For you Ai Security Dev Cloud Hardware Startups Releases General

From Tom's Hardware · 40 stories

1 source 1 report 1d ago Updated 1d ago

John Deere Skid-Steer Loader Performs Acrobatic Maneuvers Using AI Reinforcement Learning

Researchers from the U.S. and Switzerland have demonstrated a 5-ton John Deere skid-steer loader performing complex acrobatic maneuvers autonomously, trained using reinforcement learning in a simulation. This research aims to advance automation in heavy machinery industries like construction and mining, showcasing effective sim-to-real transfer for complex physical tasks.

ai robotics automation heavy machinery
3 sources 3 reports 52d ago

Xbox's Disc-to-Digital Feature Result of 15-Year Development and Disc Identification

Xbox recently announced a disc-to-digital feature, allowing users to convert physical Xbox One game discs into digital copies. This capability stems from a decision made 15 years ago by Microsoft to uniquely identify discs, enabling game preservation and potential future digital ownership models.

releases xbox gaming microsoft digitalization
3 sources 4 reports 82d ago

Local Utility Hikes and Noise Drive US Data Center Protests, Not Anti-AI Sentiment

Research indicates that public opposition to data center projects in the US is primarily driven by local environmental and economic concerns, such as increased utility costs and noise pollution, rather than broader anti-AI sentiment or political affiliations. This highlights the direct impact of data center operations on local communities, leading to widespread protests and calls for policy changes regarding infrastructure costs.

cloud data centers utilities infrastructure local impact
3 sources 3 reports 92d ago

Micron's $50 Billion Boise Expansion Transforms Local Economy and Demographics

Micron's $50 billion investment in new chip manufacturing facilities in Boise, Idaho, is driving significant economic and demographic changes in the city. This expansion, fueled by the AI boom, is creating thousands of jobs and increasing wealth for Micron employees, but also leading to rising housing costs and traffic congestion for residents.

hardware semiconductors manufacturing economic development urban planning
1 source 1 report 2d ago

Intel initiates new round of layoffs in global IT department

Intel has begun another round of layoffs within its global IT organization, following previous workforce reductions throughout the year. While Intel has not disclosed specific numbers, estimates suggest a double-digit percentage of the IT department's 2,607 employees could be affected. These cuts are part of ongoing efforts to reduce expenses and simplify operations.

general intel layoffs it workforce
1 source 1 report 2d ago

Intel 18A Process Achieves 70% Yields; Microsoft Adopts EMIB Packaging Technology

Intel announced that its 18A fabrication process has reached 70% yield rates, aligning with expectations for a new manufacturing technology ramp. Additionally, Intel signed a multi-year agreement with Microsoft for Xeon processor supply and the adoption of Intel's EMIB advanced packaging technology for Microsoft's in-house silicon.

hardware intel microsoft semiconductors manufacturing
1 source 1 report 2d ago

Silicon Motion Demos SM8466 PCIe 6.0 SSD Controller, Samples Imminent

Silicon Motion demonstrated its SM8466 PCIe 6.0 SSD controller for data center applications at the SC26 conference. This controller features a Separate Command Architecture (SCA) and 2nd Generation PerformaShape technology to improve consistent performance and predictable quality of service under concurrent workloads. Drives using the SM8466 are expected to ship in late 2027 or early 2028.

hardware ssd pcie datacenter storage
1 source 1 report 2d ago

Intel Diamond Rapids CPUs appear in live server demo and server display at SC26

Intel's next-generation Diamond Rapids Xeon CPUs were observed at Supercomputing 2026 (SC26) in a live server demonstration and a server display. This marks the first public appearance of a running Diamond Rapids chip, despite Intel not showcasing it directly.

hardware intel xeon diamond rapids cxl
7 sources 8 reports 93d ago

Comcast Activates Wi-Fi Motion Sensing on Xfinity Routers for Home Activity Monitoring

