DIY $8,500 Zeus Supercomputer Replaces Cloud Tools for AI Business
Jay’s DIY Zeus supercomputer, built for roughly $8,500, offers a high‑performance alternative to traditional cloud AI services. Powered by an AMD Ryzen 9 CPU, 128 GB of RAM, 14 TB of storage and an Nvidia 5090 GPU, it runs on Unraid OS with Docker containers for flexible workflow automation. Zeus targets data‑intensive tasks such as web scraping, email verification and model training, promising up to an 80% cut in recurring cloud expenses. Its modular architecture lets businesses scale and customize hardware as AI workloads evolve.
Trying Out Snapdragon X Elite With The Acer Swift 14 AI Laptop On Ubuntu 26.04
Testing Ubuntu 26.04 on the Acer Swift 14 AI equipped with a Qualcomm Snapdragon X Elite revealed multiple regressions. The qcom‑firmware‑extract tool now refuses the device, and the generic Linux 6.19 kernel only provides LLVMpipe software rendering, while the Qualcomm‑specific X1E kernel stalls at a gray screen....
Mid-Range Robot Vacuum Puts Flagships to Shame: Ecovacs Deebot T90 Pro Omni Review
The Ecovacs Deebot T90 Pro Omni, priced just under $900, delivers flagship‑level suction of 30,000 Pa and a 27 cm Ozmo Roller that outperforms many premium models in mopping tests. Its fast‑charging system and sizable water tanks enable extensive area coverage without...
What Is a Computer?
Modern flagship smartphones now match entry‑level laptops in CPU speed and RAM, prompting questions about their viability as primary computers. However, the Android operating system and Google’s tightening security policies prevent many essential desktop applications and limit sideloading of arbitrary...
OpenRazer 3.12 Released With Support For Newer Razer Products On Linux
OpenRazer 3.12, the latest open‑source driver suite for Razer peripherals on Linux, was released today. The update adds native kernel support for several recent Razer products, including the BlackWidow V4 Tenkeyless HyperSpeed keyboard, Mouse Dock Pro, Huntsman V3 Pro 8KHz,...
Linux 7.1 Will Bring Power Estimate Reporting For AMD Ryzen AI NPUs
Linux kernel 7.1 will introduce a new ioctl, DRM_IOCTL_AMDXDNA_GET_INFO, that exposes real‑time power‑estimate data from AMD Ryzen AI NPUs. The same update adds column‑utilization metrics, allowing user‑space tools to see how busy the NPU is. These changes arrive via the...
Faster Qubit Readings Now Avoid Unwanted Energy State Changes
Google Quantum AI researchers have added an inductive shunt to transmon qubits, eliminating offset‑charge sensitivity that caused measurement‑induced state transitions (MIST). The shunt provides an alternate current path, allowing dispersive readout without large detuning or extensive calibration. Tests at 10 mK...
A Foldable Laptop That Doesn’t Feel Like a Compromise
Huawei unveiled the MateBook Fold Ultimate Design, a foldable laptop that expands from a 13‑inch chassis to an 18‑inch OLED workspace while weighing just 1.45 kg with its detachable keyboard. Launched in mid‑2025, the device features a dual‑layer OLED panel, a...
Keychron Launches $59.99 C0 HE 8K One-Handed Gaming Keyboard
Keychron has launched the C0 HE 8K, a left‑handed, one‑handed gaming keyboard priced at $59.99. The device features 35 keys, an ABS chassis, aluminium plate, and Ultra‑Fast Lime magnetic Hall‑effect switches with 25 g actuation. It includes an ergonomic angled space bar,...
AMD Sees Agent Computers as the Next Step in AI PCs
AMD released a guide outlining two local‑AI hardware paths—RyzenClaw and RadeonClaw—designed to run large language models entirely on Windows without cloud services. The RyzenClaw configuration uses a Ryzen AI Max+ system with 128 GB memory, delivering about 45 tokens per second and supporting...
Windows 11 Insiders Get Support for >1,000 Hz Monitor Refresh Rate
Microsoft released Windows 11 Insider builds 26100.8106 and 26200.8106, adding native support for monitor refresh rates exceeding 1,000 Hz. The update marks the first time the OS handles four‑digit refresh rates, aligning software with emerging ultra‑high‑refresh displays. Philips and AOC have already...
