Marketers Keep Buying AI. The Skills Part Is Still Loading.

Digiday+ Research’s latest report, based on a survey of 142 brand and agency professionals plus interviews, shows AI investment moving from “nice-to-have” to “budget line item someone now has to explain.” Marketers reporting company investment in AI rose from 44% in 2022 to 57% in 2023, 71% in 2024, and 86% in 2025.

The org chart is also starting to show the shift. In 2024 and 2025, brands including General Motors, Mastercard and ZocDoc appointed chief AI officers, as did agencies Golin, Luckie & Co. and Horizon Media, according to Digiday.

On the consumer side, brands are already putting AI where customers can poke at it. PetSmart relaunched its membership program using AI to tailor deals based on past purchases. Guitar Center rolled out Rig Advisor, a chatbot meant to help shoppers pick products.

But the report’s central tension is that adoption is rising faster than capability. Code and Theory co-founder Dan Gardner argues the industry is mistaking “learning a new tool” for actual upskilling. M7 Innovations founder Matt Maher points to the gap between basic familiarity with tools like ChatGPT, which he says has 800 million weekly active users, or Gemini, at 400 million monthly, and using those systems “to their utmost potential.”

That gap gets louder once the infrastructure bill shows up. Huge’s global CTO Marc Maleh frames it as ROI whiplash: marketers jumped at “press release” AI, then met the real costs of GPUs, TPUs and API calls — plus questions like what “500 more seats” of Claude Code costs and whether a “30% productivity increase” pays it back.

Digiday’s survey suggests most teams are still buying more than building. Eighty-five percent of respondents said their company uses out-of-the-box AI tools, while 40% build proprietary tools on top of existing LLMs and 19% build and train their own LLMs.

The tool landscape is also getting more modular. Maleh points to out-of-the-box models from Adobe and Google, including Google’s Vertex, and to integrations that let users bring Google models into Adobe’s creative tools. Maher describes a “tech stack” era where brands might use Copilot as a base, stack Claude on top, or pair Adobe Firefly with Canva.

Which is less tidy than a single walled garden, but probably more accurate to how marketing technology usually evolves: one stack, several subscriptions, and at least one spreadsheet trying to explain why all of it is necessary.

A series of dominoes standing on a green background, with one domino in the process of falling.

Read more at Digiday.