Comcast has launched Xfinity Shield, a new home security platform that includes Wi-Fi Motion, a feature enabling compatible Xfinity routers to detect movement by analyzing disruptions in Wi-Fi signals. This update, available through the Xfinity Internet app, allows customers to monitor home activity without additional hardware, providing an added layer of awareness for Xfinity Internet users.

releases wifi iot home security comcast
1 source 3 reports 9h ago

Modder Adapts 4-kW Air Conditioner Radiator for PC Cooling Benchmarks

A hardware modder repurposed a 4-kW outdoor air conditioner radiator unit into a PC water-cooling system to test its performance with an RTX 5060, RTX 3080, and Intel Core i9 14900K. The custom setup significantly reduced GPU temperatures, but overclocking yielded minimal FPS gains, demonstrating current GPU power/performance curves.

hardware pc cooling modding overclocking pc modding
1 source 1 report 9h ago Updated 9h ago

SNES Runs ChatGPT with Super FX 2 Chip Acceleration

A 35-year-old Super Nintendo Entertainment System (SNES) has been configured to run a 2.2-million-parameter chatbot, a distilled version of OpenAI's GPT-OSS-120b. The project utilizes the SNES's Super FX 2 chip, a 21 MHz RISC coprocessor, to accelerate the chatbot's responsiveness, demonstrating AI capabilities on vintage hardware.

ai retrocomputing hardware llm
1 source 2 reports 9h ago Updated 9h ago

Homelabber Runs 2.8-Trillion-Parameter AI Model on CPU with Discontinued Optane Memory

A Reddit user successfully ran Moonshot AI's 2.8-trillion-parameter Kimi K3 model on a 40-core Xeon CPU using second-gen Intel Optane PMem 200 memory, without a GPU. This project demonstrates a method for running large AI models on consumer-grade hardware by utilizing high-capacity, secondhand persistent memory to overcome RAM limitations, despite slow inference speeds.

ai homelab optane cpu gpu
1 source 1 report 9h ago Updated 9h ago

Hardware Hacker Creates Open-Source Portable Wi-Fi NAS for 3.5-inch Floppy Disks

A hardware hacker named BigCrimping developed floppy3.5, an open-source portable Wi-Fi NAS designed to read and copy 3.5-inch floppy disks over a network. This device allows users to archive data from old floppies, including those used by vintage digital cameras, without needing a direct computer connection.

hardware open-source storage retrocomputing
1 source 1 report 2d ago

JEDEC Previews DDR6 RAM Specification: 44% Shorter Modules, 4X DDR5 Baseline Bandwidth

JEDEC released a preview of the DDR6 memory specification, detailing modules that are 44% shorter than DDR5 and offer a baseline transfer rate of 12.8 GT/s, more than double DDR5's baseline. The new design quadruples memory bandwidth per module compared to DDR5's baseline, achieving up to 153.6 GB/s, and introduces a 96-bit data bus width with four 24-bit sub-channels.

hardware memory ram jedec
1 source 1 report 3d ago

GTA VI Console Resolutions Revealed: PS5/Xbox Series X at 1440p, Series S at 720p

Rockstar Games has released technical specifications for GTA VI on consoles, detailing rendering resolutions and upscaling technologies. The PlayStation 5, Xbox Series X, and PlayStation 5 Pro render at 1440p before upscaling to 4K, while the Xbox Series S renders at 720p, upscaling to 1080p using FSR 4. All versions target a 30FPS frame rate.

releases gaming consoles gta vi rockstar
1 source 1 report 3d ago

Leaked Ohio Bill Could Allow AI Data Centers to Use Eminent Domain for Land Seizure

A leaked bill in Ohio proposes classifying data centers as public utilities, which would expand eminent domain law to allow them to seize private land for construction. This change could enable transmission developers to acquire land for data centers or their power infrastructure, potentially overriding local opposition to such developments.

