- Who: Demis Hassabis, CEO of Google DeepMind
- What: Predicted AGI (artificial general intelligence) could arrive as early as 2029
- When: Remarks made at Google I/O 2026, May 2026, and in interview with Axios
- Where: Google I/O developer conference; Mountain View, California
- Why: Growing confidence that AI has found the right technical path; AI agents are a “practice run” for AGI
- Impact: Gartner forecasts AI spending to reach $2 trillion in 2026 and $3.3 trillion by 2029; industries and governments may have just a few years to prepare
Key Takeaways
- Google DeepMind CEO Demis Hassabis now expects AGI to arrive around 2030, with 2029 as a real possibility — a narrower window than his prior estimates.
- Hassabis described humanity as standing in the “foothills of the singularity” at Google I/O 2026.
- He framed today’s AI agents as a “practice run” — a society-wide stress test for far more powerful systems still to come.
- Worldwide AI spending is projected to reach $2 trillion in 2026 and $3.3 trillion by 2029, according to Gartner.
- Hassabis warned that economists and governments are “not taking this seriously enough.”
Google DeepMind CEO Demis Hassabis said at Google I/O 2026 that AGI could arrive as early as 2029. Speaking with Axios, Hassabis said growing confidence in AI’s technical path has led him to narrow his AGI timeline from 2030–2035 to 2029–2030 — a shift with major implications for businesses, investors, and governments worldwide.
What Happened?
AGI has moved from science fiction to a near-term planning reality. Speaking at Google I/O 2026 and in a post-conference interview with Axios, Demis Hassabis — CEO of Google DeepMind and one of the world’s most credentialed AI researchers — tightened his prediction for when AGI will arrive. He now places the date at “2030, plus or minus a year,” making 2029 an explicit possibility on his timeline.
The announcement was significant not just for its content but for its tone. Hassabis described humanity as standing in the “foothills of the singularity,” a term borrowed from futurist theory to describe the point at which AI systems become smarter than humans and begin improving themselves. That threshold, he warned, is closer than most people outside tech circles appreciate.
“We can see agents really happening now and imagine what they will be in another year, and how useful they’ll be,” Hassabis told Axios. The AGI race, he suggested, is no longer a distant horizon — it is the next chapter in a story already well underway.
This represents a meaningful acceleration. As recently as late 2025, Hassabis had spoken of AGI arriving between 2030 and 2035. The updated window of 2029–2030 reflects what he called “growing confidence that the industry has found the right technical path.”
Why It Matters
The shift in Hassabis’s AGI timeline is not merely a forecast update from one executive. It signals a broad change in how AI’s top builders are speaking publicly about where the technology is headed — and how fast. For investors, policymakers, and business leaders, the 2029 scenario represents a planning horizon that is now inside a single business strategy cycle.
Hassabis chose his language deliberately. “This is partly why I use some of the terms I used, yeah, which were a little bit provocative,” he told Axios, acknowledging that phrases like “foothills of the singularity” were calculated to push governments and economists toward urgency rather than complacency.
He also raised a specific concern: recursive self-improvement. “All the leading labs are quite focused on that,” he said, describing systems capable of materially accelerating their own development. While AI has not yet reached that threshold, he noted a softer version is already visible: “These coding agents are making engineers much more productive.” The jump from soft self-improvement to hard recursive self-improvement could be sharp and fast.
At the same time, Hassabis acknowledged the risk of society being caught unprepared. He cited how Anthropic’s Mythos model caught businesses and governments off-guard as a recent example. “It was probably a good warning shot across the bow,” he said. He called for federal-level safety mandates — specifically mandatory testing before new models are released — and said discussions with other top AI lab leaders are ongoing, though he declined to offer specifics.
Expert Analysis
Narrowing the Window
The compression of AGI timelines is not limited to Hassabis. According to a June 2026 analysis by MEXC, AGI predictions across the industry have narrowed from 2060 to 2033 over just six years. What was once a 40-year problem is now being treated as a 5-to-10-year engineering challenge. This convergence of expert opinion is itself a data point: when the builders of the most advanced AI systems start agreeing on timeframes, markets and regulators tend to follow.
Hassabis’s prediction carries extra credibility given his background. He co-founded DeepMind, which produced AlphaGo (the first AI to beat a world champion Go player), AlphaFold (which solved protein structure prediction), and is now at the frontier of AI agents and reasoning models. He is not speculating about a technology he doesn’t build. His AGI estimate comes from inside the lab doing the work.
AI Agents as a Rehearsal
One of Hassabis’s most striking framings was calling today’s AI agents a “practice run” for AGI. The implication is that everything happening now — agentic AI systems booking travel, writing code, managing tasks — is a societal rehearsal for the moment when a system crosses the AGI threshold. Each time an AI agent surprises users with capability, society gets a small preview of what full AGI disruption might look like at scale.
