OpenAI’s next model is being designed by AI itself, says SoftBank’s Masayoshi Son

SoftBank's Masayoshi Son claims OpenAI's next AI model is being designed by AI itself rather than humans, suggesting accelerated development toward superintelligence. This development could fundamentally reshape industries and societal structures within years if the claim proves accurate.

SoftBank's Masayoshi Son's assertion that OpenAI is using AI to design its next generation model represents a significant inflection point in artificial intelligence development. If accurate, this indicates the field has reached a stage where machine learning systems can autonomously optimize their own architectures, compressing development timelines and potentially accelerating capability gains. This practice, sometimes called recursive self-improvement, moves beyond traditional human-guided machine learning toward systems that improve themselves with minimal human intervention.

The significance of this development lies in its implications for the AI capability curve. Historically, model improvements required substantial human expertise in architecture design, hyperparameter tuning, and training methodology. Automating this process could exponentially increase the pace of advancement, particularly if AI-designed models prove more efficient or capable than human-designed alternatives. This aligns with broader industry trends toward neural architecture search and automated machine learning, but at a scale and sophistication level that suggests approaching more general capabilities.

For markets and stakeholders, this development carries dual implications. Investors in AI infrastructure, compute resources, and enterprise AI solutions could see accelerated adoption timelines and productivity gains. However, it also intensifies concerns around AI safety and control, as faster development cycles compress timeframes for alignment research and safety testing. The claim underscores why regulatory frameworks remain incomplete—the technology is advancing faster than governance mechanisms can respond. Technology companies relying on AI moats face pressure to match this development velocity.

• →AI-designed AI models compress development timelines and potentially accelerate capability improvements beyond human-designed systems.
• →The claim reflects broader industry movement toward automated machine learning and neural architecture search at scale.
• →Faster AI development cycles heighten safety and alignment research urgency as regulatory frameworks lag behind technological advancement.
• →Markets for AI infrastructure, compute resources, and enterprise AI solutions could see accelerated growth but face increased competitive pressure.
• →The development underscores persistent gaps between AI capability advancement speed and governance mechanism readiness.