general data centers eminent domain ohio legislation
1 source 1 report 3d ago

Nvidia Denies Designing Vera CPU Specifically for Agentic AI Workloads

Nvidia's head of data center CPU business, Ian Finder, stated it is "absurd" to suggest the Vera CPU was designed with foreknowledge of the "agentic moment" in AI. Instead, Vera's development focused on co-design with Nvidia's accelerators, building on the approach used for the Grace CPU. This strategy aims to support accelerated data center and AI workloads by ensuring the CPU efficiently feeds the accelerator.

hardware nvidia cpu ai
1 source 1 report 3d ago

China Threatens Countermeasures Over New US AI GPU Rental and Export Restrictions

The US government expanded export controls on high-end AI accelerator chips, including Nvidia's Blackwell line, requiring licenses for large-quantity rentals to countries on its D:5 embargo list. China's Ministry of Commerce stated it will implement countermeasures to protect its national interests, signaling potential escalation in tech trade tensions.

general ai export controls geopolitics semiconductors
6 sources 10 reports 10d ago

Google Search Adds 1998-Style Interface Easter Egg for "Zelda" Searches

Google Search introduced an Easter egg that transforms search results into a 1998-style interface when users search for "Zelda" or "Ocarina of Time." This feature, activated by clicking a Master Sword icon, displays blue hyperlinks and brief excerpts, reminiscent of early Google search, and coincides with the upcoming remaster of The Legend of Zelda: Ocarina of Time.

releases google search easter egg zelda
6 sources 7 reports 80d ago

Humanoid Robots Surpass Usain Bolt's 100m Record at World Humanoid Robot Games

Two Chinese humanoid robots, Tiangong Ultra and Honor's Lightning, recorded 100-meter dash times of 9.39 and 9.47 seconds respectively, exceeding Usain Bolt's human record of 9.58 seconds. This development demonstrates rapid progress in bipedal robot locomotion, though the robots still lack controlled stopping mechanisms.

ai robotics humanoid competition sports
1 source 1 report 4d ago

Nvidia's Vera CPU Shows 39% Performance Gain with Software Optimization, Requires Architectural Tuning

Nvidia's Vera CPU, featuring custom Olympus cores, can achieve up to a 39% performance increase in computationally intensive workloads through architecture-specific software optimizations. While Vera offers high single-thread performance and is designed for agentic AI and data processing, it requires software adaptation to fully utilize its resources, unlike its predecessor Grace which used off-the-shelf Arm cores.

hardware nvidia cpu datacenter ai
1 source 1 report 4d ago

AMD Highlights Native FP64 on Instinct MI430X, Contrasting Nvidia's Emulation

At SC26, AMD emphasized the native FP64 double-precision support in its Instinct MI430X accelerator, positioning it against Nvidia's FP64 emulation methods. AMD argues that emulation is insufficient for most high-performance computing (HPC) workloads, which often require strong FP64 vector performance beyond dense matrix multiplies. This matters for scientific computing and simulation fields that rely on precise double-precision math.

hardware hpc gpus fp64 supercomputing
1 source 1 report 4d ago

Chinese J-35 Fighter Jet Developers Create 5-Meter 3D-Printed Load-Bearing Aircraft Structure

Engineers at China's Shenyang Aircraft Design and Research Institute, developers of the J-35 fighter jet, have successfully 3D-printed a primary load-bearing metal structure over 16.4 feet (five meters) long for military aircraft. This achievement utilizes segmented printing and additive fusion joining techniques, representing a significant advancement in additive manufacturing for large-scale aerospace components.

hardware 3d printing aerospace manufacturing titanium
1 source 2 reports 5d ago

AMD Powers 40% of TOP500 Supercomputers; China's LineShine Remains Fastest

AMD now powers 40% of the supercomputers on the TOP500 list, an increase of 14% from the previous year, totaling 202 systems. The new Lux supercomputer, developed by AMD and the U.S. Department of Energy (DOE) for the Genesis Mission, debuted at position 14. China's LineShine supercomputer retained its position as the world's fastest, achieving 2.198 FP64 ExaFLOPS.

