named jelly ice, it contains 90 percent water and ten percent gelatin, has no synthetic polymers, and doesn’t melt. The post researchers turn to hardened gelatin as substitute for non-melting, reusable ice cubes appeared first on designboom | archite

AI reasoning, inference and networking will be top of mind for attendees of next week’s Hot Chips conference. A key forum for processor and system architects from industry and academia, Hot Chips — running Aug. 24-26 at Stanford University — showcases the latest innovations poised to advance AI factories and drive revenue for the trillion-dollar data center computing market. At the conference, NVIDIA will join industry leaders including Google and Microsoft in a “tutorial” session — taking place on Sunday, Aug. 24 — that discusses designing rack-scale architecture for data centers. In addition, NVIDIA experts will present at four sessions and one tutorial detailing how: NVIDIA networking, including the NVIDIA ConnectX-8 SuperNIC, delivers AI reasoning at rack- and data-center scale. (Featuring Idan Burstein, principal architect of network adapters and systems-on-a-chip at NVIDIA) Neural rendering advancements and massive leaps in inference — powered by the NVIDIA Blackwell architecture, including the NVIDIA GeForce RTX 5090 GPU — provide next-level graphics and simulation capabilities. (Featuring Marc Blackstein, senior director of architecture at NVIDIA) Co-packaged optics (CPO) switches with integrated silicon photonics — built with light-speed fiber rather than copper wiring to send information quicker and using less power — enable efficient, high-performance, gigawatt-scale AI factories. The talk will also highlight NVIDIA Spectrum-XGS Ethernet, a new scale-across technology for unifying distributed data centers into AI super-factories. (Featuring Gilad Shainer, senior vice president of networking at NVIDIA) The NVIDIA GB10 Superchip serves as the engine within the NVIDIA DGX Spark desktop supercomputer. (Featuring Andi Skende, senior distinguished engineer at NVIDIA) It’s all part of how NVIDIA’s latest technologies are accelerating inference to drive AI innovation everywhere, at every scale. NVIDIA Networking Fosters AI Innovation at Scale AI reasoning — when artificial intelligence systems can analyze and solve complex problems through multiple AI inference passes — requires rack-scale performance to deliver optimal user experiences efficiently. In data centers powering today’s AI workloads, networking acts as the central nervous system, connecting all the components — servers, storage devices and other hardware — into a single, cohesive, powerful computing unit. NVIDIA ConnectX-8 SuperNIC Burstein’s Hot Chips session will dive into how NVIDIA networking technologies — particularly NVIDIA ConnectX-8 SuperNICs — enable high-speed, low-latency, multi-GPU communication to deliver market-leading AI reasoning performance at scale. As part of the NVIDIA networking platform, NVIDIA NVLink, NVLink Switch and NVLink Fusion deliver scale-up connectivity — linking GPUs and compute elements within and across servers for ultra low-latency, high-bandwidth data exchange. NVIDIA Spectrum-X Ethernet provides the scale-out fabric to connect entire clusters, rapidly streaming massive datasets into AI models and orchestrating GPU-to-GPU communication across the data center. Spectrum-XGS Ethernet scale-across technology extends the extreme performance and scale of Spectrum-X Ethernet to interconnect multiple, distributed data centers to form AI super-factories capable of giga-scale intelligence. Connecting distributed AI data centers with NVIDIA Spectrum-XGS Ethernet. At the heart of Spectrum-X Ethernet, CPO switches push the limits of performance and efficiency for AI infrastructure at scale, and will be covered in detail by Shainer in his talk. NVIDIA GB200 NVL72 — an exascale computer in a single rack — features 36 NVIDIA GB200 Superchips, each containing two NVIDIA B200 GPUs and an NVIDIA Grace CPU, interconnected by the largest NVLink domain ever offered, with NVLink Switch providing 130 terabytes per second of low-latency GPU communications for AI and high-performance computing workloads. An NVIDIA rack-scale system. Built with the NVIDIA Blackwell architecture, GB200 NVL72 systems deliver massive leaps in reasoning inference performance. NVIDIA Blackwell and CUDA Bring AI to Millions of Developers The NVIDIA GeForce RTX 5090 GPU — also powered by Blackwell and to be covered in Blackstein’s talk — doubles performance in today’s games with NVIDIA DLSS 4 technology. NVIDIA GeForce RTX 5090 GPU It can also add neural rendering features for games to deliver up to 10x performance, 10x footprint amplification and a 10x reduction in design cycles,  helping enhance realism in computer graphics and simulation. This offers smooth, responsive visual experiences at low energy consumption and improves the lifelike simulation of characters and effects. NVIDIA CUDA, the world’s most widely available computing infrastructure, lets users deploy and run AI models using NVIDIA Blackwell