MacStories Weekly: Issue 504
MacStories Weekly issue 504 bundles three distinct tech stories. Apple announced an iPad Air powered by the new M4 chip, promising notable performance gains over the M2 generation. The newsletter also showcases a DIY bookmark manager built in Notion that leverages...
(PR) GMKtec Unveils 2nd Gen EVO-T2 Desktop AI Supercomputer with 180 TOPS
GMKtec introduced the second‑generation EVO‑T2 Desktop AI Supercomputer, built on Intel’s Core Ultra X9 388H and X7 358H processors using the 18A (1.8 nm) Panther Lake architecture. The compact system delivers up to 180 TOPS of AI compute and can run local large language...
Smart Home Upgrades That Help Protect Your Home’s Structure
Smart home technology is now being leveraged to safeguard a house’s structural integrity. Devices such as moisture leak sensors, smart hygrometers, termite monitors, automated water shutoff valves, and foundation crack detectors provide real‑time alerts and even autonomous responses. By integrating...
NVIDIA Release RTX Powered Godot Fork
At GDC 2026 NVIDIA unveiled a custom fork of the open‑source Godot engine that adds native RTX support and real‑time path tracing. The fork, released under the MIT license on GitHub, modifies the rendering pipeline and material system to enable...
SLZB-06 vs SLZB-MR vs SMHUB: All SMLight Coordinators Compared
SMLight’s coordinator lineup now spans USB‑only dongles, PoE‑enabled units, dual‑radio MR models, and full‑stack Linux hubs. The SLZB‑07 series provides basic Zigbee or Thread connectivity, while the SLZB‑06 adds Ethernet, PoE and USB passthrough via an ESP32‑S3 controller. The SLZB‑MR...
How Chainway Help Solves RFID Linen Management at Marriott: RFID Journal Case Study
Marriott International deployed Chainway’s MC51 UHF handheld as part of an RFID Linen Management System, embedding washable tags in sheets, towels and robes. The solution automates inbound receipt, housekeeping issuance, and outbound handover through bulk scanning at key checkpoints. MC51’s...
Agentic AI and the Future of Engineering
At Synopsys' Converge keynote, Sassine Ghazi introduced "agentic AI" as a practical layer that augments engineers rather than replaces them, addressing the mounting complexity of chip and system design. He outlined a five‑level autonomy roadmap (L1‑L5) that moves from simple...
Weekly Wrap: Samsung Plans New Smart Glasses
Samsung announced plans to launch its first AI‑enabled smart glasses this month, featuring an eye‑level camera that streams directly to a paired mobile device. The move pits Samsung against Meta, whose Ray‑Ban glasses currently command 82% of the global market....
Skyrover S1: 4K, 3-Axis Gimbal, Forward Obstacle Avoidance, and Still Under 249g
The Skyrover S1 is a sub‑249 g drone that packs a 1/2‑inch Sony sensor capable of 4K 60 fps video and 8K photos, a true 3‑axis mechanical gimbal, and forward obstacle avoidance. It offers up to 40 minutes of flight per battery and a 39‑km...
2026 Apple TV 4K Is Ready, But Won’t Launch Without a Smarter Siri
Apple is postponing the 2026 Apple TV 4K and HomePod Mini launch to sync with a dramatically upgraded Siri that will debut with iOS 27 later in 2026. The new Siri promises conversational depth, contextual awareness, and tighter ecosystem integration, positioning the...
New Headphones
Matt’s latest post spotlights the Sennheiser HDB 630 Wireless Over‑Ear Headphones, praising their sound quality and comfort. He links directly to the Amazon listing and credits Philip Kaplan for the tip. The entry sits among a series of related headphone...
Raspberry Pi 5 Local AI Assistant Gains Offline Vision, Voice & Image Generation
Raspberry Pi 5 now supports sophisticated offline AI by pairing the LM AA50 accelerator and the AI Hat Plus 2 with the 2‑billion‑parameter CR3VL vision‑language model and a latent consistency generator. The hardware can interpret images, transform visuals, and create pictures from spoken or typed prompts...
Samsung Galaxy S26 Ultra Teardown: What’s Inside the New Flasgship
Samsung’s Galaxy S26 Ultra teardown reveals incremental hardware upgrades across camera, charging, and thermal management while introducing a privacy screen that sacrifices brightness. The new 50 MP telephoto lens with a wider f/2.9 aperture and prism‑based lens layout trims camera depth...
