Three years ago, Demis Hassabis was telling audiences that artificial general intelligence was five to ten years away. In March 2025, he narrowed it to 2030 to 2035. Then, after Google I/O 2026, he told Axios that he now sees 2030 as his central estimate — with 2029 as a genuine possibility.

That is a significant compression in a short time. And for businesses trying to figure out their AI strategy, it raises a practical question: does the timeline matter for decisions you need to make right now?

What Hassabis Actually Said

Speaking to Axios on May 26, following Google’s annual developer conference, Hassabis described the current moment as being at the “foothills of the singularity.” He framed the agentic AI systems Google unveiled at I/O — Gemini 3.5, Antigravity, and the broader push into autonomous agents across search, coding, and productivity tools — as a rehearsal for what comes after.

His prediction is not that everything changes overnight in 2029. It is that the technical path to AGI is clearer now than it was even a year ago, and the remaining gaps are narrower than most people outside the field appreciate.

For context: in March 2025 he was publicly framing human-level AI as a five-to-ten-year development. He is now saying three to four.

Why the Timeline Compression Matters

You could reasonably argue that AGI timelines do not affect day-to-day business decisions. If AGI arrives in 2029 or 2032 or 2036, the things a business should do with AI today are largely the same: build data foundations, train teams, deploy agents in high-ROI workflows, and keep iterating.

That argument has merit. But the timeline compression tells you something important about the rate of change in the near term, even before AGI arrives.

Hassabis connected his revised prediction directly to agentic AI — the multi-step, tool-using, autonomous AI systems that are already being deployed in enterprise environments today. The capabilities gap between a 2024 language model and a 2026 agentic system is already substantial. If the pace of progress that produced that gap continues for another three years, the systems available in 2029 will be dramatically more capable than what businesses are deploying right now.

That matters in two ways:

1. The competitive window is compressing. The advantage from early AI adoption has historically been measured in years. If the rate of capability improvement is accelerating, businesses that have not yet built meaningful AI operations — real agent workflows, not just copilot buttons in their software — are losing ground faster than the calendar suggests.

2. The skills gap is becoming urgent. When AI systems become significantly more capable, the limiting factor shifts from “what can the AI do” to “do we have people who know how to use it.” The organizations that invested in data literacy, AI fluency, and operational AI skills in 2024 and 2025 are already in a better position. By 2028 or 2029, that gap will be harder to close quickly.

The Readiness Question

Hassabis is not alone in his revised timeline. Several prominent AI researchers and leaders have moved their estimates forward in the past 12 months. The specific date matters less than the directional signal: the people who build these systems think the big capability leaps are coming sooner than the broader public expects.

The sensible response to that is not to panic or to bet your business on a specific AGI arrival date. It is to look honestly at where you stand on the fundamentals:

Data foundations. AI systems — agentic or otherwise — are only as useful as the data they can work with. Businesses that build strong data literacy pull ahead on measurable outcomes, and those with clean, accessible, well-governed data are in a structurally better position to take advantage of more capable AI as it arrives.

Operational integration. The difference between a company that has AI in a silo and a company where AI is genuinely woven into operations is large. If your AI deployment is still primarily exploratory — running pilots, testing tools, evaluating vendors — that is a reasonable 2024 position that is less reasonable in 2026. What real agent deployment looks like inside a business is quite different from what most teams picture.

Team capability. This is the hardest one to build quickly. AI fluency — understanding what AI can and cannot do, how to evaluate outputs, how to design workflows around agent capabilities — takes time to develop. The path from Excel to AI capability is a ladder, not a leap, and waiting until you need it is too late.

What This Means for Business

The Hassabis prediction is not a scare story. He explicitly framed it as an optimistic one — a case for building thoughtfully with AI now, so that organizations are ready to make the most of what comes next.

The businesses that will be well-positioned when more capable AI arrives are not the ones that waited to see what happened. They are the ones that built their data foundations, ran real agent deployments, upskilled their teams, and treated AI as a core operational capability rather than a technology experiment.

That is not a reaction to a specific AGI timeline. It is just good strategy — one that happens to become more important if Hassabis is right.

---

Enterprise DNA helps businesses build the data foundations and AI capabilities they need to stay ahead. Whether you are starting with team upskilling through EDNA Learn or deploying AI agents through Omni, the time to start is now.