anywhere. Hundreds of millions of GPUs run CUDA across the globe, from NVIDIA GB200 NVL72 rack-scale systems to GeForce RTX– and NVIDIA RTX PRO-powered PCs and workstations, with NVIDIA DGX Spark powered by NVIDIA GB10 — discussed in Skende’s session — coming soon. From Algorithms to AI Supercomputers — Optimized for LLMs NVIDIA DGX Spark Delivering powerful performance and capabilities in a compact package, DGX Spark lets developers, researchers, data scientists and students push the boundaries of generative AI right at their desktops, and accelerate workloads across industries. As part of the NVIDIA Blackwell platform, DGX Spark brings support for NVFP4, a low-precision numerical format to enable efficient agentic AI inference, particularly of large language models (LLMs). Learn more about NVFP4 in this NVIDIA Technical Blog. Open-Source Collaborations Propel Inference Innovation NVIDIA accelerates several open-source libraries and frameworks to accelerate and optimize AI workloads for LLMs and distributed inference. These include NVIDIA TensorRT-LLM, NVIDIA Dynamo, TileIR, Cutlass, the NVIDIA Collective Communication Library and NIX — which are integrated into millions of workflows. Allowing developers to build with their framework of choice, NVIDIA has collaborated with top open framework providers to offer model optimizations for FlashInfer, PyTorch, SGLang, vLLM and others. Plus, NVIDIA NIM microservices are available for popular open models like OpenAI’s gpt-oss and Llama 4,  making it easy for developers to operate managed application programming interfaces with the flexibility and security of self-hosting models on their preferred infrastructure. Learn more about the latest advancements in inference and accelerated computing by joining NVIDIA at Hot Chips.  

Japan is once again building a landmark high-performance computing system — not simply by chasing speed, but by rethinking how technology can best serve the nation’s most urgent scientific needs. At the FugakuNEXT International Initiative Launch Ceremony held in Tokyo on Aug. 22, leaders from RIKEN, Japan’s top research institute, announced the start of an international collaboration with Fujitsu and NVIDIA to co-design FugakuNEXT, the successor to the world-renowned supercomputer, Fugaku. Awarded early in the process, the contract enables the partners to work side by side in shaping the system’s architecture to address Japan’s most critical research priorities — from earth systems modeling and disaster resilience to drug discovery and advanced manufacturing. More than an upgrade, the effort will highlight Japan’s embrace of modern AI and showcase Japanese innovations that can be harnessed by researchers and enterprises across the globe. The ceremony featured remarks from the initiative’s leaders, RIKEN President Makoto Gonokami and Satoshi Matsuoka, director of the RIKEN Center for Computational Science and one of Japan’s most respected high-performance computing architects. Fujitsu Chief Technology Officer Vivek Mahajan attended, emphasizing the company’s role in advancing Japan’s computing capabilities. Ian Buck, vice president of hyperscale and high-performance computing at NVIDIA, attended in person as well to discuss the collaborative design approach and how the resulting platform will serve as a foundation for innovation well into the next decade. Momentum has been building. When NVIDIA founder and CEO Jensen Huang touched down in Tokyo last year, he called on Japan to seize the moment — to put NVIDIA’s latest technologies to work building its own AI, on its own soil, with its own infrastructure. FugakuNEXT answers that call, drawing on NVIDIA’s whole software stack —  from NVIDIA CUDA-X libraries such as NVIDIA cuQuantum for quantum simulation, RAPIDS for data science, NVIDIA TensorRT for high-performance inference and NVIDIA NeMo for large language model development, to other domain-specific software development kits tailored for science and industry. Innovations pioneered on FugakuNEXT could become blueprints for the world. What’s Inside FugakuNEXT will be a hybrid AI-HPC system, combining simulation and AI workloads. It will feature FUJITSU-MONAKA-X CPUs, which can be paired with NVIDIA technologies using NVLink Fusion, new silicon enabling high-bandwidth connections between Fujitsu’s CPUs and NVIDIA’s architecture. The system will be built for speed, scale and efficiency. What It Will Do FugakuNEXT will support a wide range of applications — such as automating hypothesis generation, code creation and experiment simulation. Scientific research: Accelerating simulations with surrogate models and physics-informed neural networks. Manufacturing: Using AI to learn from simulations to generate efficient and aesthetically pleasing designs faster than ever before. Earth systems modeling: aiding disaster preparedness and prediction for earthquakes and severe weather, and more. RIKEN, Fujitsu and NVIDIA will collaborate on software developments, including tools for mixed-precision computing, continuous benchmarking, and performance optimization. FugakuNEXT isn’t just a technical upgrade — it’s a strategic investment in Japan’s future. Backed by Japan’s MEXT (Ministry of Education, Culture, Sports, Science and Technology), it will serve universities, government agencies, and industry partners nationwide. It marks the start of a new era in Japanese supercomputing — one built on sovereign infrastructure, global collaboration, and a commitment to scientific leadership. Image courtesy of RIKEN