The Capital-Intensity Trap: The Decoupling Mechanism of Emerging Market Foundries and U.S. Fabless
The article outlines a "capital‑intensity trap" where rising energy prices and higher U.S. Treasury yields squeeze margins for hardware manufacturers while leaving asset‑light designers relatively untouched. It illustrates the mechanism with a March 13 case study: Brent crude above $100 and...
“This Is Not The Computer For You”
The article reflects on Sam Henri Gold’s essay about the MacBook Neo, noting how reviews often pigeonhole devices into narrow user categories. It points out the Neo’s modest specs—$599, A18 Pro chip, 8 GB RAM—and the common advice that it’s unsuitable for developers...
Quantum Method Processes Problems in Parallel, Cutting Solution Time by 20%
Researchers at Tokyo University of Agriculture and Technology introduced multi‑tasking quantum annealing (MTQA), a method that runs multiple combinatorial optimisation problems simultaneously on a single quantum annealer. By embedding distinct problem graphs into separate regions and using idle qubits, MTQA...
Robots as Autonomous Systems: Which IT Powers the “Brain” Of Modern Machines?
Robots are evolving into fully autonomous systems that require on‑device AI processing rather than cloud reliance. Edge computing platforms place the computational "brain" directly on the robot, cutting latency to fractions of a second. Traditional rack‑mounted servers cannot meet the...
Apple Keeps iPhone Prices Stable Despite Rising Memory Prices
Apple is reportedly planning to keep the launch prices of the upcoming iPhone 18, iPhone 18 Pro and Pro Max stable, despite a sharp rise in DRAM and NAND memory costs driven by AI data‑center demand. Analyst Ming‑Chi Kuo says...
The AI Interconnect War: Copper Fights Back Against Optics
The post examines the escalating demand for short‑reach interconnects as AI models grow larger and data centers shift to massive GPU clusters. It compares copper‑based options—DAC, ACC, and AEC—with optical solutions like AOC and emerging Co‑Packaged Optics (CPO). Active Electrical...
Hello, Claude? Are You There?
A wave of AI‑infrastructure shortages is gripping the tech sector, with CEOs from OpenAI, Oracle, Microsoft, Alphabet and Intel all flagging severe constraints on GPUs, power, memory and data‑center capacity. The scarcity, first noted in early 2025, is projected to...
Apple’s Platform Security Guide Adds a Brief Note on the MacBook Neo’s On-Screen Camera Indicator
Apple’s Platform Security Guide now highlights a new privacy safeguard on the MacBook Neo, which integrates the A18 Pro silicon and dedicated camera‑security hardware. The design guarantees that any camera activation— even by software with root or kernel privileges—must trigger...
Linux 7.0 Cpupower Now Handles Systemd Service Setting EPP, Intel P-State Turbo Boost
Today's merge adds two key power‑management enhancements to the cpupower utility in the upcoming Linux 7.0 kernel. Administrators can now set the ACPI Energy Performance Preference (EPP) directly in the cpupower systemd service file, automating performance‑efficiency tuning. The tool also gains...
Connected 594: Vibing at 50 / Warranted But Not Sexy / The Relay Account Has Signed In
Connected 594, the latest episode of the Relay FM podcast, dives into Apple’s 50‑year milestone while reviewing the newly announced MacBook Neo laptops. Hosts Stephen and Federico share hands‑on impressions, noting the device’s sleek design but questioning its price point....
CIQ Announces General Availability of RLC Pro AI, Enterprise Linux Built to Deliver More From Every GPU
CIQ has launched the general availability of Rocky Linux from CIQ Pro AI (RLC Pro AI), an enterprise Linux distribution optimized for AI inference and GPU workloads. The OS includes pre‑configured PyTorch, NVIDIA CUDA, and DOCA‑OFED stacks, and runs on...
Crusoe Unveils Crusoe Edge Zones to Deliver High-Performance AI Infrastructure
Crusoe announced the launch of Crusoe Edge Zones, modular data‑center units powered by its Crusoe Spark platform, to bring high‑performance AI compute to virtually any location. Built in the new Spark Factory, these zones can be stood up in as...
Lightmatter Joins XPO MSA as Founding Member to Accelerate High-Density Optical Interconnects
Lightmatter announced its participation as a founding member of the XPO (eXtra‑dense Pluggable Optics) Multi‑Source Agreement, a new optical transceiver standard driven by Arista Networks. The XPO specification promises up to four times the rack density of existing pluggable modules...
