📁 Hardware

This Hardware archive tracks the practical side of local AI infrastructure: GPUs, NPUs, mini PCs, edge accelerators, memory bandwidth, and power efficiency tradeoffs that directly impact LLM inference quality. We prioritize benchmark-backed updates and deployment notes useful for real build decisions, from compact home labs to enterprise pilot clusters. Use this stream to compare total cost of ownership, thermal constraints, and model-fit scenarios across current devices, then deepen with our hardware pillar guide and connected LLM coverage.

An Australian modder, Trashbench, has successfully run a PC inside a freezer, overcoming the challenges of condensation. The solution lies in the size of the freezer and the use of socks filled with silica gel. Overclocking gains were minimal at -28°C.

2026-01-13 Fonte

We tried Amazon's new AI wearable Bee. It's not for pro users yet, but more features are expected this year. The device shows promise but is still in early stages. Amazon plans to add more functionalities by the end of the year, making it more appealing to a broader audience.

2026-01-13 Fonte

Meta has established 'Meta Compute', a new organization focused on building next-generation data centers designed to power large-scale artificial intelligence (AI) applications. The initiative includes the development of custom computing architectures and infrastructures capable of supporting power consumption in the gigawatt range.

2026-01-12 Fonte

A fully 3D-printed drone, the Peregreen V4, has reclaimed the title of world's fastest drone, surpassing the previous record of 626 km/h. The quadcopter, the result of work by a father and son, sets a new speed record for drones.

2026-01-12 Fonte

Asus has revised the liquid metal application on its $4,000 ROG Matrix RTX 5090 cards. Overclocker der8auer reports that retail versions feature a 'much more professional' spread, reducing the risk of accidental liquid metal leakage. Previous samples showed poor coverage, while the metal is now distributed more evenly and closer to the IHS.

2026-01-11 Fonte

The Japanese graphics card market is under severe strain. Multiple shops are reporting that GPUs, especially higher-end models like the RTX 5060 Ti, are selling out almost as soon as they arrive. Demand far exceeds supply, creating procurement difficulties for retailers.

2026-01-10 Fonte

CES 2026 in Las Vegas is underway, featuring presentations from Nvidia, Sony, and AMD. The event showcases the latest technology innovations, opening to the public after press-exclusive days. Highlights include advancements in artificial intelligence and chip technology.

2026-01-09 Fonte

Sleepbuds maker, Ozlo, is gearing up for a year full of news. New products and AI-powered features are coming, with the aim of creating a comprehensive platform for analyzing sleep data. The company aims to improve the user experience and offer innovative solutions in the field of wellness.

2026-01-09 Fonte

At a press Q&A session at CES 2026, Nvidia CEO Jensen Huang discussed topics ranging from Rubin's design philosophy to inference economics, power constraints, and open models. Huang emphasized Nvidia's ongoing commitment to software development and maintenance.

2026-01-09 Fonte

A new whitepaper illustrates how artificial intelligence can optimize piezoelectric micromachined ultrasonic transducers (PMUTs) for biomedical applications. The approach combines cloud-based FEM simulations with neural surrogates, transforming iterative design into systematic inverse optimization. The results show significant improvements in sensitivity and bandwidth, achieved in drastically reduced times.

2026-01-09 Fonte

Former President Donald Trump and Commerce Secretary Howard Lutnick praised Intel for the launch of the new Panther Lake chips, made using the Intel 18A process. They emphasize how investments in the company are generating tens of billions of dollars for Americans and reaffirm the importance of regaining U.S. leadership in cutting-edge semiconductor manufacturing.

2026-01-09 Fonte

Chinese researchers have demonstrated a capacitor-less, DRAM-like memory cell, potentially enabling embedded or 3D stacked memory. The technology boasts fast writes and long retention, but questions remain regarding its manufacturability and commercial viability.

2026-01-09 Fonte

HBM4 memory is now expected to reach volume production no earlier than the end of Q1 2026 due to Nvidia's decision to revise its memory specs upward for its next-gen Rubin GPU platform. This delay could impact the product roadmaps relying on this advanced memory technology.

2026-01-09 Fonte

Nvidia has reportedly imposed non-refundable, full upfront payment terms on H200 GPU sales to China-based clients. This decision is reportedly aimed to hedge regulatory risks, as Chinese customers had ordered more than 2 million H200 units.

2026-01-08 Fonte

Nvidia is reportedly requiring its customers in China to pay upfront in full for its H200 AI chips. This decision comes amid uncertainty regarding approvals from both the United States and Beijing, signaling potential tensions in the semiconductor market and trade dynamics between the two nations.

2026-01-08 Fonte