Nvidia announced BlueField-4 STX at GTC 2026 on March 16, a modular reference architecture for accelerated storage designed to address the data access bottleneck limiting agentic AI inference.
Another fantastic Newegg combo deal has combined the eight-core AMD Ryzen 7 9850X3D with 32GB of Corsair Vengeance DDR5-6000 RAM and an Asus ROG Strix X870E-E motherboard for just $1,019.99, making the RAM effectively just $111 in this build.
Have you ever wanted to use a fan that's more than three times as loud as the other option while providing the same performance? If you answered in resounding joy, then this project is exactly what you've been looking for. A YouTuber 3D-printed a fan that's actually made up of 15 tiny fans, fit inside the frame of a regular 120mm fan modelled after the Noctua NF-A12x25.
Taiwan imports almost all of its energy and requires large amounts of LNG to sustain its electrical grid. That grid is then used by local chipmakers — like TSMC who is responsible for making most of the world's high-end chips. Fabrication for these chips requires helium, which Taiwan also imports and right now, the Iran-U.S. conflict has made it difficult to acquire both.
Nvidia shows off its next-generation Kyber rack-scale solution to be powered by Rubin Ultra GPUs with four compute chiplets and 1 TB of HBM4E memory per package.
Maker Will Whang designed the MK4001MTD USB Bridge to facilitate the use of the world’s smallest (0.85-inch) mechanical hard drives, originally released by Toshiba in 2004.
The Seagate FireCuda X1070 is a budget-minded SSD done mostly right. It performs well enough and has excellent support, but initial pricing is high. A Gen 4 SSD that you can just throw into any system and get the job done.
At GTC 2026, Nvidia revealed the Groq 3 accelerator and Groq LPX rack as part of the Vera Rubin platform. These SRAM-packed, inference-focused chips deliver large amounts of memory bandwidth to help Rubin deliver low-latency interactions with AI models spanning trillions of parameters and million-token contexts.
Nvidia announced more details about its new 88-core Vera data center CPUs, claiming impressive 50% performance gains over standard CPUs, fueled by a 1.5X increase in IPC from its Olympus cores. The firm also unveiled its new Vera CPU Rack architecture, which brings 256 liquid-cooled CPUs into one rack for CPU-centric workloads.
As Meta introduces its lineup of new AI chips, the company joins other tech giants in diversifying the AI accelerators used for specific workloads, and says that mainstream GPUs built for large-scale pre-training are less cost-effective for inference workloads.