DeepMind CEO Hassabis says world models are AI’s next frontier amid bubble
Gemini 3 just started reaching users, but Demis Hassabis, DeepMind’s CEO, seems already looking beyond that splash. Most of his time these days is spent on a different breed of AI, the so-called world models, which try to let machines make sense of, and forecast, whole environments rather than just single tasks. You can see the idea in projects like SIMA 2 and the video-generation system Genie 3, both of which push that kind of reasoning a step further.
Meanwhile, many observers keep talking about an AI bubble, and it’s hard to tell how that hype will shape funding for longer-term work. The push for quick product releases sits uneasily next to bets on deeper science, so Hassabis’s emphasis feels a bit unusual. It leaves me wondering: if the field really moves toward world models, what kinds of breakthroughs might we see next?
World models as the next AI frontier While Gemini 3 is rolling out, Hassabis is already focusing on the next research frontier: world models. He spends most of his research time on this topic, citing projects like SIMA 2 and the video-generation model Genie 3 as examples. According to Hassabis, these models, already used internally for training robots and other agents, will be critical to achieving AGI.
He predicts a "ChatGPT moment" for world models, but the biggest obstacles are the price and current technical hurdles. "We'd love to put Genie in the hands of more people, but it's expensive," Hassabis said, explaining that "basically, a consumer of it is another instance of the creation of it." Before scaling is possible, he noted, challenges such as "making it consistent longer than a minute" must also be solved. Hassabis warns of AI bubble, affirms Google's "Engine Room" strategy When asked about a potential AI bubble, Hassabis offered a nuanced view.
Gemini 3 Pro is out now. Still, Hassabis thinks the hard part is just beginning. He keeps pointing at world models as the next AI step - something we already see in projects like SIMA 2 and the video-generation tool Genie 3.
To his eyes, Google’s long-term plan finally shows some traction after a hype-heavy stretch that many employees and even Sundar Pichai rode. He also sounds a note of caution, saying the private-sector market looks bubbly and that money might be flowing faster than real progress. Gemini 3 aims to be a solid all-round model, yet it’s not clear if world-model work will turn into usable breakthroughs soon.
Hassabis is spending most of his research hours on that area, which suggests a shift in priorities. Whether this will reshape DeepMind’s output or just add another layer of complexity remains an open question. I guess the tug-of-war between ambition and market realities will write the next chapter.
All eyes are on what actually happens.
Related Reading
Browse all Google news →
Market Trends 60% of singles label AI relationships as cheating, sparking divorce concerns
Market Trends Riemer warns AI attacks outpace defenses as CISOs tackle 11 runtime threats
Market Trends Dell launches Pro Max with GB10 to support on‑device AI development
Related Topic Gemini 3 Pro builds screenshot-to-code app in two prompts, fixes bugs
LLMs & Generative AI
Related Topic Baidu opens multimodal AI, claims it beats GPT‑5 and Gemini, runs on one 80GB GPU
LLMs & Generative AI
Common Questions Answered
What does DeepMind CEO Demis Hassabis identify as the next AI frontier?
Hassabis says world models are the next AI frontier, emphasizing their ability to understand and predict environments holistically. He believes these models, exemplified by projects like SIMA 2 and Genie 3, are crucial for achieving artificial general intelligence.
How are the projects SIMA 2 and Genie 3 related to DeepMind's world model research?
SIMA 2 and the video‑generation model Genie 3 are concrete examples of DeepMind's world model initiatives. Both are used internally for training robots and other agents, demonstrating how predictive, environment‑aware AI can be built.
What comparison does Hassabis make between world models and a "ChatGPT moment"?
Hassabis predicts a "ChatGPT moment" for world models, suggesting a rapid adoption surge similar to what ChatGPT experienced. He expects this breakthrough to accelerate progress toward more capable, general‑purpose AI systems.
What warning does Hassabis give about the current AI market environment?
Hassabis cautions that the private‑sector AI market is experiencing a bubble, where investment may outpace sustainable technological progress. He implies that hype‑driven funding could risk overshooting realistic development timelines.