If you have used AI tools like ChatGPT or similar services, you have probably heard of tokens. Tokens represent chunks of text that AI systems analyze, and they serve as the basis for measuring usage and billing. The current market remains nascent, and during the first half of this year, escalating expenses for tokens led many organizations to scale back their AI investments.
Tokens function as the fundamental unit of measurement in large language models, with each token representing roughly four characters of text. As enterprises integrate AI into their workflows, the cost per token directly impacts budgets, leading to the development of monitoring tools and managed services that shield end users from complexity.
Some companies are already trying to fix that. Ramp recently unveiled a token spend tool, and Harvey, a legal AI startup, manages tokens for its customers instead. These products exist because the problem is real.
However, Bret Taylor - who co-founded the AI firm Sierra and chairs OpenAI - believes these difficulties are short-lived. According to Taylor, many of the "tokenomics" challenges arise because the AI industry has not yet matured.
Get the market news that matters in a five-minute read with Market Briefs, our free daily newsletter
Bret Taylor is not just any AI insider. As chairman of OpenAI, he sits at the center of the industry. And his prediction is striking: within 12 months, the average company's IT department will be so comfortable with AI that tokens become an afterthought.
"I believe where the world is going is paying for outcomes," Taylor said. During a Monday CNBC appearance, Taylor stated that eventually, other firms will handle token management.
"I think if you fast-forward 12 months from now, IT departments will be really sophisticated about the industrial applications of AI," Taylor added. "So you just don't need to think about the word token at all."
Many observers anticipate that token prices will decline as AI models improve in efficiency. A good example is the Kimi K3 model, released last week by Chinese startup Moonshot. The model drew Silicon Valley's attention because it offered performance comparable to leading models from OpenAI and Anthropic but at a lower cost.
Taylor is not fully convinced yet that Kimi K3 is truly cheaper in practice. Taylor said, "Is it cheaper to use is the most important part for anyone considering it." He added, "And I think the jury's out on that."
"We're just in the early stages of the technology, and I think it's a call for entrepreneurs to develop solutions so businesses don't need to deal with this stuff," Taylor said.
Join Market Briefs, our free daily newsletter, for a quick daily rundown of the markets
