Nvidia and Alphabet Join Forces to Advance the Future of Agentic and Physical AI
New Collaborations between Nvidia and Google
Nvidia and Google, under the Alphabet umbrella, have announced exciting new initiatives that aim to push the boundaries of artificial intelligence (AI). With such a long-standing partnership, both companies are focused on making AI more accessible and applicable across various industries, including healthcare, manufacturing, and energy.
AI Development and Research Collaboration
Engineers and researchers from both Nvidia and Alphabet are collaborating closely to address significant challenges using AI and simulations. Their joint efforts include creating robots with improved manipulation skills, enhancing drug discovery processes, and optimizing energy grids. They plan to showcase key achievements at the Nvidia GTC global AI conference, utilizing Nvidia’s innovative platforms like Omniverse, Cosmos, and Isaac.
To support these initiatives, Google Cloud is set to be an early adopter of Nvidia’s latest technologies, such as the GB300 NVL72 rack-scale solution and the RTX PRO 6000 Blackwell Server Edition GPU. These tools will bolster Google Cloud’s AI production capabilities, enabling clients to leverage advanced AI technologies effectively.
Commitment to Responsible AI and Transparency
A major focus for Google DeepMind and Nvidia is to foster trust in generative AI. This initiative involves the implementation of SynthID, a technology that embeds digital watermarks within AI-generated content like images, audio, and videos. This approach aims to preserve the authenticity of generated materials, while also guarding against potential misinformation. SynthID will be used for the first time by Nvidia, integrating seamlessly with their foundational models from Nvidia Cosmos.
Moreover, the partnership also aims to optimize Google’s Gemma models to work more effectively on Nvidia’s GPUs. The introduction of Gemma 3 marks a significant step in open-source innovation in AI, making it easier for developers to harness its capabilities.
Evolution of Intelligent Robotics
Nvidia and Alphabet’s Intrinsic, an AI company focused on robotics, are working to make robotic systems smarter and more adaptive. Traditionally, industrial robots required extensive manual programming, which can be a lengthy and costly process. The teams are now developing more intuitive workflows to support Nvidia’s Isaac Manipulator, bringing universal grasping capabilities to robots.
During the GTC event, Intrinsic will unveil its OpenUSD framework, allowing for real-time visualization of robotic operations across various platforms.
In a further collaboration, Nvidia and Google DeepMind are teaming up with Disney Research to develop Newton, a physics engine that uses Nvidia’s Warp framework. This innovation should significantly accelerate machine learning workloads for robotics, boosting efficiency immensely.
Tackling Real-World Challenges through AI
Isomorphic Labs, led by Demis Hassabis of Google DeepMind, is transforming drug discovery by employing AI technologies. With a robust drug design engine housed on Google Cloud and powered by Nvidia’s GPUs, the labs aim to create AI models that can potentially revolutionize healthcare and medicine.
Another initiative, known as Tapestry, aims to enhance the electrical grid using AI. This project is focused on developing products that make the grid greener and more reliable. The collaboration aims to improve simulations of the electric grid, particularly in accommodating new energy sources and expanding capacity to meet increasing demands.
Cutting-Edge AI-Optimized Infrastructure
Building on their commitment to advanced AI infrastructure, Google Cloud will offer Nvidia’s latest Blackwell GPU instances, including the GB300 NVL72 and the RTX PRO 6000. The new Blackwell architecture is designed to provide superior AI performance, significantly outpacing previous models.
With the launch of virtual machine instances last month, Google Cloud has positioned itself as a leading provider of Nvidia’s cutting-edge technology. The platform has also been optimized for popular open-source frameworks such as JAX and MaxText, enabling efficient model training across vast GPU networks.
Overall, the partnerships and collaborations between Nvidia and Google demonstrate their dedication to advancing AI technology while addressing real-world problems, promising a more innovative and connected future.
