Masayoshi Son says AGI will be 10x smarter than all humans by 2030

Masayoshi Son really does not want you to become a goldfish.

Speaking at the annual SoftBank World event on Wednesday, the SoftBank chief stood in front of a giant slide showing a goldfish in a bowl. He told the room that anyone avoiding AI risks falling as far behind as apes are to humans.

His core timeline is ambitious: Son predicted that artificial general intelligence (AGI) will surpass all human knowledge by the end of the decade. By 2030, he claimed, AGI will be 10 times more powerful than the collective intelligence of all humanity combined.

Big bets and missed calls

According to a report from The Wall Street Journal, Son believes this wave will bring a complete shift to autonomous driving and yield scientific breakthroughs worthy of Nobel Prizes.

His speech comes right after Arm's massive IPO. Reports last month indicated Son is now looking to deploy tens of billions of dollars directly into generative AI companies.

Of course, Son has a mixed track record with these sweeping predictions. As Reuters noted, he nailed the mobile internet boom, but his massive bets on the Internet of Things never quite panned out the way he promised.

He is also not alone in hyping the timeline. Microsoft researchers claimed earlier this year to have seen sparks of AGI in modern models, though many independent researchers remain skeptical, arguing machines still lack true comprehension of the physical world.

Corporate leaders are chiming in as well. JPMorgan Chase CEO Jamie Dimon predicted this week that AI could shrink the workweek to three and a half days and transform cancer research, while acknowledging the severe downside risks from malicious actors.

Why it matters

Son is doing what he has always done best: creating urgency to justify pouring tens of billions into a single market narrative.

We think calling non-users goldfish is clever marketing for his investor base, but banking on a tenfold human-level superintelligence by 2030 overlooks serious unsolved hurdles. Current large language models still hallucinate and struggle with physical world reasoning.

For everyday professionals and founders, the right move is not to panic about existential replacement by 2030. The immediate priority is learning how to operationalize existing narrow AI in your daily workflow.

Before AGI sweeps Nobel Prizes and clears human drivers off the road, the near-term consequences will look like Jamie Dimon and fraud experts warned: voice cloning scams in under three seconds and intense workplace pressure to produce more with smaller teams.