For you Ai Security Dev Cloud Hardware Startups Releases General

From Tom's Hardware · 40 stories

1 source 1 report 10d ago

ASRock Releases New Firmware After Previous Update Bricked Radeon RX 7900 GPUs

ASRock released VBIOS version 1.0.6 for its Radeon RX 7000-series graphics cards, following reports that the prior 1.0.5 update bricked some RX 7900 XT and RX 7900 XTX GPUs. The company has not provided a changelog or details on what the new firmware addresses, leaving affected users uncertain about warranty coverage.

hardware asrock radeon firmware gpu
1 source 1 report 10d ago

Cognichip's ACI platform uses AI to reduce chip design times from months to days

Cognichip launched Artificial Chip Intelligence (ACI), an AI-assisted platform designed to automate semiconductor engineering tasks. ACI utilizes a large dataset and specialized models to generate, verify, and optimize chip designs, aiming to significantly shorten design cycles.

ai chip design eda semiconductors
1 source 1 report 10d ago

Cyanogen Co-founder Unveils Kora Linux OS for AI Workloads Across CPU, GPU, NPU

Kora Systems Inc., led by Cyanogen co-founder Kirt McMaster, announced Kora, a new Linux desktop OS designed to distribute AI model workloads across CPU, GPU, and NPU resources, including other local network machines. The OS features a proprietary Cortex inference engine optimized for local models and is expected to launch on PCs in 2027.

releases linux operating system ai npu
1 source 1 report 10d ago

Linux Kernel Memory Management Reviewer Steps Down Due to Influx of AI-Generated Patches

Lorenzo Stoakes, a lead Linux kernel reviewer for memory management, has reduced his duties, citing an overwhelming volume of AI-generated code submissions. This departure highlights the increasing challenge open-source projects face in maintaining code quality amidst a surge of low-effort AI contributions.

dev linux open-source ai kernel
1 source 1 report 10d ago

AMD Details MEXT Technology for Server Memory Optimization, Partnering with HPE

AMD is deploying MEXT, a predictive memory engine, to reduce server DRAM requirements by offloading cold memory pages to flash storage using an AI model. AMD claims MEXT can retain 95% of server performance with 50% DRAM capacity in some internal workloads, and plans to expand its internal deployment and partner with HPE for customer availability. This technology addresses rising DRAM prices and potential memory shortages in the enterprise market.

hardware amd memory servers hpe
1 source 1 report 10d ago

Stingray Marine Solutions deploys 3,500 AI laser drones to protect Norwegian salmon farms

Stingray Marine Solutions has deployed over 3,500 AI-powered submarine drones equipped with lasers to Norwegian salmon farms, protecting over 112 million fish. These drones identify and eliminate sea lice, offering an alternative to chemical treatments and improving fish welfare and growth. The deployment represents one of Scandinavia's largest commercial AI operations.

ai aquaculture drones robotics
1 source 1 report 11d ago

Microsoft Lobbies Against Potential Ban on Chinese Optical Networking Gear for US Data Centers

Microsoft and the Information Technology Industry Council (ITI) are lobbying against a potential FCC ban on Chinese optical transceivers for US data centers. A ban could significantly disrupt data center expansion due to China's 66% share of the global market and limited US-based alternatives, impacting the deployment of AI infrastructure.

cloud data centers supply chain networking ai infrastructure
1 source 1 report 11d ago

Frontier Model Forum Releases Jailbreak Rating Framework for AI Models

The Frontier Model Forum (FMF) has introduced a new Jailbreak Rating Framework to classify the severity and ease of deployment of AI model jailbreaks. This framework, supported by Anthropic, Google, Microsoft, and OpenAI, aims to establish a cross-vendor standard for evaluating how AI models bypass safeguards and the potential impact of such breaches.

ai security llm framework
1 source 1 report 11d ago

Prusa 3D Printer Vulnerabilities Discovered, Including Remote Code Execution

Software engineer Yannick Dixken identified 23 defects in Prusa MK4S 3D printers, including a high-severity remote code execution (RCE) vulnerability in Prusa Connect. All discovered vulnerabilities have been patched in Prusa's Buddy firmware versions greater than 7.0.0. Dixken responsibly disclosed the findings to Prusa, which coordinated a fix.

security 3d printing vulnerability firmware
1 source 1 report 12d ago

Former Ethereum Foundation Employee Sues Over Mandatory 'Source Seppuku' Pledge

A former Ethereum Foundation employee is suing the nonprofit for alleged disability discrimination, retaliation, and unfair employment practices. The lawsuit stems from a mandatory "Source Seppuku" pledge included in a 38-page internal mandate, which required employees to vow to take their own lives if they failed to uphold certain pledges related to open-source code.

