News from the AI & ML world
DeeperML - #nvidiaai
|
staff@insideAI News
//
Saudi Arabia is making major strides in artificial intelligence, unveiling deals with several leading U.S. technology firms including NVIDIA, AMD, Cisco, and Amazon Web Services. These partnerships are primarily formed through HUMAIN, the AI subsidiary of Saudi Arabia’s Public Investment Fund (PIF), which controls about $940 billion in assets. As part of these collaborations, Saudi Arabia’s Crown Prince Mohammed bin Salman has launched ‘Humain’ with the intent of establishing the kingdom as a global leader in artificial intelligence. This initiative aligns with the Kingdom’s Vision 2030 plan to diversify its economy and reduce dependence on oil revenues.
NVIDIA has partnered with HUMAIN to construct AI factories in Saudi Arabia. The partnership underscores HUMAIN’s mission to position Saudi Arabia as an international AI powerhouse, and will have a projected capacity of up to 500 megawatts. The initial phase includes the deployment of 18,000 NVIDIA GB300 Grace Blackwell AI supercomputers with NVIDIA InfiniBand networking. AMD has also signed an agreement with HUMAIN where the parties will invest up to $10 billion to deploy 500 megawatts of AI compute capacity over the next five years. In addition to chip manufacturers, networking and cloud service providers are also involved. Cisco will partner with HUMAIN AI enterprise to power AI infrastructure and ecosystem growth, with new investments in research, talent, and digital skills. Amazon Web Services (AWS) and HUMAIN plan to invest over $5 billion to build an “AI Zone” in the kingdom, incorporating dedicated AWS AI infrastructure and services. These efforts are supported by the U.S. government easing AI chip export rules to Gulf states, which had previously limited the access of such countries to high-end AI chips. Recommended read:
References :
Isha Salian@NVIDIA Blog
//
References:
developer.nvidia.com
, BigDATAwire
Nvidia is pushing the boundaries of artificial intelligence with a focus on multimodal generative AI and tools to enhance AI model integration. Nvidia's research division is actively involved in advancing AI across various sectors, underscored by the presentation of over 70 research papers at the International Conference on Learning Representations (ICLR) in Singapore. These papers cover a diverse range of topics including generative AI, robotics, autonomous driving, and healthcare, demonstrating Nvidia's commitment to innovation across the AI spectrum. Bryan Catanzaro, vice president of applied deep learning research at NVIDIA, emphasized the company's aim to accelerate every level of the computing stack to amplify the impact and utility of AI across industries.
Research efforts at Nvidia are not limited to theoretical advancements. The company is also developing tools that streamline the integration of AI models into real-world applications. One notable example is the work being done with NVIDIA NIM microservices, which are being leveraged by researchers at the University College London (UCL) Deciding, Acting, and Reasoning with Knowledge (DARK) Lab to benchmark agentic LLM and VLM reasoning for gaming. These microservices simplify the deployment and scaling of AI models, enabling researchers to efficiently handle workloads of any size and customize models for specific needs. Nvidia's NIM microservices are designed to redefine how researchers and developers deploy and scale AI models, offering a streamlined approach to harnessing the power of GPUs. These microservices simplify the process of running AI inference workloads by providing pre-optimized engines such as NVIDIA TensorRT and NVIDIA TensorRT-LLM, which deliver low-latency, high-throughput performance. The microservices also offer easy and fast API integration with standard frontends like the OpenAI API or LangChain for Python environments. Recommended read:
References :
@techstrong.ai
//
Nvidia has unveiled new tools and capabilities designed to streamline AI deployment and enhance efficiency for enterprises. A key component of this release is the general availability of NVIDIA NeMo microservices, empowering companies to construct AI agents that leverage data flywheels to improve employee productivity. These microservices provide an end-to-end platform for developers, allowing them to create and continuously optimize state-of-the-art agentic AI systems through the integration of inference, business, and user feedback data. The company also highlighted the importance of maintaining a constant stream of high-quality inputs to ensure the accuracy, relevance, and timeliness of AI agents.
