Google DeepMind CEO Predicts AGI by 2030
Google DeepMind CEO Demis Hassabis estimates that artificial general intelligence (AGI) could arrive by 2030, warning that governments and industries have only a few years left to prepare for its impact.
Speaking at recent industry events, including the Stanford Graduate School of Business and Sequoia Capital’s AI Ascent, Hassabis reaffirmed his timeline for AGI—systems capable of matching or exceeding human cognition across a wide range of tasks. He described the current state of artificial intelligence as the “foothills of the singularity,” referring to the theoretical threshold where AI begins to improve itself autonomously.
“I believe that we’re only a few years away from that, maybe 2030, plus or minus a year,” Hassabis told the Stanford audience. “Which is astounding to think, really.”
The Industry Timeline Debate
Hassabis’s forecast places him near the middle of an increasingly aggressive timeline among technology executives. As AI laboratories continue scaling up their computing infrastructure, the consensus around AGI’s arrival has narrowed.
OpenAI CEO Sam Altman has suggested that digital superintelligence is imminent, while Anthropic CEO Dario Amodei has pointed to late 2026 or 2027 as the window when AI systems could outperform humans at almost every task. Tesla and xAI chief executive Elon Musk recently estimated that AGI could emerge within the next two years.
Despite the varying predictions, the broader AI research community remains divided on whether current architectures—primarily large language models based on transformers—can actually achieve general intelligence. Meta’s chief AI scientist, Yann LeCun, has consistently dismissed imminent AGI claims, arguing that current systems lack the underlying logic and physical understanding of the world required for human-level cognition.
DeepMind’s chief AGI scientist, Shane Legg, maintains a more measured estimate, placing a 50 percent probability on reaching “minimal AGI” by 2028.
The Path to Intelligence
Hassabis founded DeepMind in 2010 with a two-part mission: first, solve intelligence, and second, use that intelligence to solve everything else. Google acquired the London-based research lab in 2014, and it has since produced some of the industry’s most significant technical milestones, from mastering the complex board game Go to predicting the 3D structures of nearly all known proteins.
The lab merged with Google’s Brain division to form Google DeepMind, consolidating the tech giant’s AI research to compete directly with OpenAI. During a recent panel at Bloomberg House in Davos, Hassabis noted the ferocity of this corporate race. He described the current environment as “maybe the most intense competition there has ever been in technology.”
To maintain its edge, Google has integrated its full hardware and software stack, relying on its custom Tensor Processing Units (TPUs) to train increasingly large models. While competitors in the United States remain the primary challengers, Hassabis acknowledged that Chinese technology companies, such as ByteDance, are trailing the frontier by only about six months.
Physical Robotics and the Real World
While AGI represents a software milestone, Hassabis also expects rapid progress in physical AI systems. Current large language models operate entirely in the digital realm, but tech companies are racing to embed these systems into humanoid robots and industrial machines.
Hassabis predicted a significant breakthrough in physical robotics within the next 18 to 24 months. Google recently partnered with Boston Dynamics to integrate advanced AI models with physical hardware, aiming to create machines that can navigate and interact with the physical world with the same fluidity that chatbots process text.
If AI systems achieve both digital superintelligence and physical autonomy, the economic implications would be vast. Hassabis has theorized that AGI could eventually lead to a “post-scarcity world,” where intelligent systems manage resource allocation, optimize energy grids, and dramatically lower the cost of physical labor.
Labor and Human Skill Shifts
The accelerated timeline has intensified anxiety around job displacement. Generative AI systems already write code, draft legal documents, and synthesize research, encroaching on specialized knowledge work that was previously considered safe from automation.
Hassabis acknowledged these fears but emphasized that human adaptability will remain crucial. As AGI systems begin executing complex logic and administrative tasks, human workers will need to lean heavily on traits that machines cannot easily replicate.
In a recent interview, Hassabis noted that people possessing “taste, design sensibility, original thinking,” and the ability to synthesize concepts across multiple disciplines will hold a significant advantage over the next five years.
“We’re general intelligences ourselves, don’t forget,” Hassabis said. “Look at what we built around us—it’s incredible—with our hunter-gatherer brains. Why would we stop here?”
Rather than eliminating the need for human input, Hassabis expects AGI to lower the barrier to entry for creative and scientific pursuits. He compared future AI tools to “superpowers,” suggesting that individuals will soon be able to execute large-scale projects that currently require entire teams.
Regulatory Push
Because of the timeline he anticipates, Hassabis is pushing for more structured oversight. He recently published a framework proposing a new regulatory body to monitor advanced AI systems, suggesting the United States should lead the initiative.
Instead of a traditional government agency, which he argued would move too slowly, Hassabis proposed a public-private organization modeled after the Financial Industry Regulatory Authority (FINRA). This body would be funded by the tech industry but answerable to the federal government.
Under his proposal, AI models that cross specific capability thresholds would be classified as “frontier-class”. The companies developing them would automatically inherit strict testing and reporting requirements, regardless of whether the software is proprietary or open-source. Developers would be required to submit their models for review up to 30 days before public release, providing experts time to evaluate potential security risks.
What Happens Next
The focus at DeepMind remains on expanding AI capabilities in scientific research. The company recently achieved breakthroughs in biology with AlphaFold, which models protein structures, and is now applying AI to materials science and drug discovery. Hassabis believes that AGI could eventually compress drug discovery timelines from a decade down to a matter of weeks.
While the technical hurdles to AGI remain substantial, the engineering path forward is clearer than it was even two years ago. AI developers are moving beyond simple chatbots, building autonomous agents capable of using software tools, reasoning through multi-step problems, and executing long-term plans.
As these systems become more integrated into daily business operations, the focus is shifting from technical feasibility to societal readiness. For Hassabis, the priority is ensuring that global institutions adapt quickly enough to manage a transition that is now less than a decade away.
In this interview, Demis Hassabis outlines his 2030 timeline for artificial general intelligence and discusses the specific human skills that will retain value in the future labor market.