general ethereum legal employment nonprofit
1 source 1 report 12d ago

DEEPX Demonstrates AI Chip Running 40 Degrees Cooler Than Competitors

South Korean startup DEEPX showcased its DX-M1 AI chip, which reportedly runs 25 to 40 degrees Celsius cooler than rival chips under the same AI workload. This thermal efficiency could address significant heat, power, and cooling challenges in AI data centers, potentially reducing infrastructure costs and environmental impact.

hardware ai chips data centers
1 source 1 report 12d ago

Nvidia RTX PRO 6000 GPU cables show heat damage despite reduced power draw

A Reddit user reported frayed and stiff braided sleeving on 12V-2x6 cables for two Nvidia RTX PRO 6000 GPUs, indicating heat damage even though the GPUs were limited to 500W, below their 600W maximum. This incident highlights ongoing issues with 12V-2x6 connectors overheating, potentially due to uneven current distribution across pins. The user's GPUs were undamaged, but the cables exhibited clear signs of thermal stress.

hardware nvidia gpu power delivery
1 source 1 report 12d ago

Smartphone Makers Increase Use of Recycled Memory to Offset Rising Costs

Smartphone manufacturers are increasing their adoption of reballed memory, salvaged from old devices, to manage production costs. This practice, currently at 3-4% penetration in entry-level phones, is projected to reach 7-9% by 2027 as DRAM prices continue to rise, potentially expanding to more premium segments.

hardware smartphones memory manufacturing cost management
1 source 2 reports 12d ago

Envision Energy deploys 800V DC power system for AI data center in Inner Mongolia

Envision Energy has deployed an 800-volt DC power system at its Galaxy AI data center in Ulanqab, Inner Mongolia, China. This system converts 10kV AC directly to 800V DC, reducing power conversion stages and supporting higher computing density for AI workloads. The technology is intended to scale the Galaxy campus to 3GW capacity and support up to one million AI accelerators.

cloud data centers power systems ai energy efficiency
1 source 1 report 12d ago

Huawei Launches Desktop PCs with Kirin X90 Processor and HarmonyOS, Citing Domestic Production

Huawei has released its Qinyun HM650 and HM650g desktop PCs, featuring its 10-core Kirin X90 processor and HarmonyOS Enterprise Edition. The company states these systems use entirely Chinese-manufactured components, aiming to reduce reliance on foreign technology amidst sanctions. While the CPU is made by SMIC on a 7nm-class process, the claim of 'entirely domestic' is debated due to reliance on foreign manufacturing equipment, materials, and IP.

hardware huawei desktop kirin harmonyos
1 source 1 report 12d ago

Wearable Sleep Trackers Use Brainwave Data for AI Training Without Explicit Consent

An investigation by Science and the Pulitzer Center's AI Accountability Network found that companies behind wearable sleep trackers are using customer brainwave data (EEG recordings) to train AI models. Several companies, including Interaxon and Emotiv, did not explicitly disclose this use in their privacy policies, raising concerns about user consent and data privacy. Interaxon has since updated its policy, acknowledging the need for explicit user information.

ai privacy wearables eeg
1 source 1 report 12d ago

LG WOLED TVs manufactured between 2020-2022 show 'edge rot' pixel degradation

RTings' long-term burn-in test identified a new issue, 'edge rot,' affecting 14 out of 16 LG WOLED TVs from 2020-2022, where pixels darken around the edges. This problem appears distinct from burn-in and has been reported by other users, raising concerns about the longevity of these specific LG panels.

hardware oled lg tv display
1 source 1 report 13d ago

Diamond Foundry develops first 200mm single-crystal diamond wafer for chip cooling

Diamond Foundry announced the creation of the world's first 200mm single-crystal diamond (SCD) wafer, a material with high thermal conductivity and electrical insulation. This development aims to address the thermal management challenges of high-power AI chips by providing an efficient cooling solution at a wafer scale comparable to semiconductor manufacturing processes.

hardware diamonds semiconductors cooling ai
1 source 1 report 13d ago

Ukraine develops combat robots for underground gas pipeline warfare against Russian troops

Ukraine's 2nd National Guard Corps, Khartiia, is developing combat robots to operate within underground gas pipelines. These unmanned ground vehicles (UGVs) are designed for reconnaissance and combat in narrow spaces, addressing a new category of warfare following an underground assault near Kupiansk.

hardware robotics military ugv ukraine
1 source 1 report 13d ago

Microsoft VP releases open-source GPU task manager for VRAM usage insights

Microsoft VP Scott Hanselman released gputaskman, an open-source GPU task manager for Windows, Linux, and macOS. The tool provides per-process GPU usage, VRAM consumption, active runtimes, and AI/LLM task-specific information, addressing the need for detailed GPU memory insights.