Nvidia is also introducing the G-Assist plug-in builder, a tool enabling customization of AI on GeForce RTX AI PCs. This plug-in builder expands the functionality of G-Assist by allowing users to add new commands and connect external tools, which can range from large language models to simple functions like controlling music. Developers can use coding languages like JSON and Python to create tools integrated into G-Assist, and they can submit plug-ins for review and potential inclusion in the NVIDIA GitHub repository, making new capabilities available to others. G-Assist can be modified to perform different actions with different LLMs and software, gaming-related or not. Researchers at the University College London (UCL) Deciding, Acting, and Reasoning with Knowledge (DARK) Lab are leveraging NVIDIA NIM microservices in their new game-based benchmark suite, BALROG, designed to evaluate the agentic capabilities of models on challenging, long-horizon interactive tasks. By using NVIDIA NIM, the DARK lab was able to accelerate their benchmarking process, deploying and hosting the DeepSeek-R1 model without needing to do so locally. This showcases the flexibility and efficiency of NIM microservices, which can be deployed across various platforms, including cloud environments, data centers, and local workstations, enabling seamless integration into diverse workflows and handling workloads of any size. Recommended read:
References :
@the-decoder.com
//
Nvidia is significantly expanding its presence in the AI hardware landscape, announcing plans to mass produce AI supercomputers in Texas and shifting more of its AI production to the United States. This initiative includes building its own factories in the U.S., marking a notable shift in strategy. Working with manufacturing partners like Foxconn in Houston and Wistron in Dallas, Nvidia has commissioned over one million square feet of production space. This move is fueled by a planned investment of up to $500 billion in U.S. AI infrastructure over the next four years. Blackwell AI chips have already begun production at TSMC's facilities in Phoenix, Arizona, indicating rapid progress towards realizing these ambitious goals.
Mass production at the Texas plants is expected to ramp up within the next 12 to 15 months. Nvidia's commitment to US-based production involves partnerships with companies like Amkor and SPIL for packaging and testing operations in Arizona, creating a complex AI chip and supercomputer supply chain. These AI supercomputers are seen as engines for a new type of data center designed specifically for AI processing, referred to as "AI factories." Nvidia anticipates the construction of dozens of these "gigawatt AI factories" in the coming years, further solidifying its role in the rapidly expanding AI industry. In addition to its supercomputing endeavors, Nvidia is intensifying its push into the robotics sector with the launch of its Jetson Thor platform, slated for the first half of 2025. This compact computing platform is specifically designed for humanoid robots, positioning Nvidia to capitalize on the potentially lucrative future of the robotics industry. While Google is also active in this space, Nvidia aims to offer a comprehensive technology platform for "physical AI," encompassing both the necessary software and robotic chips. Their existing Jetson Orin series chips have already seen adoption by humanoid robot manufacturers, and the upcoming Jetson Thor promises further enhanced capabilities. Recommended read:
References :
@the-decoder.com
//
Nvidia is making significant advancements in artificial intelligence, showcasing innovations in both hardware and video generation. A new method developed by Nvidia, in collaboration with Stanford University, UCSD, UC Berkeley, and UT Austin, allows for the creation of AI-generated videos up to one minute long. This breakthrough addresses previous limitations in video length, where models like OpenAI's Sora, Meta's MovieGen, and Google's Veo 2 were capped at 20, 16, and 8 seconds respectively.
The key innovation lies in the introduction of Test-Time Training layers (TTT layers), which are integrated into a pre-trained Transformer architecture. These layers replace simple hidden states in conventional Recurrent Neural Networks (RNNs) with small neural networks that continuously learn during the video generation process. This allows the system to maintain consistency across longer sequences, ensuring elements like characters and environments remain stable throughout the video. This new method has even been showcased with an AI-generated "Tom and Jerry" cartoon. Furthermore, Nvidia has unveiled its new Llama-3.1 Nemotron Ultra large language model (LLM), which outperforms DeepSeek R1 despite having less than half the parameters. The Llama-3.1-Nemotron-Ultra-253B is a 253-billion parameter model designed for advanced reasoning, instruction following, and AI assistant workflows. Its architecture includes innovations such as skipped attention layers, fused feedforward networks, and variable FFN compression ratios. The model's code is publicly available on Hugging Face, reflecting Nvidia's commitment to open-source AI development. Recommended read:
References :
James McKenna@NVIDIA Newsroom
//
References:
NVIDIA Newsroom
, NVIDIA Technical Blog
NVIDIA's Omniverse platform is gaining traction within industrial ecosystems as companies leverage digital twins to train physical AI. The Mega NVIDIA Omniverse Blueprint, now available in preview, empowers industrial enterprises to accelerate the development, testing, and deployment of physical AI. This blueprint provides a reference workflow for combining sensor simulation and synthetic data generation, enabling the simulation of complex human-robot interactions and verification of autonomous systems within industrial digital twins.
