When will any company achieve AGI?

My independent view is that this is a live but far from favored outcome. The key distinction is that the market is not asking whether AGI truly exists in some objective sense, but whether any company will choose to publicly say it has achieved AGI before Oct. 1, 2027. That makes the bar easier than consensus-grade scientific validation, and it also means a company could move first if it believes it has crossed a sufficiently broad capability threshold.

The strongest case for Yes is that elite AI leaders have materially shortened their public timelines, with some pointing to the 2026 to 2028 window, and competitive dynamics create a strong incentive for a lab that thinks it has crossed the line to claim the milestone. The lack of a universally accepted AGI definition matters a lot here: a company does not need to persuade all experts, only enough observers to accept that its own standard has been met. In a fast-moving race, that can bring an announcement forward well before a broad research consensus would normally form.

The strongest case against Yes is that broad expert surveys still cluster much later, and the technical leap from impressive frontier models to systems that are robustly general, autonomous, and reliable across domains is still uncertain. Even if capability improves quickly, companies may avoid the AGI label because it invites regulatory scrutiny, public backlash, and heavy claims-testing. Compared with the current market price of 34%, I think the market is a bit low on announcement risk because self-serving or loosely defined claims are more likely than a strict scientific consensus would suggest, but the short time horizon keeps this well below a coin flip.