Huawei Targets AI Data Centre Reliability with Xinghe AI Fabric 2.0
At Mobile World Congress, Huawei unveiled Xinghe AI Fabric 2.0, a three‑layer AI‑centric architecture designed to improve reliability and performance of AI‑heavy data centres. The suite introduces Rock‑Solid Architecture 2.0 with an AI Eagle‑Eye engine that can monitor up to 200,000 service...
(PR) Sama Introduces V62 Curved Panoramic Glass PC Case Series
Sama launched the V62 Curved Panoramic Glass PC case, featuring a 270° tempered‑glass panel and integrated ARGB lighting. The chassis ships with seven pre‑installed ARGB fans and supports up to 360 mm radiators on both top and bottom for high‑performance cooling....
Samsung Reportedly to Use 2 Nm Process on HBM4E Base Die
Samsung is set to fabricate the base die of its next‑generation HBM4E memory using a 2 nm process, following the recent launch of the industry’s first commercial HBM4. The move coincides with a redesign of the HBM4E power‑delivery network, raising the...
Benchmark Telecom Introduces Ranger XL Mobile Mast for Temporary Coverage
Benchmark Telecom, a Belgian telecom infrastructure provider, launched the Ranger XL, a mobile transmitter designed for temporary network capacity. The telescopic mast extends up to 30 meters and can be operational in just 30 minutes. It targets high‑traffic events such as festivals and...
Huistone Patent Targets Large Area Resin Printing Economics
Fujian Huistone 3D Technology has filed Chinese patent CN121608383A describing a 3D printer that relocates a UV light source across the build area instead of using a fixed, full‑size array. The moving‑light engine is driven by dual servos and screw mechanisms,...
Ravi Subramanian on Trends that Are Shaping AI at Synopsys
Ravi Subramanian, Synopsys' Chief Product Management Officer, explained how AI is driving the convergence of silicon design and systems engineering, a shift highlighted at the Synopsys Converge event. He noted the industry’s move from throughput‑focused metrics to efficiency‑centric measures such...
To Sparsify or To Quantize: A Hardware Architecture View
Hardware architects face a trade‑off between sparsity and quantization for compute‑bound generative AI models. Unstructured sparsity offers maximal pruning but forces complex routing and poor SIMD utilization, prompting a shift toward structured patterns like N:M and block‑sparse attention. Quantization reduces...
(PR) Imec Launches University Consortium Around Next Generation of Chips
Imec announced a European university consortium of 26 institutions to develop CMOS 2.0, a post‑CMOS scaling paradigm that leverages fine‑grain wafer stacking and heterogeneous integration. The partnership will fund 26 PhD researchers who will remain at their home universities while accessing...
AMD HDR/Color Improvement For Their Linux Driver & KDE - Co-Developed By Claude Code
AMD engineer Harry Wentland announced new HDR and color‑pipeline enhancements for the AMDGPU Linux driver, built on the DRM Color Pipeline API introduced in Linux 6.19. The work adds a color‑space conversion (CSC) operation via the `drm_colorop` patch and integrates it...
AMD, NVIDIA, OpenAI & Others Form An Optical Scale-Up Consortium
AMD, Broadcom, Meta, Microsoft, NVIDIA and OpenAI have launched the Optical Compute Interconnect (OCI) Multi‑Source Agreement consortium. The group aims to create an open, multi‑vendor ecosystem for optical scale‑up interconnects that replace copper in AI data‑center clusters. OCI’s specification combines...
Scaling AI for Everyone
OpenAI announced a $110 billion financing round that values the company at $730 billion pre‑money, with $30 billion each from SoftBank and Nvidia and $50 billion from Amazon. The capital backs a strategic partnership with Amazon and next‑generation inference compute from Nvidia, expanding global...
Akave Raises $6.65M to Challenge Traditional Cloud Storage with Compute-Agnostic Platform Built for AI Applications
Akave has launched Akave Cloud, a decentralized, S3‑compatible storage platform built on Avalanche’s Layer 1 blockchain, targeting AI and analytics workloads. The service offers flat‑rate pricing at $14.99 per terabyte per month with zero egress fees, on‑chain verifiable audit trails, and...
Micron Sets New Benchmark with the World’s First High-Capacity 256GB LPDRAM SOCAMM2 for Data Center Infrastructure
Micron Technology announced the shipment of its 256 GB SOCAMM2 LPDRAM module, the first server memory built on a monolithic 32 Gb LPDDR5X die. The new module delivers one‑third the power consumption and footprint of comparable RDIMMs while offering 1.33 times more capacity...