At Hannover Messe, leaders from manufacturing, warehousing, and supply chain sectors are showcasing their adoption of the blueprint to simulate robots like Digit from Agility Robotics. They are also demonstrating how industrial AI and digital twins can be used to optimize facility layouts, material flow, and collaboration between humans and robots. NVIDIA ecosystem partners like Delta Electronics, Rockwell Automation, and Siemens are also announcing further integrations with NVIDIA Omniverse and NVIDIA AI technologies at the event, further solidifying Omniverse's role in physical AI development. Recommended read:
References :
Hassan Shittu@Fello AI
//
References:
Fello AI
, lambdalabs.com
Nvidia is making significant strides in healthcare and AI infrastructure, particularly through the development of specialized large language models (LLMs). Their DNA LLM exemplifies this, aiming to revolutionize genomic research and drug discovery. This highlights AI's potential to transform medical science by enabling faster analysis and interpretation of biological data.
Lambda has been recognized as NVIDIA's 2025 Healthcare Partner of the Year for accelerating AI innovation in healthcare and biotech. John Snow Labs introduced the first commercially available Medical Reasoning LLM at NVIDIA GTC, optimized for clinical reasoning and capable of verbalizing its thought processes. Nvidia's involvement in this has helped lead the way for these healthcare specific Large Language Models. Recommended read:
References :
Cierra Choucair@The Quantum Insider
//
Nvidia CEO Jensen Huang unveiled the company's latest advancements in AI and quantum computing at GTC 2025, emphasizing a clear roadmap for data center computing, AI reasoning, robotics, and autonomous vehicles. The centerpiece was the Blackwell platform, now in full production, boasting a 40x performance leap over its predecessor, Hopper, crucial for inference workloads. Nvidia is also countering the DeepSeek efficiency challenge, with focus on the Rubin AI chips slated for late 2026.
Nvidia is establishing the NVIDIA Accelerated Quantum Research Center (NVAQC) in Boston to integrate quantum hardware with AI supercomputers. The center will collaborate with industry leaders and top universities to address quantum computing challenges. NVAQC is set to begin operations later this year, supporting the broader quantum ecosystem by accelerating the transition from experimental to practical quantum computing. NVAQC will employ the NVIDIA GB200 NVL72 systems and CUDA-Q platform to power research on quantum simulations, hybrid quantum algorithms, and AI-driven quantum applications. Recommended read:
References :
Ellie Ramirez-Camara@Data Phoenix
//
Nvidia is making significant strides in the realm of AI agents, highlighted at this year's GTC 2025 conference. CEO Jensen Huang emphasized the transformative impact of agentic AI and reasoning models, predicting that these technologies will revolutionize industries and automate processes. To support this shift, Nvidia unveiled the Blackwell Ultra platform, designed to handle the demanding requirements of AI reasoning, agentic AI, and physical AI applications. The platform, which includes the GB300 NVL72 rack-scale solution and the HGX B300 NVL16 system, offers substantial performance improvements over previous generations, with the GB300 NVL72 delivering 1.5x more AI performance.
In addition to hardware advancements, Nvidia launched NVIDIA Dynamo, an open-source inference framework, to optimize reasoning AI services across thousands of GPUs. This framework is designed to maximize token revenue generation for AI factories deploying reasoning AI models by orchestrating and accelerating inference communication across GPU clusters. Major cloud providers and server manufacturers are expected to offer Blackwell Ultra-based products starting in the second half of 2025. These developments position Nvidia as a key player in the emerging landscape of AI agents and reasoning models, promising to drive significant advancements in AI capabilities and applications. Recommended read:
References :
Walter Sun@SAP News Center
//
References:
SAP News Center
SAP and NVIDIA are deepening their collaboration to deliver advanced AI capabilities to businesses worldwide. SAP is integrating NVIDIA's Llama Nemotron reasoning models to enhance the reasoning capabilities of its AI agents. This strategic move aims to improve accuracy and equip AI agents with advanced decision-making and execution skills. By incorporating these models, SAP's Joule agents will be better equipped to handle complex business challenges through deeper contextual reasoning and seamless interaction with enterprise data and systems, fostering more intelligent and autonomous operations.
The partnership between SAP and NVIDIA has already yielded AI innovations transforming business operations. SAP Joule for Consultants, enhanced with NVIDIA NeMo Retriever microservices, enables consultants to quickly access relevant insights from SAP-exclusive content, reducing time spent on documentation and troubleshooting. For developers, Joule for developers, powered by NVIDIA NIM microservices, accelerates ABAP code generation, improving code quality and accelerating innovation. These advancements are revolutionizing how enterprises implement SAP solutions and develop applications on SAP Business Technology Platform. Recommended read:
References :
|
BenchmarksBlogsResearch Tools |