Chipmaker AMD has announced its acquisition of World Labs, a startup specializing in physical AI, co-founded by renowned AI expert Fei-Fei Li. The all-stock transaction, valued at $8.2 billion, will grant AMD access to cutting-edge technology in a critical area of artificial intelligence, which is anticipated to enhance the capabilities of autonomous systems like robots and self-driving vehicles.
This strategic acquisition is set to bolster AMD's position in the competitive AI landscape, especially as the company produces GPUs essential for training and deploying AI models. It will also allow AMD to refine its offerings to meet evolving AI needs, providing a stronger counter to rival Nvidia, which is simultaneously expanding its own initiatives in physical AI.
AMD CEO Lisa Su expressed her excitement about the merger, stating on X, "Together, we’ll merge World Labs’ profound expertise in AI and world models with AMD’s computing prowess to drive the future of AI and support the open AI framework."
This deal highlights a significant collaboration between two prominent female leaders in the field: AMD's Su and Li, who is often celebrated as the "godmother of AI" for her pivotal role in developing ImageNet, a foundational visual database that spurred advancements in deep learning two decades ago. Following the acquisition, Li will assume the role of executive vice president and chief scientist at AMD, directly reporting to Su.
In a detailed post on X, Li emphasized the importance of dedicated hardware efforts for optimizing AI efficiency and scalability. She praised AMD’s Su for being a supportive friend and an early backer of World Labs.
The partnership between AMD and World Labs began last year, focusing on optimizing model training and inference utilizing AMD’s GPU technology. Li noted, "As our teams collaborated, it became clear how naturally our software and hardware ecosystems, foundation models, and applications aligned."
World models present an alternative to the large language models (LLMs) developed by companies like OpenAI, Anthropic, and Google. Instead of relying on text-based datasets, world models leverage video data depicting the three-dimensional physical world. This approach aims to equip AI with a deeper understanding of how objects behave in real-world scenarios, an aspect many industry experts believe is essential for advancing AI in robotics and other physical applications.
World Labs has garnered approximately $1 billion in funding and is one of several well-capitalized startups exploring the potential of world models. Notably, Yann LeCun, former chief AI scientist at Meta, launched AMI Labs earlier this year. Nvidia is also actively involved, having developed its own world models and collaborating with startups like Wayve, which focuses on self-driving technology.
“Our company was founded on the understanding that language alone is inadequate for model development. The universe consists of tangible objects, not just words,” Li remarked.
While the specific number of World Labs employees joining AMD has not been disclosed, the companies confirmed that the entire team will transition as part of the acquisition. The deal is anticipated to finalize by the end of the current year, pending regulatory approvals.
In after-hours trading on Monday, AMD's stock remained relatively stable.
