Sam Altman says AI is an economic force like electricity, requiring products and computing infrastructure
Sam Altman compared AI's economic role with electricity and transistors, saying usable products and vast computing infrastructure are needed to turn research into public benefits.

AI is becoming an economic force comparable to electricity, enabling businesses, technologies and scientific discoveries, Sam Altman said. Turning research into benefits for society requires both usable products and computing infrastructure on a vast scale.
Speaking at Stanford Graduate School of Business on October 21, OpenAI co-founder and CEO Altman said AI had developed differently from the team's initial expectations, requiring it to build a large product company beyond its origins as a research lab.
Altman said the early assumption was that developing superintelligence would lead to people using it and benefiting from it, without the team having to work out how to deliver it effectively. As the technology developed, building products became an essential part of making the research useful.
He described that development as a better outcome because AI increasingly functions as infrastructure that supports activity elsewhere in the economy. Comparing it with electricity and transistors, Altman said making AI available required products such as ChatGPT alongside computing capacity larger than any project humanity had previously undertaken.
By late 2020 or early 2021, he said, it had become clear that OpenAI would need to build a large company. The speed of that transition was unexpected, with ChatGPT's launch quickly convincing the team that the business would grow very rapidly.
Behind that decision was his confidence in the effects of scale. Altman said the AlexNet result in 2012 and subsequent progress by 2014 showed that deep learning improved across several dimensions as it grew. Although the team did not yet understand how predictable those improvements would become, the results gave it reason to keep pursuing the research.
Altman acknowledged that conviction sometimes faltered, with the team repeatedly wondering whether it had reached a limit. He said his experience of unexpected gains from scale helped sustain the decision to continue.