Nvidia's Cosmos, trained on 20 million hours of video, can be used alongside the Omniverse platform to explore potential AI model outcomes. The move is part of Nvidia's strategy to diversify its revenue sources amid competition from tech giants like Amazon and Google, who are developing their own chips. Huang anticipates that autonomous vehicles will become a significant robotics industry, with Nvidia partnering with companies like Toyota to advance these technologies. The success of Cosmos could be crucial in achieving widespread deployment of robots and autonomous vehicles in everyday life.
Key takeaways:
- Jensen Huang unveiled Nvidia's Cosmos platform to simulate scenarios for training real-world robots and autonomous vehicles.
- Cosmos generates virtual environments to provide synthetic data for reinforcement learning, helping AI models understand the physical world.
- Nvidia aims to diversify its income sources as Big Tech companies develop their own chips, reducing dependence on Nvidia's products.
- Huang believes autonomous vehicles could become the first multi-trillion dollar robotics industry, with Nvidia partnering with Toyota to advance this goal.