The cost of artificial intelligence became a central focus at the recent Allen & Co. Sun Valley Conference, with top industry figures urgently seeking ways to make AI technology more affordable. OpenAI CEO Sam Altman emphasized that this year marked a significant shift, as efficiency and return on investment took center stage among participants.
Addressing CNBC during the event, Altman explained that companies are increasingly scrutinizing their AI spending to maximize value. OpenAI’s latest release, the GPT-5.6 family, was designed with these concerns in mind. It includes three variants: the Sol model, aimed at high performance; the Terra model, which balances daily use; and the Luna model, focused on cost efficiency. Altman highlighted that the GPT-5.6 Sol model improved token efficiency by 54% for coding tasks, a key factor in controlling operational expenses.
As the Sun Valley conference brought together influential executives such as Netflix’s Ted Sarandos, Uber’s Dara Khosrowshahi, Apple’s Tim Cook, and Amazon’s Andy Jassy, the topic of AI expenditure remained a major theme. Altman noted that virtually every enterprise represented there is now evaluating how much value they gain relative to their AI investments.
Other tech leaders echoed this trend. Coinbase CEO Brian Armstrong has tested implementing less expensive Chinese AI models as defaults to manage costs effectively, while Vercel CEO Guillermo Rauch recommended companies diversify their AI usage across multiple platforms—including OpenAI, Anthropic, Gemini, and Chinese providers—to optimize spending and performance.