- In The Loop AI
- Posts
- Nvidia and Samsung Invest in Robotics Startup Skild AI
Nvidia and Samsung Invest in Robotics Startup Skild AI
Meta’s new V-JEPA 2
In Today’s Issue:
Nvidia and Samsung Invest in Robotics Startup Skild AI
Meta’s new V-JEPA 2
Read time: 3 minutes
Nvidia and Samsung Invest in Robotics Startup Skild AI

Skild AI
Nvidia and Samsung are investing in Skild AI, a robotics software startup, as part of its Series B round valuing the company at about $4.5 billion. Nvidia is contributing $25 million, while Samsung is investing $10 million, alongside a $100 million lead investment from SoftBank, according to sources. Samsung sees the move as a way to monitor Skild’s development and talent while staying competitive with other Korean firms like LG and Hanwha, which are also investing smaller sums. Though Samsung reportedly views parts of Skild’s tech as lagging competitors, it remains focused on expanding in robotics. Nvidia is positioning “physical AI” — robots and autonomous systems — as a key future revenue stream and has invested in companies like Figure AI and Serve Robotics. Samsung, which already sells robot vacuums and owns a major stake in Rainbow Robotics, is developing more advanced consumer robots, including Ballie, a home projector bot launching with Google later this year.
Meta’s new V-JEPA 2

Meta has introduced V-JEPA 2, an upgraded "world model" designed to help AI agents understand and predict physical interactions in the real world. Building on last year’s V-JEPA, the new model was trained on over a million hours of video to teach AI common-sense reasoning, such as anticipating the trajectory of a moving object or the next step in a physical task. In examples shared by Meta, the model can predict that a robot holding a spatula and plate near a stove might next serve cooked eggs. Meta claims V-JEPA 2 runs 30 times faster than Nvidia’s Cosmos model, though the companies may use different benchmarks. Meta’s chief AI scientist Yann LeCun said world models like V-JEPA 2 could enable robots to perform real-world tasks without massive datasets, paving the way for practical home and service robots.
How was today's newsletter? |
Reply