It may take longer than market expectations for AI implementation to lead to increased profit margins.

Torsten Srock, chief economist at Apollo Global Management, points out that investment in AI does not necessarily lead to immediate profit growth for companies, and that outside of the technology sector, it may take a long time for the return on investment (ROI) to become apparent.

AI: The ROI Runway Could Be Long Outside the Tech Sector | The Daily Spark

https://www.apollo.com/wealth/insights-news/insights/daily-spark/ai-the-roi-runway-could-be-long-outside-the-tech-sector

Slok explains that the valuation of AI companies' stocks incorporates the expectation that the profit margins of the S&P 493 (the S&P 500 excluding seven major technology companies) will increase in the future. However, as of June 2026, there are no signs of profit margins rising outside of the technology sector. While expectations for AI are driving up stock prices, the path to actually increasing profits for companies using AI remains unclear.

When considering the profitability of AI implementation, the 'token cost' incurred each time the AI is run is crucial. High token costs cannot be justified unless the profits of companies that implement AI actually improve, and if the amount of AI used does not increase unless token costs are reduced, the revenue of major cloud providers and others may not grow as much as expected. Slok states that the reason why discussions surrounding token costs, the appropriate use of models, and the token market are important is because they are deeply linked to the future revenue of AI companies.

In the technology sector, there are many cases where AI can be quickly integrated into existing products and internal processes. On the other hand, in many non-technology sectors such as healthcare, banking, insurance, energy, utilities, defense, aerospace, pharmaceuticals, life sciences, manufacturing, logistics, construction, real estate, education, legal affairs, and the public sector, regulatory compliance, data management, and changes to business design are necessary, which may cause AI-driven productivity improvements to lag behind market expectations.

Furthermore, before actually implementing AI, it takes time not only to integrate AI into internal processes but also to create rules that are based on AI. For example, even if AI can speed up document verification, it won't lead to overall time savings unless it's decided who will approve the verification results, who will be responsible if errors occur, and to what extent internal data can be shared with the AI.

Even if AI is eventually introduced into sectors other than technology, the stock market may have already factored in AI-driven profit growth far too early, and delays in its implementation could become a problem. Slok points out that if productivity increases over five years rather than in five months, stock prices that are based on the assumption of immediate profit growth may need to be re-evaluated.

Furthermore, if companies cannot quickly see the results of their AI investments, they may reduce their AI-related spending. Slok explains that the focus on optimizing AI to reduce processing costs is an early warning that AI adoption may be a more bumpy road than anticipated.

While AI is expected to be a tool to speed up operations, boosting profit margins requires that the operations, data, regulatory compliance, and cost management of companies implementing AI are all in sync. Slok says that the gap between the return on investment (ROI) that investors expect from AI companies and the time it takes for companies to actually realize an ROI from their AI investments can have a significant impact on the valuation of AI companies.