Researchers Walk Away From Bezos-Backed Prometheus to Build Physics AI Processing 5 Trillion Data Points
7 Sources
[1]
EXCLUSIVE: The AI founders who walked away from Bezos-backed Prometheus to model the universe
SAN FRANCISCO, Aug 25 (Reuters) - A research duo, once pitched to lead the Jeff Bezos-backed Project Prometheus, on Tuesday unveiled what they've built on their own: an AI model that can take in and produce information so vast that a lone computer might not handle it. Their company, Accelerated
[2]
Billionaire Jeff Bezos came calling. These researchers walked away to build physics AI, leaving a $2 mn salary behind
Two AI researchers declined a lucrative offer from Jeff Bezos' Project Prometheus. They instead chose to develop their own artificial intelligence system focused on physics. This new AI model processes significantly more data than current language-based systems. Their company, Accelerated
[3]
Founders once linked to Bezos project pitch AI that models physics, not language
SAN FRANCISCO - A research duo, once pitched to lead the Jeff Bezos-backed Project Prometheus, on Tuesday unveiled what they've built on their own: an artificial intelligence model that can take in and produce information so vast that a lone computer might not handle it. Their company, Accelerated
[4]
AI model for physics: The AI founders who walked away from Bezos-backed Prometheus to model the universe
The difference is that this new system is not built for understanding language. It is built for physics, said co-founders Anima Anandkumar and Benedikt Jenik, in an exclusive interview ahead of Tuesday's corporate launch. A research duo, once pitched to lead the Jeff Bezos-backed Project
[5]
AI Researchers Walk Away From Bezos-Backed Prometheus to Build Physical-World AI
Two artificial intelligence researchers who were once approached to lead have chosen a different path, launching their own company to develop technology focused on understanding the physical world rather than language. Anima Anandkumar and Benedikt Jenik, the co-founders of Accelerated
[6]
The AI founders who walked away from Bezos-backed Prometheus to model the universe: Reuters exclusive
A research duo, once pitched to lead the Jeff Bezos-backed Project Prometheus, on Tuesday unveiled what they've built on their own: an AI model that can take in and produce information so vast that a lone computer might not handle it. Their company, Accelerated Understanding Inc, says it has
[7]
The AI founders who walked away from Bezos-backed Prometheus to model the universe
SAN FRANCISCO, Aug 25 (Reuters) - A research duo, once pitched to lead the Jeff Bezos-backed Project Prometheus, on Tuesday unveiled what they've built on their own: an AI model that can take in and produce information so vast that a lone computer might not handle it. Their company, Accelerated
Share
Copy Link
Two AI researchers declined a lucrative offer from Jeff Bezos' Project Prometheus, including a $2 million salary and 35% stake, to launch Accelerated Understanding. Their physics AI model processes 5 trillion data points using neural operators instead of language-based Transformer architecture, targeting enterprise applications in chip design and robotics.
Researchers Reject Bezos-Backed Prometheus for Independent Physics AI Venture
Two AI researchers, Anima Anandkumar and Benedikt Jenik, turned down a significant offer from Jeff Bezos' Project Prometheus to pursue their own vision of artificial intelligence. The duo unveiled Accelerated Understanding Inc on Tuesday, introducing an AI model for physics that processes 5 trillion pieces of data in a single prompt
1
2
. This represents approximately 5 million times the capacity of flagship models from Anthropic and Google, comparable to reading Tolstoy's "War and Peace" 5 million times in one sitting3
. The offer from Bezos-backed Prometheus included a combined $1 million annual salary that would double to $2 million after three months, plus a 35% stake in the company for both founders4
. The proposal outlined more than $2 billion in committed funding through Series B from investors including Jeff Bezos1
.Physics AI Challenges Language-Centric AI Paradigm
Unlike ChatGPT and other language-centric AI systems built on the Transformer architecture, this AI model focuses on understanding and predicting physical phenomena across space and time. The system uses neural operators, a technology that Anandkumar helped pioneer years ago during her tenure at Nvidia
1
. "The language-centric view of intelligence is humans at the center. Putting physics at the center is a nature-centric view," explained Anandkumar, a Caltech professor of computing and mathematical sciences4
. This physical-world AI represents a fundamental shift from text-based models that predict the next word in a sentence to systems that can predict physical phenomena even the eye cannot see1
. Various companies, including startups overseen by AI leaders Yann LeCun and Fei-Fei Li, are pursuing similar world models that understand spatial reality better than AI trained on text4
.Enterprise Applications Target Chip Design and Beyond
Accelerated Understanding aims to solve complex problems across multiple industries through a single AI that can handle any physics query. In chip design, the company believes an intrinsic grasp of physics is essential for optimizing materials and temperatures for chip performance, reducing trial and error in laboratory settings
1
. The AI that models physics can also power robotics, predict extreme weather, and sift through geological data for energy companies4
. Instead of brittle and bespoke mathematical models for each application, Accelerated Understanding's approach offers a more generally capable form of neural operators1
. The company is focused on enterprise deals rather than consumer offerings initially, targeting businesses that need to predict physical phenomena with unprecedented accuracy4
.Related Stories
Nvidia Origins and Computing Infrastructure
The idea for this physics AI originated during Anandkumar's time at Nvidia, where she worked from 2018 as a director leading scientists researching how graphics processing units could be used for frontier AI
1
. An early project demonstrated how AI could speed up extreme weather prediction with accuracy matching complex traditional forecasting methods, which amazed Nvidia CEO Jensen Huang2
. Huang encouraged Anandkumar to pursue the neural operators concept, responding enthusiastically: "I want it to eat all their lunches" after seeing her work2
. While Anandkumar declined to discuss funding details, she confirmed partnerships with computing providers who furnished hardware clusters to develop and run the AI1
. Meanwhile, Bezos and co-founder Vik Bajaj proceeded with Project Prometheus, raising $12 billion in Series B funding in June 2026 to target AI that can automate the manufacturing of complex physical systems4
.References
Summarized by
Navi
[1]
[5]
Related Stories
Jeff Bezos's Project Prometheus Recruits xAI Co-Founder to Build Industrial AI Systems
07 Apr 2026•Technology
Jeff Bezos Returns as Co-CEO of $6.2B AI Startup Project Prometheus
17 Nov 2025•Startups
Jeff Bezos' Prometheus raises $12 billion to build an artificial general engineer for physical AI
11 Jun 2026•Startups
Recent Highlights
1
AI Hallucination Nearly Sparked US Military Confrontation With Chinese Ship in Middle East
Technology
2
Google Gemini Hacked Three Real Companies During Cybersecurity Testing, Raising AI Safety Concerns
Technology
3
OpenAI Claims AI Breakthrough on Navier-Stokes Problem, Sparking Math Community Crisis
Science and Research
Recent Highlights
Today's Top Stories
Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Follow topics that matter to you and stay ahead.
