hardware

Summer savings are heating up. From the Steam Summer Sale to GeForce NOW membership discounts, this week’s GFN Thursday delivers double the deals and more ways to get the most value from cloud gaming.

James Attwood / Bloomberg: The US awards 250M in CHIPS funds to a startup co-founded by billionaire mining magnate Robert Friedland to develop silicon-carbide chips and pulsed-power tech — I-Pulse Inc

This analysis examines the growing power demands of behind-the-meter datacenters in the US, projecting capacity could exceed 40GW by 2028, and discusses how grid constraints and infrastructure limitat

Without a doubt: A world running foundation models on a billion robots won't be doing it locally. None of the embedded compute will look like it does today. All of the embedded compute will be wireles

arXiv:2606.24369v1 Announce Type: new Abstract: Reinforcement learning (RL) has become a dominant post-training paradigm, driving the emergence of high-performance RL systems such as veRL for autoregr

NVIDIA NeMo AutoModel is a tool designed to accelerate the fine-tuning process of transformer models by automating model selection and configuration. The solution leverages NVIDIA's NeMo framework to

This article discusses a collaboration between AMD and Vultr to implement AI solutions in the property insurance sector, leveraging AMD's processors and Vultr's cloud infrastructure. The partnership l

Is Argentum AI building the financing layer for the AI infrastructure boom? Yesterday, Barron’s reported that Argentum AI, the infrastructure startup led by Andrew Sobko and backed by Supermicro, had

arXiv:2606.16333v2 Announce Type: replace Abstract: Most existing approaches either fix the container in advance or optimize only a single container dimension through an outer search loop, leaving the

arXiv:2606.24075v1 Announce Type: cross Abstract: Although deep learning-based methods can achieve high accuracy in automatic modulation recognition (AMR) tasks, their high computational cost makes it

arXiv:2604.03345v2 Announce Type: replace Abstract: Kolmogorov-Arnold Networks (KANs) have recently emerged as a powerful architecture for various machine learning applications. However, their unique

Building AI systems at scale is demanding, requiring low-latency inference, fast vector search, strong GPU price-performance and infrastructure that can grow without multiplying operational complexity

OpenAI and Broadcom unveiled Jalapeño, OpenAI's first Intelligence Processor designed specifically for LLM inference , with early testing showing substantially better performance per watt than current

OpenAI: OpenAI and Broadcom unveil Jalapeño, an LLM-optimized inference chip developed from design to manufacturing tape-out in nine months, aided by OpenAI's models — Loading... - Early testing shows

OpenAI and Broadcom announced a collaboration on a specialized inference chip designed to optimize the performance and efficiency of large language model deployments. The chip, referred to as 'Jalapen

OpenAI Group PBC today revealed a custom chip called Jalapeño that it will use to power its large language models. The processor is the fruit of a collaboration with Broadcom Inc., which is no strange

Katherine Doherty / Bloomberg: Ornn, which plans to launch a marketplace for GPU capacity designed to function like an exchange to trade oil contracts, raised a 33M seed led by a16z — Ornn raised 33 m

Stephanie Palazzolo / The Information: RunPod, which rents access to non-Nvidia servers, raised 100M led by Summit Partners at a 1B valuation, a source says up from $100M after its seed in 2024 — By s

The Information: SambaNova Executive Chairman Lip-Bu Tan says the AI chip startup is set to raise 800M to 1B, sources say at a ~10B valuation, up from 2B in February — AI server chips that effectively

arXiv:2606.24039v1 Announce Type: new Abstract: Robotics increasingly relies on GPUs for parallel simulation, large-scale learning, and neural-network inference. For model predictive control (MPC) to

New research explores how game theory can optimize NVIDIA Dynamo disaggregated serving architectures, revealing critical performance thresholds and reducing worst-case AI inference latency by up to 7.

arXiv:2509.04827v3 Announce Type: replace-cross Abstract: The energy cost of Large Language Model (LLM) inference is rapidly becoming a barrier to sustainable and scalable deployment. Although modern

arXiv:2606.22447v1 Announce Type: cross Abstract: Explanation requires ground truth: to verify an account of a system we must know its inner functioning-just what is missing where explainable AI (XAI)

arXiv:2606.22091v1 Announce Type: new Abstract: Per-scene 3D Gaussian Splatting (3DGS) enables high-fidelity rendering, but practical robotic and AR scene capture pipelines often depend on external ge

