TheCorporateCounsel.net

You really can’t go anywhere these days without talking about AI – and as you might expect, that’s especially true of earnings calls and SEC disclosures. But as the landscape changes, so do the questions and conversations. This season, as folks get more sophisticated about the use cases and implications of AI, analysts for some companies are asking about token expense – as well as ROI. This Bloomberg article flags the trend. Here’s an excerpt:

The cost associated with using artificial intelligence has gone up sharply for many businesses, with [big-name corporations] limiting access to AI tools to rein in spending. After pushing employees to make use of AI applications, companies are finding that additional productivity comes with rising expenses — and growing scrutiny from analysts and investors.

The article gives examples of how companies are addressing these questions, so it’s worth a read if you are preparing for an upcoming call. It also points out:

There’s no reporting requirement for companies to break out these costs. Still, chief financial officers face the challenge of deciding how much of that spending (and its returns) they want to disclose to the outside world.

Companies are taking different approaches to reporting token-spend – e.g., some are breaking it out into a separate category and implementing controls to track it while others are discussing the business impact of AI in a more holistic way, in light of the cost savings it may also bring. Likewise, some companies are discussing margin and ROI in general terms while others are not yet providing that information. And as this WSJ article shares, some companies are also getting more “tokenomical” – shifting budgets to low-cost and/or open-source models when available in order to tamp down expenses.

Like any emerging issue, AI-related spending is also triggering a conversation about quarterly risk factors. In addition to what you might find about cap-ex, pre-AI risk factors about access to service and availability of adequate tech infrastructure, network capacity and compute power could serve as a model for risks relating to power and compute. Of course, every company will need to think through whether there are material risks and – if so – describe them in a way that’s tailored to their particular business and circumstances. In the past, these types of risk factors were mainly found in tech companies’ filings, but now we might see them in other industries as well.

– Liz Dunshee