Techme101.com

Nvidia today introduced its Earth-2 cloud platform at the GTC 2024 event in San Jose, California, aimed at advancing global climate change predictions through AI-powered simulations. According to VentureBeat, the platform, which was first mentioned in 2021, has now reached full operational status.
During a keynote address, Nvidia CEO Jensen Huang highlighted Earth-2’s role in addressing the frequent occurrence of climate disasters such as droughts, hurricanes, and floods. The Earth-2 platform, according to Huang, offers APIs that enable large-scale simulation and visualization of weather and climate data, providing a tool to better prepare for and mitigate the effects of extreme weather events.
The technology behind Nvidia’s Earth-2
Earth-2 relies on a combination of Nvidia’s CUDA-X microservices software and AI models, including the CorrDiff generative AI model. This integration allows for simulations that are both faster and more energy-efficient than traditional numerical models. Rev Lebaredian, Nvidia’s vice president of simulation, underscored the economic impact of extreme weather, which results in $140 billion in losses annually, and the need for detailed, kilometer-scale simulations to effectively address this challenge.
One of the first organizations to utilize Earth-2 is Taiwan’s Central Weather Administration, which aims to use the platform’s models to improve typhoon forecasting and evacuation planning. Additionally, Earth-2’s capabilities are being extended through integration with Nvidia Omniverse, enabling users to develop 3D workflows that incorporate actual weather data into their digital twin environments.
The Weather Company, an entity focused on weather data forecasting and insights, announced plans to integrate Earth-2 APIs to enhance their weather modeling products and services. This collaboration is intended to provide more accurate weather intelligence for enterprise clients and improve the overall understanding and simulation of weather impacts.
At the GTC 2024 conference, Nvidia also showcased other innovations such as the NIM software platform for streamlined AI deployment, and advancements in humanoid robotics with Project GR00T.
The post Nvidia debuts Earth-2 platform for enhanced climate forecasting appeared first on ReadWrite.

At its GTC conference today, Nvidia unveiled NIM, a revolutionary software platform designed to seamlessly integrate both custom and pre-trained AI models into production environments.
Alongside a number of announcements today at Nvidia’s GTC conference, NIM harnesses Nvidia’s expertise in AI model inferencing and optimization, offering a streamlined approach for developers. By merging AI models with an optimized inferencing engine and encapsulating them into containers accessible as microservices, NIM drastically reduces deployment time. According to TechCrunch reporting, what would traditionally take months can now be accomplished swiftly, bypassing the need for extensive in-house AI expertise.
This innovative platform supports models from notable entities such as NVIDIA, A121, and Getty Images, alongside open models from tech giants like Google and Meta. Nvidia’s collaboration with Amazon, Google, and Microsoft aims to integrate NIM microservices into major cloud services, enhancing accessibility for developers across the board.
NIM’s backbone: Nvidia’s inferencing engines
At the heart of NIM lies the Triton Inference Server, alongside TensorRT and TensorRT-LLM, underscoring Nvidia’s commitment to providing a robust foundation for AI applications. The platform also features specialized microservices, such as Riva for speech and translation adjustments, cuOpt for routing optimizations, and the Earth-2 model for simulations in weather and climate.
Manuvir Das, head of enterprise computing at Nvidia, emphasized the efficiency and enterprise-grade quality that NIM brings to the table, allowing developers to focus on building enterprise applications without the overhead of model management.
NIM stands as a testament to Nvidia’s vision of transforming enterprises into AI-driven entities, equipped with a suite of containerized AI microservices. With the backing of industry giants and an ecosystem of partners, Nvidia’s NIM is poised to revolutionize the way AI models are deployed and utilized across various sectors.
Jensen Huang, Nvidia’s CEO, highlighted the transformative potential of NIM, envisioning a future where every enterprise leverages AI to enhance their operations and innovation capacity.
The post Nvidia launches NIM to simplify AI model deployment appeared first on ReadWrite.

Nvidia’s ambition to lead in the realm of AI extends beyond its renowned chip technology, as CEO Jensen Huang expresses enthusiasm for developing foundation models for humanoid robots, according to a recent TechCrunch report. The technology giant is leveraging its influence in AI hardware to pioneer innovations in robotics, marked by the announcement of Project GR00T at its annual GTC developer conference. This initiative underscores Nvidia’s strategic move into the competitive sphere of humanoid robotics, hinting at its potential impact on future AI applications.
Project GR00T emerges as Nvidia’s response to the growing interest in humanoid robots, proposing a general-purpose foundation model aimed at supporting a wide array of companies in this sector, such as Agility Robotics, Boston Dynamics, and XPENG Robotics. This move signifies Nvidia’s commitment to facilitating advancements in robotics, offering a platform that promises to accelerate development across various applications, from industrial automation to daily human assistance.
According to TechCrunch, Jonathan Hurst of Agility Robotics and Geordie Rose of Sanctuary AI lauded the initiative, highlighting the transformative potential of human-centric robots like Digit in reshaping labor and addressing major societal challenges. Nvidia’s partnership approach, focusing on collaboration and shared innovation, sets a promising stage for the evolution of embodied AI.
Accompanying Project GR00T, Nvidia introduced Jetson Thor, a computing marvel designed for humanoid robotics applications. This new hardware, featuring Nvidia’s Blackwell architecture, emphasizes the company’s efforts to push the boundaries of AI and robotics performance, facilitating the development of more capable and versatile robots.
Nvidia also unveiled two additional programs: Isaac Manipulator and Isaac Perceptor, targeting advancements in robotic dexterity and vision processing. These initiatives reflect Nvidia’s comprehensive strategy to influence a broader range of robotic functionalities, from manufacturing automation to autonomous navigation.
The post Nvidia ventures into humanoid robotics with Project GR00T appeared first on ReadWrite.