Instagram Trims AI Projects to Curb Rising Costs
Instagram is scaling back its artificial intelligence expenditure by eliminating inefficient projects, CEO Adam Mosseri said. Speaking on a recent episode of “Lenny’s Podcast,” Mosseri noted that while AI remains a core focus, the company is reining in costs by cutting “silly” initiatives and moving away from industry trends like token-consumption leaderboards.
Rethinking Metrics and Infrastructure
Enterprise AI spending has become a focal point for tech leadership as organizations face mounting bills from high-frequency usage and complex, agentic AI tasks. Mosseri described the process of managing AI expenses as similar to allocating traditional infrastructure resources like GPUs, storage, and RAM.
“We’ve managed to get the costs reined in a little bit by shutting down the silly things that we were doing,” Mosseri said. He dismissed the practice of using token-consumption leaderboards—a trend once popular in Silicon Valley for tracking AI usage—as a “terrible idea.”
The Looming AI Pricing War
While Instagram’s AI costs may continue to climb as employee adoption grows, Mosseri anticipates that broader market competition will eventually drive down pricing. “All of these frontier models are going to be in a bit of a pricing war,” he said. “So we’ll see. I think it’ll be a bit of a roller coaster.”
Budgeting for Future Consumption
As AI integration matures, Mosseri suggested that companies may eventually need to implement formal budget caps for AI consumption. He compared the potential future cost of AI tokens to the current cost of employment for an engineer.
“I think that you can imagine, at least in a year or two coming, that the burn rate of a strong engineer might be the same as their salary or their cost of employment,” Mosseri noted. He added that while Instagram has not yet implemented hard caps, he believes they could be a “healthy” way to ensure that AI usage remains tied to positive return on investment.
Restructuring into Lean Generalist Pods
Beyond budgeting, Instagram has significantly reorganized its internal team structure. Mosseri stated that the company has moved away from larger, specialized teams—previously consisting of about 13 people—to smaller “pods” of six or seven employees.
This shift has changed the composition of individual teams. In the past, a typical team included a mix of specific engineers, data scientists, and project managers. Now, Instagram is prioritizing generalist roles. A typical pod currently consists of four to six generalist engineers, a member of the product staff, and one specific specialist relevant to the project’s immediate needs.
“What’s clearly happening is all of the functions are starting to bleed into each other, and the whole industry’s wrestling with what that means,” Mosseri said. He credited these smaller, leaner teams with the ability to move faster and make decisions with less “design by committee.”