releases supercomputing amd hpc ai
1 source 1 report 5d ago

Nvidia Announces Rubin HPC GPU with Over 50,000 CUDA Cores for Agentic Workflows

Nvidia unveiled the Rubin HPC GPU at Supercomputing 2026, featuring over 50,000 CUDA cores and NVLink 7 interconnect. This new GPU is designed to accelerate agentic workflows and simulations for scientists, engineers, and analysts working with large datasets, aiming to improve performance in fields like climate research and genomic analysis.

hardware nvidia gpu hpc supercomputing
1 source 1 report 5d ago

Intel to acquire Classiq, shifting quantum computing efforts toward commercialization

Intel plans to acquire quantum computing company Classiq, marking a strategic shift from research to commercialization of its quantum efforts. This acquisition will integrate Classiq's quantum software development platform with Intel's hardware, creating a full hardware-to-software stack for quantum computing. The deal aims to establish Intel as a leading vendor in the quantum computing market, leveraging Classiq's existing customer base and revenue streams.

hardware quantum computing acquisition intel software
1 source 1 report 5d ago

Jailbroken PS4 runs 9-billion-parameter AI model offline at up to 43 tokens/second

Developer Leonardo Alegría's open-source project, Garlic LLM, enables a jailbroken PlayStation 4 to run the Qwen3.5-9B AI model offline on its 2013-era GPU. This allows the console to process AI tasks, including voice conversation and dictation, at speeds up to 43 tokens per second, by directly accessing the GPU and fitting the model into the console's memory.

ai gaming open-source hardware
1 source 1 report 5d ago

Fable 5.2 AI Generates 96KB Browser-Based Underwater World Demo

A front-end developer used Fable 5.2 to create "Abyssal Garden," a 96 KB browser-based 3D underwater world demo. The demo utilizes procedural generation and shaders for visuals, running on WebGPU and WebGL, demonstrating the AI's ability to generate complex graphics from a concise prompt.

ai webdev graphics fable
1 source 1 report 6d ago

DARPA tests autonomous ground vehicles in simulated combat for pack hunting capabilities

DARPA conducted tests at Fort Knox with autonomous ground vehicles (UGVs) operating in packs for reconnaissance and surveillance in simulated combat. The exercise, part of the RACER program, demonstrated UGVs negotiating terrain and identifying targets autonomously, with human operators retaining control over kinetic engagements. This development allows UGVs to perform battlefield tasks without manual driving, addressing requests from European allies.

releases robotics defense autonomy darpa
1 source 1 report 6d ago

Intel Reclaims Overall CPU Market Share from AMD in Q3 2026 Due to Desktop Performance

Intel regained overall x86 CPU market share from AMD in Q3 2026, marking its first lead since Q2 2025, driven by strong desktop PC performance. This occurred despite an overall decline in x86 CPU shipments quarter-over-quarter and year-over-year, primarily due to reduced client processor demand, while server CPU sales increased significantly.

hardware cpu market share intel amd
1 source 1 report 6d ago

University of Rochester Converts Ender 3 3D Printer into $450 Bioprinter for ELM Research

Microbiologists at the University of Rochester modified a $200 Ender 3 FDM 3D printer into a functional 3D bioprinter for engineered living material (ELM) research, costing approximately $450 in total. This development significantly reduces the cost barrier for bioprinting, making ELM research more accessible compared to commercial bioprinters that typically exceed $100,000.

hardware bioprinting 3d printing synthetic biology research
1 source 1 report 7d ago

Gears of War: E-Day Uses SSD as RAM Extension to Reduce Memory Usage

Gears of War: E-Day uses storage technology that leverages fast SSDs as an extension of system memory to dynamically load and offload complex game data. This method allows the game to run its visually detailed world on systems with limited RAM, as confirmed by testing on both high-end and more modest PC configurations.