dev gpu open-source task manager vram
1 source 1 report 13d ago

World of Warcraft: Forever Launch Utilized 70,480 CPU Cores and 251 TB RAM

Blizzard deployed 3,924 virtual machines with 70,480 CPU cores and 251 TB of RAM to ensure a smooth launch for World of Warcraft: Forever. This infrastructure investment resulted in one of Blizzard's smoothest game launches, demonstrating the impact of significant computing resources on high-demand online services.

cloud gaming infrastructure blizzard
1 source 1 report 14d ago

TSMC Denies Report of Exploring 3D DRAM Production at Nvidia's Request

TSMC has denied a report suggesting it is exploring 3D DRAM production following a request from Nvidia, stating it will remain a pure-play logic foundry. The report indicated TSMC was preparing plans for 3D DRAM manufacturing due to AI hardware demand and memory shortages, which would significantly expand its role in the AI supply chain.

hardware tsmc nvidia dram manufacturing
1 source 1 report 14d ago

EA Discovers and Digitizes 20,000 Archival Media Assets, Leading to Game Remasters

An EA producer discovered a physical vault containing nearly 20,000 archival media assets, including master tapes for classic games. This led to a digitization project and the subsequent release of 30th Anniversary Editions for "Wing Commander III" and "Privateer 2," incorporating original footage and community-developed enhancements.

releases ea game preservation remasters wing commander
1 source 1 report 15d ago

GlobalFoundries Details FDX Fusion Technology for 7nm-Class Performance Without EUV

GlobalFoundries is developing FDX Fusion, a new FD-SOI process technology that aims to deliver 7nm-class performance using planar transistors, without requiring EUV lithography. The company plans for commercial production to begin in 2028, targeting high-volume manufacturing by 2030, and intends for FDX Fusion to serve connectivity and edge processor applications.

hardware semiconductors manufacturing fd-soi chip design
1 source 1 report 15d ago

Tesla Unveils Revamped AI Accelerator Roadmap, Adopts 'SI' Branding

Tesla has released its updated AI accelerator roadmap, detailing significant changes to its chip development. The company's 6th Generation inference accelerator is projected to deliver 80 times more compute performance than the current 4th Generation chip, and the D3 training accelerator will transition to a multi-chiplet design with stacked HBM, moving away from its previous SRAM-centric architecture. Tesla is also rebranding its AI accelerators with "SI" monikers, standing for "super intelligence."

hardware ai chips tesla
1 source 1 report 15d ago

Nagare Labs develops Z2RAM, a memory technology combining SRAM speed with DRAM density

Nagare Labs, a Georgia Tech spinout, introduced Z2RAM, a new memory technology that aims to provide SRAM-level speed at DRAM density, directly on AI chips. This technology addresses the memory bandwidth limitations of AI chips by offering higher density and speed with lower power consumption compared to existing solutions like HBM and SRAM. Z2RAM uses a novel two-transistor, no-capacitor cell design and a 3D stacking process that avoids interposers and bonding.

hardware memory ai semiconductors
1 source 1 report 16d ago

India Aims to Develop Homegrown GPU Within Two Years, Commercialize SHAKTI and VEGA Processors

India's Ministry of Electronics and Information Technology (MeitY) announced plans to develop an indigenous GPU within two years as part of its India Semiconductor Mission 2.0. This initiative aims to reduce reliance on foreign chip suppliers and build a domestic hardware and software ecosystem to support AI development. The country also intends to commercialize its existing SHAKTI and VEGA processors.

hardware semiconductors gpu india ai
1 source 2 reports 16d ago

Micron Taiwan Awards 700 Employees Bonuses Equivalent to Five Years' Salary Amid Union Disputes

Micron Taiwan is distributing significant bonuses to its employees, with 700 workers receiving a payout equivalent to five years' salary, while other staff will also receive three bonuses in October. This action occurs as Micron faces ongoing union disputes and worker demands for increased pay, driven by record profits from the AI-driven memory chip boom.

hardware micron bonuses labor semiconductors
1 source 1 report 16d ago

Congresswoman Calls for Ban on Artificial Superintelligence After Midterm Elections

Congresswoman Yassamin Ansari (D-Ariz.) called for a ban on artificial superintelligence and regulation of the AI industry following the U.S. midterm elections. This move aligns with other legislative efforts to regulate AI, including proposals from Senator Bernie Sanders, but contrasts with the Trump administration's focus on AI development as a race against China.

ai policy regulation government
1 source 1 report 17d ago

T-Mobile launches 5 Gbps fiber internet plan for $90/month with 10-year price guarantee

T-Mobile introduced a new 5 Gbps fiber internet tier priced at $90 per month, including a 10-year price guarantee and a mesh Wi-Fi extender. This new offering expands its home fiber internet service, which previously offered speeds up to 2 Gbps. The plan aims to compete in the home fiber market by providing symmetrical speeds and promotional offers for new customers.