The scramble by tech companies for top AI talent is using unorthodox methods that imperil Silicon Valley’s startup culture.

Of around 7,000 languages in the world, a tiny fraction are supported by AI language models. NVIDIA is tackling the problem with a new dataset and models that support the development of high-quality speech recognition and translation AI for 25 European languages — including languages with limited available data like Croatian, Estonian and Maltese. These tools will enable developers to more easily scale AI applications to support global users with fast, accurate speech technology for production-scale use cases such as multilingual chatbots, customer service voice agents and near-real-time translation services. They include: Granary, a massive, open-source corpus of multilingual speech datasets that contains around a million hours of audio, including nearly 650,000 hours for speech recognition and over 350,000 hours for speech translation. NVIDIA Canary-1b-v2, a billion-parameter model trained on Granary for high-quality transcription of European languages, plus translation between English and two dozen supported languages. NVIDIA Parakeet-tdt-0.6b-v3, a streamlined, 600-million-parameter model designed for real-time or large-volume transcription of Granary’s supported languages. The paper behind Granary will be presented at Interspeech, a language processing conference taking place in the Netherlands, Aug. 17-21. The dataset, as well as the new Canary and Parakeet models, are now available on Hugging Face. How Granary Addresses Data Scarcity To develop the Granary dataset, the NVIDIA speech AI team collaborated with researchers from Carnegie Mellon University and Fondazione Bruno Kessler. The team passed unlabeled audio through an innovative processing pipeline powered by NVIDIA NeMo Speech Data Processor toolkit that turned it into structured, high-quality data. This pipeline allowed the researchers to enhance public speech data into a usable format for AI training, without the need for resource-intensive human annotation. It’s available in open source on GitHub. With Granary’s clean, ready-to-use data, developers can get a head start building models that tackle transcription and translation tasks in nearly all of the European Union’s 24 official languages, plus Russian and Ukrainian. For European languages underrepresented in human-annotated datasets, Granary provides a critical resource to develop more inclusive speech technologies that better reflect the linguistic diversity of the continent — all while using less training data. The team demonstrated in their Interspeech paper that, compared to other popular datasets, it takes around half as much Granary training data to achieve a target accuracy level for automatic speech recognition (ASR) and automatic speech translation (AST). Tapping NVIDIA NeMo to Turbocharge Transcription The new Canary and Parakeet models offer examples of the kinds of models developers can build with Granary, customized to their target applications. Canary-1b-v2 is optimized for accuracy on complex tasks, while parakeet-tdt-0.6b-v3 is designed for high-speed, low-latency tasks. By sharing the methodology behind the Granary dataset and these two models, NVIDIA is enabling the global speech AI developer community to adapt this data processing workflow to other ASR or AST models or additional languages, accelerating speech AI innovation. Canary-1b-v2, available under a permissive license, expands the Canary family’s supported languages from four to 25. It offers transcription and translation quality comparable to models 3x larger while running inference up to 10x faster. https://blogs.nvidia.com/wp-content/uploads/2025/08/Canary-demo.mp4 NVIDIA NeMo, a modular software suite for managing the AI agent lifecycle, accelerated speech AI model development. NeMo Curator, part of the software suite, enabled the team to filter out synthetic examples from the source data so that only high-quality samples were used for model training. The team also harnessed the NeMo Speech Data Processor toolkit for tasks like aligning transcripts with audio files and converting data into the required formats. Parakeet-tdt-0.6b-v3 prioritizes high throughput and is capable of transcribing 24-minute audio segments in a single inference pass. The model automatically detects the input audio language and transcribes without additional prompting steps. Both Canary and Parakeet models provide accurate punctuation, capitalization and word-level timestamps in their outputs. Read more on GitHub and get started with Granary on Hugging Face.

