'We Created a Monster': Leading Firms Curb AI Spending Amid Surging Costs

As the initial euphoria surrounding generative AI transitions into a more sober operational phase, some of the world's largest companies are beginning to rein in their usage of the technology. Giants including Amazon, Walmart, Meta, and Uber have reportedly introduced caps or discouraged "wasteful" AI experiments as the costs associated with running large language models strain corporate budgets. After a year of rapid adoption, executives are now demanding clearer evidence of return on investment (ROI) before approving further large-scale deployments.

The "monster" of AI costs stems from the high fees charged by cloud providers for GPU compute time and the premium prices for top-tier model licenses. In response, many firms are pushing internal teams to use smaller, cheaper open-source models for routine tasks rather than the most advanced versions of GPT or Claude. This cost-conscious shift is a reaction to the realization that while AI can improve productivity, the current pricing structure remains too high for universal application across every business unit.

At the same time, the social and economic implications of AI-driven automation are coming into sharper focus. Policymakers in China are warning that robots and autonomous systems could replace up to 700,000 delivery workers in the near future, threatening the livelihoods of millions in the gig economy. To prevent a wider backlash, some analysts and business leaders are advocating for a "redeployment tax credit" to encourage companies to reinvest AI-related profits into worker retraining and apprenticeships. The debate is shifting from what AI can do to who will pay for its implementation and its societal consequences.