AI Founders Anima Anandkumar and Benedikt Jenik Unveil Physics-First AI Capable of Processing 5 Trillion Data Points
SAN FRANCISCO: Anima Anandkumar and Benedikt Jenik, co-founders of Accelerated Understanding Inc., have introduced a groundbreaking AI model designed to process an unprecedented 5 trillion data points in a single prompt. This innovative technology, unveiled on Tuesday, aims to revolutionize applications in fields such as chip design, robotics, weather forecasting, and enterprise solutions.
The AI model developed by Accelerated Understanding is distinct from conventional systems. Unlike popular models such as those from Anthropic and Google, which are primarily language-based, this new AI focuses on physics. In practical terms, the model’s capacity to handle 5 trillion data points is equivalent to reading “War and Peace” 5 million times in one sitting, showcasing its immense processing power.
A System Built for Physics, Not Language
Anandkumar and Jenik emphasized that their AI system is fundamentally different from language-centric models. While models like ChatGPT are trained on vast amounts of text data to predict the next word in a sentence, their AI has been designed to predict physical phenomena across space and time. This approach discards the traditional Transformer architecture, which is prevalent in many AI systems, opting instead for neural operators—a technology that Anandkumar has been instrumental in developing.
Anandkumar, a professor of computing and mathematical sciences at Caltech, articulated the philosophical shift in their approach. She stated that a language-centric view places humans at the center of intelligence, whereas a physics-centric view aligns more closely with the natural world.
Anima Anandkumar, Caltech Bren Professor of Computing, attends the Reuters NEXT conference, in New York City, New York, U.S., December 11, 2024. (Reuters)
World Models and Their Implications
The concept of “world models” is gaining traction in the AI community, with various companies, including those led by prominent figures like Yann LeCun and Fei-Fei Li, exploring ways to enhance AI’s understanding of spatial reality. Accelerated Understanding’s approach aims to create a more versatile neural operator capable of predicting phenomena that are not immediately visible, thereby enhancing business applications.
One of the key areas of interest is chip design. While many companies utilize AI to program computers and analyze text for semiconductor applications, Accelerated Understanding posits that a deeper understanding of physics is essential for optimizing materials and temperatures in chip performance. This could lead to significant reductions in the trial-and-error process typically associated with laboratory testing.
The potential applications of this AI extend beyond chip design. The founders believe it can also enhance robotics, predict extreme weather events, and analyze geological data for energy companies. Instead of relying on specialized mathematical models for each application, their AI could provide a unified solution for various physics-related inquiries in the business sector.
The company’s initial focus is on enterprise partnerships rather than consumer products, as stated by the co-founders.
Interest from Bezos and Project Prometheus
The innovative work of Anandkumar and Jenik has attracted attention from notable figures in the business world, including Jeff Bezos, the founder of Amazon. In late 2024, during a dinner in Los Angeles, investor Vik Bajaj, who later co-founded Project Prometheus with Bezos, discussed potential collaboration with the duo. Anandkumar’s previous experience as a scientist at Amazon and her tenure as a director at Nvidia positioned her as a key player in AI discussions.
Bajaj later presented an offer letter titled “Project Prometheus,” which proposed that Anandkumar could serve as the public face of the company, a board member, and the custodian of its scientific vision. The offer included a 35% equity stake and a substantial salary package. However, Anandkumar and Jenik chose to continue developing their technology independently.
In June 2026, Prometheus successfully raised $12 billion in a Series B funding round, focusing on AI technologies aimed at automating the manufacturing of complex physical systems.
Accelerated Understanding Inc’s co-founders Anima Anandkumar and Benedikt Jenik pose for a picture, in Pasadena, California, U.S., August 23, 2026. (Reuters)
Origins at Nvidia and Future Prospects
Although Anandkumar did not disclose specific funding details, she confirmed existing partnerships with computing providers that have supplied hardware clusters for the development and operation of Accelerated Understanding’s AI. Nvidia, where Anandkumar worked prior to launching her company, has been a significant influence on her research. She led a team that explored how Nvidia’s graphics processing units could be utilized for advanced AI applications, including improving weather prediction accuracy.
Nvidia CEO Jensen Huang has been supportive of Anandkumar’s initiatives, expressing enthusiasm for the potential of AI to transform fields traditionally dominated by human expertise. Anandkumar recounted Huang’s excitement when discussing the implications of AI on physics research.
For further insights into this development, visit Emirates247.
Read all the latest developments and breaking updates in the Latest News section.
Published on 2026-08-25 15:16:00 • By the Editorial Desk
