AMD Acquisition Boosts Fei-Fei Li's Net Worth to $1.6 Billion
According to Bloomberg, after AMD's all-stock acquisition of AI startup World Labs valued at $8.2 billion, co-founder and CEO Fei-Fei Li's net worth is expected to rise to at least $1.6 billion upon completion of the deal, an increase of over 50% from before. The transaction is pending regulatory approval, with a target closing date by the end of 2026.
Bloomberg Billionaires Index estimates she holds about 19% of World Labs. Li first became a billionaire after the company completed a funding round in February this year that valued it at over $5 billion; Reuters reported that this funding round was approximately $1 billion, with AMD also participating. World Labs was founded in early 2024 in San Francisco, focusing on spatial intelligence and physical AI, generating, reconstructing, and simulating interactive 3D environments from text, images, and videos using models. Its product line includes the world generation model Marble, set to launch in 2025, aimed at design, simulation, and robotic learning.
After the closing, Li will serve as AMD's Executive Vice President and Chief Scientist, reporting to CEO Lisa Su, with the research team continuing to develop models and leveraging AMD's computing power. The two parties have already begun collaboration on training and inference optimization, and Li attended AMD's CES event earlier this year. She told Bloomberg Television that the goal is to accelerate the software and hardware development flywheel; Su stated that the better the understanding of end-to-end workloads, the better the systems that can be created.
The $8.2 billion valuation is about 60% higher than the over $5 billion valuation in February, marking it AMD's second-largest acquisition in history, following Xilinx. The chip company seeks to define cutting-edge workloads in reverse for its roadmap, while the lab aims to scale the world model from papers and demos to trainable clusters. Regulatory approvals and closing conditions are still pending, and the net worth figures will be based on shareholding and stock compensation at the time of closing.
In market terms, this is a chip manufacturer exchanging its stock for the modeling team and the rights to define physical AI scenarios. The buyer is AMD, which needs differentiated training loads to catch up with NVIDIA's "chip plus world model" combination, while the seller is World Labs' shareholders, who have proven the commercial viability of spatial models and need computing power to scale. Funds are shifting from the market capitalization of the public company to the founding team and early investors. Beneficiaries include Li, who holds about 19% of the shares, and other investors in the round, while existing AMD shareholders face the pressure of equity dilution and must prove the premium with self-developed models; robotics and simulation software companies will encounter a world model supplier now integrated into a chip manufacturer.
ABAB AI Insight
Fei-Fei Li, known for her work with ImageNet and computer vision, has long been a professor at Stanford. In 2024, she left the mainstream focus on "only large language models" to establish World Labs, arguing that general intelligence must be grounded in physical space. Within months, the company's valuation reached $1 billion, and with the release of Marble in 2025, it surpassed a $5 billion valuation in February 2026, placing her on the billionaire list. This isn't AMD's first approach: they invested in training and inference optimization last year and have now acquired the entire lab, placing her in the Chief Scientist role reporting to Lisa Su. The trajectory from scholar to entrepreneur to head of research at a chip company follows the same logic: "whoever defines the workload defines the next chip."
The capital path is an all-stock acquisition, not a cash-out. The 19% stake corresponds to a billion-dollar asset at the $8.2 billion valuation, shifting her wealth from "paper riches based on private equity valuation" to "public stock that fluctuates with AMD's stock price." AMD was already on the list for the $1 billion funding last year, effectively locking in a relationship with minority equity before acquiring the team at a premium. The money is not flowing into new independent financing but rather into integrating model research, system software, and silicon roadmap. The motivation is clear: synthesizing a 3D world and robotic simulation lacks real data, and the world model will become the blueprint for the next wave of computing orders.
The comparison is clear. NVIDIA has already sold world models alongside chips; Google and Meta have kept large model labs within their cloud and advertising machines; Microsoft uses OpenAI to define Azure workloads. AMD's acquisition of World Labs is the second time, after Xilinx, that it has used acquisition to fill the gap in "seeing future workloads," still positioning itself as a follower rather than a leader with a CUDA-level ecosystem. Marble proves that generative 3D can be productized, but it has yet to demonstrate the stability to consume wafer capacity like training clusters. The industry phase is that physical AI is entering the narrative of capital expenditure in chip companies.
Structurally, this represents a reconstruction of the industry chain driven by technological substitution. The substitute is not the existing GPUs themselves, but rather the chip planning approach that "only consumes text and video tokens"; the world model writes new training forms for 3D interaction, robotic learning, and simulation data generation, forcing a rearrangement of interconnects, memory, and software stacks. The mechanism is: whoever locks the lab into the chip company first can write the yet-to-scale physical intelligence into next year's SKUs and software roadmap. Pricing power shifts from the next round valuation of independent model companies to the equity percentage that chip companies are willing to pay for defining rights.