Evan Greenberg, Chairman and Chief Executive Officer (CEO) of global insurer Chubb, said the company’s AI token costs represent only a minor fraction of the efficiencies and improvements it gains.
During Chubb Limited’s Q2 2026 earnings call, Greenberg was asked how the company can balance the rising costs of technology deployment with its ability to capture savings and improve margins.
The question comes as businesses grapple with rising artificial intelligence (AI) costs, with a recent J.P. Morgan analysis stating that these expenses are being driven by the price of AI “tokens”, prompting companies to reassess whether productivity gains outweigh the expense.
However, Greenberg said Chubb’s experience has been different, with the insurer finding that the costs associated with AI token usage remain small relative to the efficiencies, insights, and improvements generated by the technology.
“That token usage is really about the vast token usage among tech companies. AI and tech companies use vast amounts in model development. That comment is not really applying to general businesses,” said Greenberg.
He explained that Chubb tracks its token costs as part of its broader expense management framework.
“We know our token costs. Frankly, it’s within our economic model and how we measure expenses. Our token costs and the usage of that is a minor fraction relative to the efficiencies and the insights and the improvements that we gain, and we measure it in hard dollars.”
Later in the call, Greenberg also discussed technology use, stating that “I think that with technology, the data, and scale, and size, and the breadth of capability that brings you, and insight is a competitive advantage,” adding that he believes this represents “a structural, secular advantage.”
As AI token costs rise and companies seek to extract efficiencies from AI deployments, there is also a notable shift away from large foundation models towards cheaper open-weight and open-source models, whose capabilities are catching up fast.
