Musk poached people from Nvidia to develop AI games!

The competition among tech giants in the field of "world models" is welcoming a high-profile new entrant: Elon Musk's xAI.

The Financial Times recently reported that, to gain an advantage in the increasingly fierce AI arms race, xAI successfully poached several key researchers from chip giant Nvidia this summer to build its own elite R&D team.

The news has sparked widespread industry concern: Will leading institutions like Meta and Google DeepMind in the field of world models face their strongest challenger? First, secret poaching, targeting Nvidia's core team.

As AI labs around the world compete to develop world models, xAI's entry is particularly eye-catching: it directly targets Nvidia's senior R&D talent. It is reported that at least two researchers from Nvidia joined xAI this summer: Zeeshan Patel and Ethan He. Zeeshan Patel received his master's degree from the University of California, Berkeley, in May of this year. His research focuses on deep learning, generative models, and physical AI.

Prior to joining xAI, he worked on foundational model development at Apple's AI/ML department and later at NVIDIA Research, focusing on building generative world models, with a particular expertise in large-scale multimodal models and training frameworks.

Ethan He boasts an equally impressive resume: he graduated from Xi'an Jiaotong University with a bachelor's degree and then earned a master's degree in computer vision with full honors from Carnegie Mellon University.

His research has been cited nearly 8,500 times on Google Scholar. He previously worked as a research engineer at Facebook AI, leading large-scale self-supervised video learning and foundational video models. After joining NVIDIA in 2023, he focused on research in hybrid expert models, multimodal understanding, and world models.

Notably, both individuals were deeply involved in the core development of NVIDIA's Omniverse digital twin platform. This background undoubtedly provides critical technical support for xAI's efforts to build highly realistic virtual worlds.

Second, xAI's first focus is gaming; Musk reiterates his "2026 promise."

According to sources familiar with the matter, xAI's first application of its world model will be in video games. The team is currently developing an AI system capable of automatically generating highly realistic, interactive 3D scenes, adapting content in real time based on player behavior to create truly living virtual worlds.

This direction aligns closely with Musk's own goals. A few days ago, he reaffirmed on the social platform X that xAI will launch a revolutionary AI-generated game powered by its world model by the end of 2026.

Third, xAI is urgently establishing an "omnimodal" team and recruiting game mentors with high salaries.

To achieve this vision, xAI is urgently establishing an internal "omnimodal team" dedicated to developing integrated image, video, and audio understanding and generation technologies.

Recent job postings indicate that the team is recruiting for several key positions:

In addition, a position titled "video game mentor" has sparked heated discussion within the industry. It's speculated that this position aims to help AI systems gain a deep understanding of game design logic and player psychology. One X platform user commented, "AI will greatly improve game development efficiency and unleash human creativity. This makes perfect sense."

Fourth, "Understanding the Nature of the Universe": World Models Become a Necessary Path

Musk has publicly stated numerous times the ultimate mission of xAI—"to enable AI to understand the nature of the universe." World models are the core path to achieving this vision. They are not only a tool for generating content; they also aim to enable AI to internally construct a "digital universe" capable of simulating the laws of real physics and society.

If successful, this technology will not only power the next generation of games but will also be widely applied in areas such as autonomous agents, fully autonomous driving, and embodied robotics, completely rewriting the boundaries of human-computer interaction.

The entry of xAI into the field has undoubtedly injected new variables into the global research and competition on world models. Can Musk deliver on his promise of AI gaming by 2026? This "model war" for the future of AI has just begun.