releases t-mobile fiber internet broadband telecom
1 source 1 report 17d ago

US Army Flies VICTOR Hybrid Electric Drone with 3D-Printed Heavy-Fuel Rotary Engine

The U.S. Army Combat Capabilities Development Command (DEVCOM) Army Research Laboratory (ARL) successfully flew the VICTOR hybrid electric drone, which uses a 3D-printed heavy-fuel rotary engine. This engine design allows the drone to operate on JP8 military fuel, improving performance and reducing the need for a liquid cooling system.

hardware drones 3d printing hybrid military
1 source 1 report 18d ago

Intel Foundry releases 14A process design kit 0.9, moving closer to production

Intel Foundry released its 14A process design kit (PDK) version 0.9, a step in the development of its next-generation manufacturing technology. This PDK is based on actual silicon measurements, allowing chip designers to develop products with greater confidence in performance and power consumption. Intel plans for PDK 1.0 in Q1 2027 and mass production in 2028.

hardware intel foundry semiconductors manufacturing
1 source 1 report 18d ago

xAI Powers Mississippi Data Center with Converted Jet Engines, Faces Environmental Concerns

xAI is using eight converted jet engine turbines to power its Mississippi data center, a move that has received state-level permission but is drawing criticism from environmental groups. Concerns center on increased pollutant emissions, with the Southern Environmental Law Center challenging the permits and accusing xAI of undercounting projected emissions.

ai data centers energy environment
1 source 1 report 19d ago

Tachyum, developer of Prodigy processor, files for bankruptcy in Slovakia

Tachyum, the company developing the Prodigy universal processor, has filed for bankruptcy in Bratislava, Slovakia, following missed development and production milestones. While the Slovak subsidiary is bankrupt, the ultimate parent company in the Cayman Islands, which holds the intellectual property, is also facing liquidation proceedings. This situation could lead to the sale of Prodigy's patents to repay creditors, potentially ending the processor's development.

hardware tachyum prodigy bankruptcy processor
1 source 1 report 19d ago

New Data Shows Over 90% of Public Comments on Data Centers Are Negative

A new website, American Sentiment by Prophecy Gov, analyzed 10,202 public comments from 417 local government meetings, revealing that 90.5% of public feedback on data center proposals is negative. This data highlights growing public opposition to new data center construction across 48 US states, with concerns primarily focused on water usage, local quality of life, and transparency.

general data centers public opinion infrastructure ai
1 source 1 report 19d ago

Nvidia's Brev Cloud Platform Suffers Security Incident, SSH Keys Potentially Exposed

Nvidia's Brev cloud development platform experienced a security incident between October 4 and October 20, potentially exposing customer SSH keys. Nvidia replaced affected keys and updated customer machines, stating no evidence of improper use was found, but did not detail attacker access methods or whether customer data was accessed. This incident follows a trend of security breaches on AI development platforms in October.

security cloud nvidia breach
1 source 1 report 19d ago

Military Explores Commercial 5G Networks for Drone Detection and Defense

NATO conducted experiments in Latvia using existing 5G masts for passive drone detection and tracking, while AT&T and Ericsson demonstrated similar capabilities in the U.S. These initiatives explore repurposing commercial 5G infrastructure to enhance national security and counter drone threats.

security 5g drones military ai
1 source 1 report 20d ago

Neuralink's 'Blindsight' for restoring sight is engineered and awaiting regulatory approval

Elon Musk announced that Neuralink's second product, 'Blindsight,' designed to restore sight to blind individuals, is ready from an engineering standpoint and awaiting regulatory approval. This technology aims to create an artificial visual pathway by converting camera-captured images into brain-understandable signals via a BCI.

ai neuralink bci biotechnology neuroscience
1 source 1 report 20d ago

Lenovo's 'Encrypted' Fingerprint USB Drive Found to Be Unencrypted SD Card

A security researcher discovered that Lenovo's ThinkPlus FU100 fingerprint USB stick, marketed with AES-256 encryption, was an unencrypted SD card, allowing trivial data access. Lenovo has since issued fixes for several affected drives, requiring users to contact service for replacements or firmware upgrades.

security vulnerability lenovo usb
1 source 1 report 20d ago

San Bernardino County deploys autonomous firefighting drones with 500-gallon foam payload

The San Bernardino County Fire Protection District acquired a fleet of five autonomous firefighting drones from Seneca Systems. These drones can deliver over 500 gallons of firefighting foam and are designed to attack wildfires within 10 minutes of detection. The deployment aims to stop 90% of fires before they exceed 10 square feet, enhancing rapid initial attack capabilities.

releases drones wildfire public safety ai
More stories →