Google DeepMind chief says AI marks foothills of the singularity

Demis Hassabis used Google I/O’s closing stretch in Mountain View to frame Google’s AI campaign as something larger than a routine product update: the “foothills of the singularity.” In plain English, that is the early slope of a climb technologists believe could lead to machines that move beyond human-level performance. Hassabis, who leads Google DeepMind and shared the 2024 Nobel Prize in Chemistry for AlphaFold, gave the remark scientific heft even as the claim remained fundamentally speculative.

The setting mattered. Google I/O 2026 took place at Shoreline Amphitheatre in Mountain View, California, and online at io.google, with keynote addresses, fireside chats, product demos and updates across Gemini, Android and other lines. Google said the company was releasing two new models, Gemini Omni and Gemini 3.5, making clear that the event was not just about philosophical talk but about shipping products into a market where AI has become the core of Google’s strategy.

That message echoed Sundar Pichai’s opening keynote, which described the past year as a period of “relentless shipping” and said AI remained central to Google’s mission. Google has also been trying to anchor its AGI language in scale. In its Gemini 3 announcement, the company said it was taking “another big step on the path toward AGI,” and pointed to product adoption that is already measured in the billions: AI Overviews now reach 2 billion users every month, while the Gemini app has surpassed 650 million monthly users.

For investors, competitors and policymakers, the practical question is not whether a slogan lands, but whether the underlying technology keeps improving in ways that matter economically. Google DeepMind has long used AlphaGo and AlphaFold as reference points for what advanced models can do when they solve narrow but difficult problems. Hassabis has also tied AGI to science, healthcare and productivity in recent public appearances, signaling that Google wants AI viewed not only as a consumer feature but as infrastructure for research and work.

That is where the gap between marketing and measurable progress becomes important. Google’s new model launches and rising usage figures show a real commercial base. Hassabis’s “foothills of the singularity” language points to a much bigger wager: that current gains are still the opening stage of a deeper transformation. Whether that transformation arrives on the timeline technologists suggest, or at all, remains the central unanswered question.