Employees in tech are some of the smartest and best-paid white-collar workers, but AI models might now be smarter than many of them.
Y Combinator CEO Garry Tan has suggested that the intelligence of OpenAI’s latest models has surpassed that of many individuals he has hired during his career, signaling a pivotal new era for business and employment. Tan’s remarks frame a powerful moment in the evolution of AI, assigning a tangible, human-centric metric to a machine’s cognitive ability.
“This is like actually a really powerful moment because, literally, you have a model that is about 130 IQ,” he stated, referring to OpenAI’s o3 model. He even speculated on its future iterations, adding, “Maybe o3 Pro can be even smarter than that.”
The Y Combinator CEO then drew a direct and startling comparison to the human talent he has evaluated over the years. “When I really think about that, it’s like, ‘Oh yeah. A lot of the people who I’ve ever hired in my lifetime… o3 is smarter than that person now.'”
However, Tan’s analysis quickly pivoted from a simple intelligence ranking to a strategic blueprint for entrepreneurs. He argued that the true power lies not just in the model’s raw intellect but in its application to specialized domains. “You can basically take that and, you know, connect it to the proprietary data systems of almost any niche,” he explained.
He emphasized that the more obscure the field, the greater the opportunity to build a defensible business. “And the more weird and unlikely for someone like someone in this room to know about it, the more likely that will be a durable enough moat that you can get a foothold,” Tan advised. “You can basically get… you can wedge yourself in there, and then basically all you need is a wedge. And then you just basically expand that wedge until you have the pie.”
Tan’s assertion, while jarring, crystallizes a trend that has been accelerating for some time. An IQ of 130 places an individual in the “gifted” range, a level of cognitive ability that AI is now arguably reaching. This is evidenced by the performance of OpenAI’s o3 and Google’s Gemini, which have demonstrated the ability to pass notoriously difficult professional exams, including the IIT exam, the Bar exam and medical licensing boards, often scoring in the top percentiles. Tan’s statement suggests that the latest, perhaps yet-to-be-released models, are pushing these boundaries even further. The broader implication is a fundamental re-evaluation of the value of human intellect in the workplace. His strategic advice points to a future where competitive advantage is gained not by building foundational AI, but by cleverly applying these powerful, generalist models to specific, niche problems. This “wedge” strategy, as he calls it, democratizes the AI revolution, allowing savvy entrepreneurs to create formidable “moats” around their businesses by leveraging unique, proprietary data, effectively carving out their piece of the economic “pie” in an AI-powered world. For the tech workforce, the message is a clear call for adaptation: the future lies in complementing, not competing with, these increasingly intelligent systems.