All DOGE required was contact information of the recipients to confirm that funding was not fraudulent. No validated medical funding was stopped. Anything that appeared to be legitimate lifesaving fun

arXiv:2606.22218v1 Announce Type: cross Abstract: In this paper, we investigate untrained recurrent models from the Reservoir Computing (RC) paradigm for audio surveillance, focusing on bidirectional

Cristina Criddle / Financial Times: An interview with Nvidia VP of Healthcare Kimberly Powell on how AI can ease doctors' workloads, help address trained medical staff shortages, and more — The chipma

arXiv:2606.22089v1 Announce Type: new Abstract: Breast arterial calcification (BAC) on screening mammograms is an emerging cardiovascular risk biomarker, but quantitative use requires reproducible seg

arXiv:2606.21712v1 Announce Type: cross Abstract: Batch inference has become a central mode of AI computation, yet existing inference engines still rely on execution models designed for interactive se

DFlash is an open source block diffusion model for speculative decoding that significantly accelerates LLM inference on NVIDIA Blackwell GPUs by drafting entire token blocks in parallel and verifying

The NVIDIA BioNeMo Agent Toolkit enables AI scientists to use structure prediction, molecular generation, docking, sequence analysis, design, and genomics as callable tools , allowing AI agents to exe

.@cartesia runs one of the hardest inference workloads: real-time voice. Their stack has to keep long-lived streams moving, serve millions of audio minutes a day, and hold model latency around 90ms. T

China Xin Memory Technologies (CXMT) is positioning itself as a competitor to established DRAM manufacturers through domestic production capabilities and technological development. The article likely

arXiv:2606.22939v1 Announce Type: cross Abstract: Radio frequency jamming poses a critical threat to the availability of wireless Industrial Internet of Things (IIoT) networks. Existing detection and

arXiv:2606.23521v1 Announce Type: cross Abstract: Long-running LLM agents keep valuable state resident on GPUs: KV caches, request schedulers, communication state, and sometimes online adapters. Losin

Patrick Temple-West / Financial Times: Eli Lilly plans to use its $7.3B cash reserve to fund an “App Store” for biotech scientists after launching a data center with 1,016 Blackwell chips in 2025 — Mo

arXiv:2606.21448v1 Announce Type: new Abstract: As large language models scale, memory bandwidth for key-value caches and retrieval-augmented generation systems becomes a critical bottleneck. While 1-

arXiv:2606.22180v1 Announce Type: cross Abstract: Graph embedding maps graph nodes into low-dimensional vectors to support applications such as recommendation, fraud detection, and graph-based retriev

arXiv:2509.17390v2 Announce Type: replace Abstract: While 3D Gaussian Splatting (3DGS) has emerged as a strong representation for photorealistic rendering, its vast ecosystem of assets remains difficu

arXiv:2606.21222v1 Announce Type: new Abstract: Large-scale autonomous fleets rely on teleoperation to resolve rare failures, yet streaming raw sensor data from many vehicles is costly, and remote ope

Genesis Workbench is a collaborative framework developed by Databricks and NVIDIA designed to accelerate AI adoption in the life sciences industry by providing integrated tools and best practices. The

arXiv:2606.23419v1 Announce Type: new Abstract: Autoregressive decoding with LLMs is primarily bottlenecked by GPU memory bandwidth, especially in edge-computing settings. While quantization is essent

arXiv:2606.21633v1 Announce Type: new Abstract: Diffusion LLMs (dLLMs) improve GPU utilization over autoregressive decoding by generating multiple tokens per forward pass, but their KV cache still gro

Telecommunication operators deploy autonomous agents built on telecom-domain models and running inside secure execution runtimes connected to tools, digital twins, and shared skills to understand netw

i dont think anyone is correctly doing the math around how SpaceX, the NeoCloud+NeoLab, is currently going to market? SpaceX has already recouped about HALF its investment in Cursor, in compute deals.

LLMs write fast single-GPU kernels. Ask for a multi-GPU one and they fall apart. ParallelKernelBench (PKB) measures how they fail by benchmarking against 87 problems pulled from real codebases includi