Adam Mosseri says Instagram has 'reined in' AI costs after shutting down 'the silly things' that were burning tokens

In a recent appearance on "Lenny's Podcast," Instagram CEO Adam Mosseri revealed that the company has successfully reduced its artificial intelligence expenditures. He explained that this was achieved by eliminating unnecessary projects, which he referred to as 'silly things.' While he remained vague about the specifics of these projects, he emphasized that managing costs was not overly challenging. When discussing token leaderboards, a feature that was part of the early tokenmaxxing trend, Mosseri expressed strong disapproval, labeling it as 'a terrible idea.' This comes amid broader concerns in the tech industry about rising AI expenses, particularly as organizations increasingly adopt AI technologies for complex tasks. Mosseri noted that as more employees utilize AI, the demand for tokens—the fundamental units of data for large language models—will likely grow. This could lead to increased costs for Instagram. However, he remained optimistic that, as competition intensifies among companies, pricing for AI services could eventually decrease. He explained that Instagram's budgeting now incorporates AI tokens in a manner similar to how resources like GPUs and RAM are allocated. Mosseri predicted that in the near future, the cost associated with hiring skilled engineers could be comparable to the expense of token usage. He mentioned that if spending on tokens escalates significantly, it would be reasonable to consider implementing caps on usage, ensuring they align with a company's confidence in achieving a positive return on investment. Furthermore, Mosseri shared insights into changes within Instagram's team structure. Previously, teams consisted of around thirteen members, in line with the well-known 'two-pizza rule' championed by Amazon's Jeff Bezos. Currently, teams—referred to as 'pods'—have shrunk to six or seven members. This shift allows for quicker decision-making and improved collaboration, as fewer individuals are involved in the process. The new structure includes a mix of engineers with broader skill sets and a product staff member, indicating a departure from traditional roles. Mosseri concluded by highlighting the ongoing evolution within the tech industry and the blending of job functions as companies adapt to the growing influence of AI technologies.