There are all manner of predictions on when AGI will be reached, but a prominent investor believes that it’s already here.
Marc Andreessen, co-founder of venture capital firm Andreessen Horowitz (a16z), made a striking claim on The Joe Rogan Experience: that artificial general intelligence — AI that is as smart as a person — has already been achieved. And it happened roughly three months ago.
“The technology advances in the last three years have been mind-boggling — crazy, amazing, impressive,” Andreessen said. “People talk about this concept called AGI, which means artificial general intelligence, which basically means an AI that’s as smart as a person. And I actually think we crossed that about three months ago.”
He pinpointed the breakthrough to the most recent releases from the leading AI labs. “I think it was with the very latest versions of the leading models,” he said.
Andreessen also offered an explanation for why so many people haven’t noticed. “One of the reasons people are having a hard time understanding what’s happening in AI is because it’s moving so fast that if you don’t use the latest thing, you don’t understand what’s happening — because you’re not seeing it. A lot of people used ChatGPT last year or the year before, and they’re not actually seeing the new thing.”
The “new thing,” he specified, is a cluster of recent frontier model releases: GPT-5.5 from OpenAI, Claude 4.6 from Anthropic, Gemini 3.0 from Google, and Grok 4.3 from xAI. “In each case, I think with those releases, they kind of hit this threshold where all of a sudden…” he said. “In my line of work, 99% of the time, the answer I’m getting from the most advanced AI models is better than I would get from talking to basically almost any expert I have access to.”
That’s a significant claim from someone with access to some of the best minds in technology and business.
Andreessen’s remarks add his voice to a fast-growing chorus of tech leaders willing to call, or nearly call, the AGI moment. OpenAI’s VP of Research Aidan Clark has hinted that AGI may have already arrived, and Elon Musk has suggested AGI could come by end of 2026. Anthropic CEO Dario Amodei has reportedly placed AGI arrival in the 2026–2027 window. Not everyone is in a hurry to plant a flag, though — Google DeepMind’s Demis Hassabis has argued that current systems still fall short because of inconsistencies and easily exploitable weaknesses, and AI pioneer Fei-Fei Li has called AGI more of a marketing term than a scientific one, noting that no one has ever agreed on a single definition. The lack of a shared benchmark makes the debate hard to settle — and easy to win by simply moving the goalposts.
What’s harder to dispute is Andreessen’s secondary point: most people are evaluating AI on outdated versions. The gap between frontier models and the ones most users interact with daily has never been wider, and it’s growing. If the leading labs have genuinely crossed a qualitative threshold in the last few months, the public and most enterprises are still catching up to the previous generation. That lag in perception may matter as much as the underlying capability — at least for now.