releases gaming storage unreal engine 5 pc hardware
1 source 1 report 7d ago

Malaysia Plans New Law to Track Data Center GPU Imports and Power Use to Combat Smuggling

Malaysia is developing a new law requiring data centers to report AI accelerator imports and power consumption to prevent restricted hardware from being smuggled to countries like China. The legislation also aims to optimize power infrastructure use by charging data centers for reserved but unused electricity capacity. This initiative seeks to ensure AI hardware imported into Malaysia is used domestically and to deter illicit re-exportation.

general malaysia data centers ai hardware smuggling
1 source 1 report 7d ago

Pentagon struggles to account for $31 billion in F-35 parts, GAO report finds

A Government Accountability Office (GAO) report indicates the Pentagon cannot track 26 million F-35 aircraft parts, valued at $31 billion, due to an outdated inventory system. This issue raises concerns about the readiness and reliability of the F-35 fleet, following a recent incident where F-35 parts were reportedly diverted through Hong Kong and potentially seized by China.

general pentagon f-35 gao inventory
1 source 4 reports 8d ago

Intel's Nova Lake desktop CPUs to use 18A-P node, more details at CES

Intel confirmed its Nova Lake desktop processors will be the first client products to use the enhanced 18A-P node, with more information expected at CES in January. The company is also reallocating Panther Lake production capacity to increase Granite Rapids data center CPU output.

hardware intel cpus manufacturing processors
1 source 1 report 9d ago

Techie Uses Voltage Glitching to Convert Nvidia CMP-170HX into A100-Equivalent GPU

A tech enthusiast successfully converted an Nvidia CMP-170HX cryptomining GPU into a card with performance comparable to an 80GB A100 GPU through voltage glitching. This modification unlocked disabled VRAM and features on the GA100 silicon, originally present but factory-disabled for lower-tier models. The process involved exploiting the card's eFuse controller using a custom circuit board and an STM32G4 microcontroller.

hardware gpu modding nvidia
1 source 1 report 9d ago

Marijuana Farmers Convert Cultivation Facilities into AI Data Centers Amid Industry Decline

Cannabis farmers facing declining revenue and oversupply are exploring converting their cultivation facilities into "mini data center nodes." These facilities are suitable for data centers due to existing power, water, and HVAC infrastructure, offering a new use for struggling agricultural properties.

ai data centers cannabis real estate
1 source 1 report 10d ago

AMD to Reintroduce 3D V-Cache to Data Centers with Venice-X CPUs in 2025

AMD will bring its 3D V-Cache technology back to data center HPC with Venice-X CPUs in the second half of 2025, after a four-year absence. Venice-X will offer 96 cores, 1,152 MB of L3 cache, 5.3 GHz boost clocks, PCIe 6.0, and significantly increased memory bandwidth, combining high frequency with large cache for HPC workloads.

hardware amd epyc hpc 3d v-cache
1 source 1 report 10d ago

Prusa Releases PETG ESD Filament with Carbon Fiber and Nanotubes for Electronics Manufacturing

Prusa has launched Prusament PETG ESD, a new 3D printing filament containing carbon fiber and carbon nanotubes, designed for electrostatic-dissipative (ESD) applications in electronics production. This filament offers improved dimensional accuracy and reduced warping compared to other ESD filaments, and is priced lower than competitors while being the first to include carbon fiber reinforcement.

releases 3d printing materials electronics manufacturing
1 source 1 report 10d ago

OpenAI's $30 Billion Georgia Data Center Faces RICO Racketeering Lawsuit

Residents of Effingham County, Georgia, filed a RICO lawsuit against OpenAI, county commissioners, and other entities involved in the "Project Camellia" data center. The lawsuit alleges fraud, concealment, improper land transactions, and zoning violations in the approval process for the $20-30 billion project. Plaintiffs claim the data center has negatively impacted property values and was advanced through confidential negotiations.

startups openai data centers rico lawsuit georgia
More stories →