AnswersPublished18 sources
At CoRL 2026, Lisa Su and Fei-Fei Li Put Robotics at the Center of AI’s Next Compute Wave
Su and Li framed physical AI as a potential new source of computing demand, from cloud training to edge devices. World Labs’ Atlas is described by CoRL as an autoregressive frame generation model that combines pixel generation with 3D scene reconstruction.
Published byEdited with GPT-6 LunaImages generated with GPT Image 2
Research answer
Key takeaways
- Su and Li framed physical AI as a potential new source of computing demand, from cloud training to edge devices.
- World Labs’ Atlas is described by CoRL as an autoregressive frame generation model that combines pixel generation with 3D scene reconstruction.
- The available coverage does not verify several more specific claims about robot performance, hardware, AGI, or research compute shortages.
AI Prompt
openai.comCreate a landscape editorial hero image for this Studio Global article: What did AMD CEO Lisa Su and World Labs founder Fei-Fei Li reveal at their November 12 fireside chat at the CoRL 2026 Conference on Robot Le. Article summary: Their central message was that robotics and physical AI could drive the next wave of computing demand, while today’s models still have not solved the system-level problems of interacting with the physical world. [7] The . Topic tags: general, news, general web, user generated. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts w
At CoRL 2026, AMD CEO Lisa Su and World Labs founder Fei-Fei Li discussed robotics as a potential new driver of computing demand. The core idea is that physical AI needs more than model training in the cloud: it also depends on computation closer to real-world devices. But reporting on the size of the opportunity conflicts, and several technical details circulating around the conversation are not established by the available sources. 19
20
22
The case for robotics as a computing frontier
Li said robotics could increase demand for computing at both the training and edge levels. Coverage of the discussion also noted a continuing challenge: language and vision-language models have not solved the broader problems involved in interacting with the physical world. 19
20
That makes the opportunity more than a question of building larger models. Robots operate in physical environments, so the computing needs extend beyond training. The available reporting, however, does not verify specific claims about how Su said AMD would coordinate compute, memory, software, and system design—or about particular power and latency targets.
Su’s market forecast comes with a reporting caveat
Some reports attribute to Su a forecast that physical AI, particularly robotics, could become a market worth hundreds of billions of dollars in the coming years. 19
20 Another report gives a substantially smaller estimate, describing the opportunity as tens of billions.
22
The disagreement means the larger figure should be treated as a reported forecast, not a settled or independently verified market estimate. The supplied coverage does not resolve the discrepancy.
What CoRL says about World Labs’ Atlas
The CoRL program describes Atlas as an autoregressive frame-generation model that brings together pixel generation and 3D scene reconstruction. 30 That description supports a connection to world modeling: generating visual frames while reconstructing a three-dimensional scene.
More specific claims about Atlas—including particular input combinations, a 15-frames-per-second performance figure, or demonstrated uses in robot navigation—are not substantiated by the conference description or reporting excerpts available here. They should not be presented as established results on this evidence.
AMD’s World Labs agreement is relevant context
AMD announced on September 28 that it had entered into an agreement to acquire World Labs in an all-stock transaction valued at about $8.2 billion. The company said the deal would add World Labs’ researchers and model expertise as AMD develops hardware, software, and systems for emerging AI models and applications. 1
14
The agreement had not yet been reported as completed: coverage described a year-end 2026 target, subject to regulatory approval and other conditions. It also reported that Li was expected to become AMD’s executive vice president and chief scientist, reporting to Su, after the transaction closes. 10 Reports described the CoRL conversation as the pair’s first joint public appearance since the acquisition announcement.
7
What remains unclear
The sources support the broad themes of robotics-related compute demand, Atlas’s stated design, and the acquisition agreement. They do not establish several other claims raised around the event, including specific industrial deployments of AMD embedded FPGAs or AI engines; public research-compute shortages; Su’s views on imminent AGI; or a detailed AMD strategy against Nvidia.
The most reliable takeaway is narrower: Su and Li presented robotics and physical AI as a possible next frontier for computing, while the available coverage leaves the forecast’s scale—and many of the technical details—uncertain.
Studio Global AI
Continue your research
This page includes a source-backed answer you can continue inside Studio Global.
People also ask
What is the short answer to "At CoRL 2026, Lisa Su and Fei-Fei Li Put Robotics at the Center of AI’s Next Compute Wave"?
Su and Li framed physical AI as a potential new source of computing demand, from cloud training to edge devices.
What are the key points to validate first?
Su and Li framed physical AI as a potential new source of computing demand, from cloud training to edge devices. World Labs’ Atlas is described by CoRL as an autoregressive frame generation model that combines pixel generation with 3D scene reconstruction.
What should I do next in practice?
The available coverage does not verify several more specific claims about robot performance, hardware, AGI, or research compute shortages.
Sources
aifollow.newsFei-Fei Li and Lisa Su discuss robotics computing needs and model limits
news.futunn.comFei-Fei Li in Dialogue with Lisa Su: Robotics to Propel Computing Power into Next Stage of Development
kucoin.comLi Fei-Fei and Lisa Su Discuss Robotics and Compute at ...
2026.corl.orgCoRL 2026 - Keynotes