AI Publications

Elon Musk’s artificial intelligence startup xAI has hired two former Nvidia researchers to accelerate the development of advanced “world models”—AI systems designed to understand and navigate real environments, beyond traditional text-based language models. This move positions xAI in direct competition with tech giants like Meta and Google in the race to create AI that can interact with and simulate the real world.

The company recruited Zeeshan Patel and Ethan He, both AI researchers experienced in world modeling during their tenure at Nvidia. According to reports from the Financial Times and confirmation from Patel’s personal website, he now works as a “Technical Staff Member at xAI focused on multimodal/world model research.” Nvidia has led in this field with its Omniverse platform, which creates and runs simulations for digital worlds.

World models represent a significant leap beyond current video generation systems like OpenAI’s Sora, which create content by predicting visual patterns from training data. Instead, world models offer real-time causal understanding of physics and how objects interact in different environments. The technology is trained with video material and robotic data to understand real-world dynamics, movement, and cause-and-effect relationships.

Musk confirmed on X that xAI plans to launch “a great AI-generated game before the end of next year,” reiterating a goal first set in 2024. The company is actively hiring for its “omni team,” which aims to create multimodal AI systems, including images, video, and audio, with salaries ranging from $180,000 to $440,000 per year. A specialized “video game tutor” position offers $45-$100 per hour to help train Grok, xAI’s chatbot, in game design.

Broader Ambitions in Robotics

Beyond video games, sources familiar with xAI’s plans indicate that the same world model technology could eventually power robotics applications, teaching machines to understand and design real spaces. This aligns with wider industry trends, as Nvidia has suggested that the potential market for world models could be as big as the global economy.

The timing coincides with xAI’s recent launch of its new image and video generation model, Grok Imagine, which the company describes as offering “massive improvements” and is provided to users for free. However, industry veterans remain skeptical about the role of AI in gaming. Michael Douse, publishing head at Larian Studios, argued that the main challenge for the gaming industry is not technical but “leadership and vision,” suggesting there’s a need for greater creative expression rather than “mathematically produced and psychologically trained gameplay loops.”

Nvidia CEO publicly confirms investment in xAI

Nvidia CEO Jensen Huang publicly confirmed his company’s investment in Elon Musk’s xAI during an interview with CNBC on Wednesday, while expressing that his biggest regret was not investing more money in the project. “I’m very excited about the funding opportunity they’re creating,” Huang said on CNBC’s Squawk Box. “My only regret with xAI—we’re already investors—is that I didn’t give more money.” Huang praised Musk’s entrepreneurial trajectory, saying, “almost everything Elon is involved in, you really want to be part of too.”

Nvidia CEO calls Musk the ‘ultimate GPU’ for AI leadershipNvidia CEO Jensen Huang called Elon Musk the “ultimate GPU” during a recent podcast appearance, praising the billionaire’s ability to manage complex AI infrastructure projects while his companies remain among Nvidia’s biggest clients. Speaking on the BG2 podcast late last month, Huang compared Musk’s cognitive abilities to one of Nvidia’s graphics processing units, highlighting his capacity to retain vast amounts of information and coordinate interdependent systems across multiple companies. “These AI supercomputers are complicated things,” Huang explained. “The technology is complex. Acquiring it is complex due to financing issues. Securing the land, the energy and the housing, powering it all, is complicated too.” When podcast hosts compared Musk to a “large GPT” or supercomputer, Huang responded: “He is the ultimate GPU.” He emphasized Musk’s unique advantage for developing xAI’s ambitious Colossus II project: “All these systems are interoperating and the interdependencies reside in a single head, including funding.”