When will any company achieve AGI?

The key question is not only whether systems will become highly capable by late 2027, but whether a company will be willing to make the loaded claim that it has achieved AGI. The recent evidence points to faster-than-expected progress at frontier labs, with prominent leaders sounding more aggressive about timelines than they did a few years ago, and the most likely announcers would be firms already inclined to frame breakthroughs as near-AGI milestones. If a major lab reaches a clearly general, economically valuable agent that outperforms humans across a wide range of tasks, a public AGI declaration becomes quite plausible.

Even so, the bar for a public announcement is higher than the bar for a technically impressive system. AGI has no single accepted definition, and the term carries major reputational, regulatory, and competitive consequences, so companies may prefer softer language such as general-purpose reasoning, advanced agents, or frontier-level intelligence even if their internal teams believe they are near the line. Broader survey-based forecasts still cluster much later than 2027, which means the market is effectively betting on an unusually fast compression of capabilities and on a company choosing to say so publicly rather than waiting for consensus.

Against the current market price of 34%, my view is slightly more conservative. The market is correctly recognizing that announcement risk is easier than definitive scientific proof, but I still think it overweights the chance that a company both crosses a sufficiently convincing threshold and decides to use the specific AGI label before the deadline. That combination feels more like a one-in-four to one-in-three outcome than a one-in-three to two-in-five outcome, so I see modest downside to the current price rather than a major mispricing.