AGI and Market Impact
AI Spending on a Steep Curve
The financial stakes of the AGI race are enormous. Gartner forecasts worldwide AI spending will reach over $2 trillion in 2026, rising to $3.3 trillion by 2029 — a compound annual growth rate of approximately 22%. The AI agent market alone is expected to grow from $7.84 billion in 2025 to over $52 billion by 2030, representing a 46.3% CAGR, according to Vention Teams’ State of AI 2026 report.
Enterprise adoption is accelerating in parallel. Gartner predicted that 40% of enterprise applications will feature task-specific AI agents by 2026, up from less than 5% in 2025. For technology companies, semiconductor manufacturers, cloud providers, and any business relying on data or knowledge work, these numbers represent both opportunity and competitive pressure.
The AGI narrative also directly affects capital flows. Companies seen as leading the AGI race — including Google DeepMind, Anthropic, and OpenAI — continue to attract record investment. Qualcomm is in early talks to acquire AI chip startup Tenstorrent for $8–10 billion, a move that signals large-scale commitment to AI infrastructure. Meanwhile, Meta is restructuring around AI with 7,000 employees reassigned to AI-focused teams even as it reduces its broader headcount.
According to YouTube coverage by AI research channel Demis Hassabis — Three Quarters of the Way to AGI (posted April 2026), the consensus view inside top labs is that the remaining AGI challenges are primarily about scale, alignment, and recursive self-improvement — not fundamental scientific breakthroughs. That framing shifts the question from “if” to “when,” with capital markets repricing accordingly.
AI Perspective
AGI raises questions that go beyond markets or policy — it touches on what it means for human work, creativity, and decision-making to be replicated at scale by a machine. Hassabis has been clear that he views AGI as having the potential to solve humanity’s greatest problems — from drug discovery to climate modelling — but equally clear that the risks of deploying it without adequate safety infrastructure are severe.
His call for mandatory model testing before release echoes proposals from AI safety researchers who argue that the speed of capability development currently outpaces the speed of alignment research. The question for 2026 and beyond is not whether AGI will arrive — it is whether the legal, economic, and ethical frameworks to manage it will be ready in time. Hassabis’s narrowed timeline suggests the answer to that question demands more urgency than it is currently receiving.
Frequently Asked Questions
What is AGI?
AGI (artificial general intelligence) refers to a hypothetical AI system capable of performing any intellectual task that a human can. Unlike today’s AI systems, which excel at specific tasks (language, image recognition, code), an AGI would possess broad, flexible reasoning across all domains without task-specific training.
When does Demis Hassabis predict AGI will arrive?
Hassabis said at Google I/O 2026 that he expects AGI to arrive around 2030, “plus or minus a year,” making 2029 a real possibility on his timeline. This narrows his earlier estimate of 2030–2035, which he gave in late 2025.
What does “foothills of the singularity” mean?
Hassabis used the term at Google I/O 2026 to describe the current moment in AI development — a phase close to but not yet at the technological singularity, the point where AI begins rapidly self-improving beyond human-level intelligence. He was warning that society is not prepared for how quickly this transition may happen.
How will AGI affect businesses and investors?
Gartner projects AI spending will reach $3.3 trillion by 2029. Companies at the frontier of AI development — and those that adopt AI agents early — are likely to gain significant competitive advantages. Businesses that delay AI adoption may find themselves unable to compete as AGI-level systems begin automating knowledge work at scale.
What safety measures are being discussed?
Hassabis supports mandatory AI model testing before release, a policy he described as being discussed among leaders at top AI labs. He also called on governments to accelerate AI safety work, saying “safety needs to be accelerated” and that the current moment is “a good time to strike while the iron is hot.”
Conclusion
The AGI prediction from Demis Hassabis is not a forecast made in isolation — it is a signal from one of the most credible voices in AI that the timeline for transformative artificial general intelligence has compressed significantly. For investors, the message is about capital allocation and competitive positioning in an AI-first economy. For policymakers, it is a call to build safety infrastructure before the window closes. For businesses, it is a reminder that AI adoption is no longer optional.
Whether AGI arrives in 2029 or 2031, the preparation required is the same — and according to Hassabis, the preparation needed is not happening fast enough. The “foothills of the singularity” may be a dramatic phrase, but the substance behind it is straightforward: the world is closer to AGI than most people outside tech circles realise, and the time to act is now.
Sources
- Axios: DeepMind CEO: AI agents are a “practice run” for AGI
- Sherwood News: Hassabis: AGI is 3 to 4 years away
- Fast Company: Demis Hassabis isn’t shying away from AI’s biggest questions
- Vention Teams: State of AI 2026
- YouTube: Demis Hassabis — Three Quarters of the Way to AGI (Sequoia, April 2026)
This article is for informational purposes only and does not constitute financial or investment advice.
