SoftBank’s chief says AI is already shaping the design of future models, signaling a move toward artificial superintelligence where machines begin designing their own successors.

In an interview, he warned engineers may lose the ability to design next-gen AI once AI systems start producing subsequent models, marking the dawn of superintelligence.

He described this shift as the beginning of a regime where AI is exponentially smarter than humans and participates in creating its own next generations.

He has updated his timeline to reflect a faster pace for these developments than previously expected.

The forecast is framed within SoftBank’s broader AI investment footprint, including advances like AI-powered personalized news feeds and more interactive search and notification experiences.

These forecasts come amid safety concerns about rapid AI progress, highlighted by scholarly warnings from Anthropic about risks in accelerated development.

SoftBank aims to invest about $65 billion in OpenAI for roughly a 13% stake, having committed $41 billion of that by the end of 2025.

SoftBank’s stock has surged in 2026 on optimism around AI tech and related infrastructure projects such as Stargate AI.

The company frames the current AI revolution as exponentially more impactful than the internet boom, predicting it to be roughly 50 times larger.

OpenAI accounts for a little over 20% of SoftBank’s net asset value, underscoring the scale of its stake in the AI wave.

Son reiterates that the AI revolution will be fifty times bigger than the dot-com era, signaling sweeping technological and societal change.

He personally uses ChatGPT daily and says AI already surpasses him in many areas, with expectations that AI could outperform humans in the bulk of subjects soon and be about ten times more capable in those domains.