports have made Linux security mailing list 'almost entirely unmanageable' — private list 'a waste of time for everybody involved' in switch to new public system

Linus Torvalds says flood of duplicate AI-generated vulnerability reports have made Linux security mailing list 'almost entirely unmanageable' — private list 'a waste of time for everybody involved' in switch to new public system

Linus Torvalds expressed frustration with the Linux kernel's private security mailing list being overwhelmed by duplicate vulnerability reports generated by AI tools. He emphasized the need for a new public system for handling AI-detected bugs, directing researchers to submit findings directly to maintainers as public disclosures. The volume of duplicate reports has strained the existing triage process, prompting Torvalds to urge researchers to provide more value by creating patches along with bug reports. The Linux kernel project recently established a policy allowing AI-generated code with strict disclosure requirements, ensuring human accountability for all AI-assisted contributions.

Tom's Hardware•

We use cookies

We use cookies to ensure you get the best experience on our website. For more information on how we use cookies, please see our cookie policy.