NVIDIA is partnering with the U.S. National Science Foundation (NSF) to create an AI system that supports the development of multimodal language models for advancing scientific research in the United States. The partnership supports the NSF Mid-Scale Research Infrastructure project, called Open Multimodal AI Infrastructure to Accelerate Science (OMAI). “Bringing AI into scientific research has been a game changer,” said Brian Stone, performing the duties of the NSF director. “NSF is proud to partner with NVIDIA to equip America’s scientists with the tools to accelerate breakthroughs. These investments are not just about enabling innovation; they are about securing U.S. global leadership in science and technology and tackling challenges once thought impossible.” OMAI, part of the work of the Allen Institute for AI, or Ai2, aims to build a national fully open AI ecosystem to drive scientific discovery through AI, while also advancing the science of AI itself. NVIDIA’s support of OMAI includes providing NVIDIA HGX B300 systems — state-of-the-art AI infrastructure built to accelerate model training and inference with exceptional efficiency — along with the NVIDIA AI Enterprise software platform, empowering OMAI to transform massive datasets into actionable intelligence and breakthrough innovations. NVIDIA HGX B300 systems are built with NVIDIA Blackwell Ultra GPUs and feature industry-leading high-bandwidth memory and interconnect technologies to deliver groundbreaking acceleration, scalability and efficiency to run the world’s largest models and most demanding workloads. “AI is the engine of modern science — and large, open models for America’s researchers will ignite the next industrial revolution,” said Jensen Huang, founder and CEO of NVIDIA. “In collaboration with NSF and Ai2, we’re accelerating innovation with state-of-the-art infrastructure that empowers U.S. scientists to generate limitless intelligence, making it America’s most powerful and renewable resource.” The contributions will support research teams from the University of Washington, the University of Hawaii at Hilo, the University of New Hampshire and the University of New Mexico. The public-private partnership investment in U.S. technology aligns with recent initiatives outlined by the White House AI Action Plan, which supports America’s global AI leadership. “The models are part of the national research infrastructure — but we can’t build the models without compute, and that’s why NVIDIA is so important to this project,” said Noah Smith, senior director of natural language processing research at Ai2. Opening Language Models to Advance American Researchers  Driving some of the fastest-growing applications in history, today’s large language models (LLMs) have many billions of parameters, or internal weights and biases learned in training. LLMs are trained on trillions of words, and multimodal LLMs can ingest images, graphs, tables and more. But the power of these so-called frontier models can sometimes be out of reach for scientific research when the parameters, training data, code and documentation are not openly available. “With the model training data in hand, you have the opportunity to trace back to particular training instances similar to a response, and also more systematically study how emerging behaviors relate to the training data,” said Smith. NVIDIA’s partnership with NSF to support Ai2’s OMAI initiative provides fully open model access to data, open-source data interrogation tools to help refine datasets, as well as documentation and training for early-career researchers — advancing U.S. global leadership in science and engineering. The Ai2 project — supported by NVIDIA technologies — pledges to make the software and models available at low or zero cost to researchers, similar to open-source code repositories and science-oriented digital libraries. It’s in line with Ai2’s previous work in creating fully open language models and multimodal models, maximizing access. Driving U.S. Global Leadership in Science and Engineering  “Winning the AI Race: America’s AI Action Plan” was announced in July by the White House, supported with executive orders to accelerate federal permitting of data center infrastructure and promote exportation of the American AI technology stack. The OMAI initiative aligns with White House AI Action Plan priorities, emphasizing the acceleration of AI-enabled science and supporting the creation of leading open models to enhance America’s global AI leadership in academic research and education.