OpenAI CEO: The cost of running AI at all levels will drop 10 times per year

Reprinted from jinse
02/10/2025·14DAuthor: Martin Young, CoinTelegraph; Translated by: Tao Zhu, Golden Finance
OpenAI CEO Sam Altman said the cost of using any specific level of artificial intelligence will drop by about 10 times each year and could lead to a significant drop in commodity prices.
"The cost of using a specific level of AI drops about 10 times every 12 months, and the price drop will lead to more use," OpenAI CEO Sam Altman said in a February 9 blog post on the economics of artificial intelligence. ”
Altman mentioned that the cost has dropped by about 150 times from the company's GPT-4 model in early 2023 to the GPT-4o in mid-2024.
This is much faster than Moore's Law, he said, which refers to the number of transistors in integrated circuits doubling about every two years, thereby increasing processing power, efficiency and reducing the cost of electronic devices.
“In some ways, artificial intelligence may be economically like a transistor—a major scientific discovery that has good scalability and penetrates almost every corner of the economy,” he said.
Altman predicts that prices for many commodities will eventually drop sharply, adding: “Currently, intellectual costs and energy costs restrict a lot of things.” However, the prices of luxury goods and some inherently limited resources, such as land, “may rise.” Faster,” he said.
According to Moore's Law, the evolution of transistors. Source: ResearchGate
Altman said he is willing to try various ways to make AI reach everyone around the world, such as providing a "computing budget."
“We are willing to take on some weird-sounding ideas like providing some 'computing budget' that allows everyone on the planet to use a lot of AI, but we also see that as long as we persist in minimizing the cost of intelligence, Can produce the expected effect."
Altman concluded that continually reducing AI costs will help democratize AI capabilities, with the goal of anyone getting the same intelligence as everyone in 2025 by 2035.
“Everyone should be able to use infinite genius to guide everything they can imagine.”
In January this year, the latest low-cost AI model launched by Chinese developer DeepSeek shocked the stock market, with American companies that produce high-cost hardware such as Nvidia being hit hard, and AI costs became the focus of people's attention.
Meanwhile, Chinese automakers, technology companies and leading telecommunications companies have integrated the DeepSeek AI model into their products, Reuters reported on February 9.