AI is transforming the way that we understand and treat diseases. And nowhere is this more evident than in how we study proteins. It used to take months of lab experiments to determine the structure of a protein.With the release of AlphaFold 2 in 2021. That process can be reduced to minutes with deep learning. This capability can now fit on your desk.The NVIDIA DGX Spark brings data center class performance to protein AI, powered by the Grace Blackwell architecture, with up to one Petaflop of compute and 12 ...

Industrial AI Adoption & Market Opportunity - Industrial AI 正在经历极端的采用 [1] - NVIDIA 认为物理 AI 和工业 AI 是一个 10 万亿美元的行业 [2] - 工业 AI 不仅仅是数字孪生，它贯穿整个汽车生命周期，从前期设计协作到性能数字孪生 [4] - 工业 AI 被注入到数字孪生的创建中 [14] - 工业 AI 的创建从两个角度考虑：使用 Omniverse 创建工厂或仓库的虚拟版本，并使用 Cosmos 创建无限的变体 [15][16] Digital Twin Transformation - 行业正在从规划数字孪生转型到运营数字孪生，并注入 AI [5][8] - 运营数字孪生中，AI 可用于预测性维护，实时零件规划和机器人路径规划 [7] - 通过 AI 协同，用户可以通过提问来监控生产量，生产线问题等 [12][13] Mobility & Virtual Experience - 使用虚拟配置器查看不同的颜色和装饰组合 [9] - 能够实时体验车辆在不同环境条件下的性能，并最终通过 XR 体验坐在车辆中 [10] - 可以在 XR 中体验自动驾驶的未来 [10] AI-Powered Tools & Platforms - Metropolis 平台连接数千个摄像头，以帮助识别数字孪生如何实时运行 [22] - Isaac 和 Groot 平台用于机器人开发，帮助训练和测试 AMR，AGV，ARM 和人形机器人 [22] - Blueprint 旨在解决查找符合特定场景的数据，创建无限变体并基于数据训练模型的难题 [20][21] Real-World Applications & Partnerships - 已经与 BMW，Keon，Kanti，Rockwell Automation 和 Foxconn 合作实现了数字孪生 [5][25][26] - 与 Accenture 合作，将技术应用于 Keon 的仓库，实时感知数字孪生的每个操作 [24] - 设想与 Service Now 建立深度合作伙伴关系，在工厂车间触发实时警报，从而调试机器人或人员来处理事件 [19]

Modeling and simulation are core pillars of science and engineering workflows. AI surrogate models can enable nearrealtime simulation workflows. To develop such models, it's critical to integrate domain knowledge.But building these knowledge guided models isn't easy. Training a model only on available data restricts its usability as it has very limited generalizability. Existing frameworks also don't cater to approaches that fuse data with domain knowledge.NVIDIA Physics Nemo is an open-source AI physics pl ...

NVIDIA's AI Solutions for Automotive - NVIDIA pioneers accelerated computing, assisting companies with AI transformation, particularly in the automotive sector, collaborating with almost all automakers [1] - NVIDIA's solutions span various automotive GenAI use cases, including enterprise transformation, dealership support, in-cabin experience enhancement, self-driving technology, and design/manufacturing improvements [2] - NVIDIA provides an AI factory encompassing compute, software, and tools to handle diverse AI verticals like generative AI, physical AI, and HPC workflows [4] - NVIDIA's AI factory offers optimized use cases with a lower Total Cost of Ownership (TCO) [5] Enterprise AI Applications - Only 1% of enterprises have mature AI deployments, indicating a significant adoption challenge that NVIDIA aims to address with its ecosystem of partners [4] - NVIDIA is helping companies set up IT hubs for employee efficiency, connected vehicle data management, factory floor optimization, and code/test case generation [5] - NVIDIA assists customers with the V-model for design and development, introducing AI agents to improve requirements management [6][7][8] - NVIDIA partners with ServiceNow to introduce AI agents on the factory floor, cutting resolution time from hours to minutes, potentially saving customers hundreds of thousands of dollars per minute [8][9][10][11][12] In-Vehicle AI Applications - China is leading in in-vehicle AI innovation, focusing on personalization, multimodality, companion features, and sentry mode [13][14] - NVIDIA provides solutions for developing AI agents that operate both inside the vehicle and in the cloud, enabling features like predictive maintenance and roadside assistance [16][17][18][19] - NVIDIA offers full-stack platforms, including CUDA, to ensure seamless transition and migration from cloud to car, allowing customers to differentiate their end products [21][22] - Ford is using AI agents for design (V-model), contact center customer service, and in-vehicle repair manuals, demonstrating significant time savings (hours to seconds) [26][27]

Proteins are the machines of life. Understanding how proteins fold and interact with other molecules is a key part of understanding humanity's most pressing biological challenges. From designing new drugs to growing safer, healthier food, and even designing new cosmetics, resolving the 3D structure of proteins once took years in the lab.AI now makes this possible in minutes. With the NVIDIA RTX Pro 6000 Blackwell Server Edition, protein structure inference is over 100 times faster than before, unlocking the ...

AI has truly crossed the chasm from back office proof of concepts to full mainstream production. And the way that AI is getting deployed and used is through inference. So when we think about inference, it's actually two workloads.There's the context processing which we call prefill and then there's the generation of tokens which we call decode. In order to serve those two disparate processes more efficiently, the industry has developed disagregated serving. It's a method by which you can basically separate ...

General Overview - The document is a musical piece or a reference to music [1]

NVIDIA accelerated computing starts with DGX, the world's AI supercomputer, the engine behind the large language model breakthrough. IHand delivered the world's first DGX to open AI. Since then, half of the Fortune 100 companies have installed DGX AI supercomputers. DGX has become the essential instrument of AI. The GPU of DGX is eight H100 modules.H100 has a transformer engine designed to process models like the amazing chat GPT which stands for generative pre-trained transformers. The eight H100 modules a ...

Watch NVIDIA’s Quantum-X Photonics switch come to life in an #AIFactory. The NVIDIA Quantum-X Co-Packaged Optics (#CPO) Q3450 switch and ConnectX-8 SuperNICs connect NVIDIA’s GB300 racks with OSFP pluggable optical modules demonstrating NVIDIA's scale-out topology using NVIDIA #SiliconPhotonics, the world’s most advanced networking solution for the era of #AgenticAI. Learn more: https://www.nvidia.com/silicon-photonics/ https://youtu.be/kS8r7UcexJU?si=HOt4QJ8by1EYjVUB ...

AI Model & Infrastructure Demands - AI 模型在规模和复杂性上不断增长，对 AI 和数据中心基础设施提出了更高的要求 [1] - 行业需要在各个层面进行创新，包括硬件平台和算法方面 [1] New Technology & Performance - 公司推出了一种新的 FP4 格式，称为 MVFP4，旨在提供比 MXFP4 更高的性能，并保持与 FPA 或更高精度相同的水平 [2] - MVFP4 允许使用更少的能源和空间，并传输更少的数据，从而更快地训练模型 [2] - GB300 平台将提供比当前基于 Hopper 的平台高 7 倍的性能 [3] Industry Impact & Collaboration - 该技术对整个行业具有重大影响，公司正在与 AWS、Perplexity、OpenAI 等合作 [3] - 性能的提升将为在每个科学领域和每个 AI 学科中提高效率和性能开辟一个全新的机会 [3]