C-suites have decided: it's time to put AI on a diet

In a significant shift from the earlier exuberance surrounding artificial intelligence, companies are now tightening their budgets as costs for AI tools continue to climb. After a period of unrestricted use, many organizations are reevaluating their AI spending habits, leading to a more cautious approach in the summer of 2026. Take Coinbase, for example. Rob Witoff, the executive overseeing infrastructure at the crypto exchange, observed a dramatic increase in AI tool usage following the release of Anthropic's advanced coding model, Opus 4.6. However, as prices have spiked, Coinbase has implemented weekly spending caps tailored to employees' roles, ranging from $500 to $5,000. Witoff explained, "Once people understand what's possible, usage takes off on its own. The focus has now shifted to ensuring that AI is being used effectively." This newfound restraint is not unique to Coinbase. Executives across multiple sectors are recognizing the need to be more strategic with their AI investments. Major players in the AI field, including OpenAI and Anthropic, are facing pressure as companies reconsider their reliance on expensive models, especially as they prepare for potential IPOs. Salesforce's CTO, Parker Harris, admitted that while the company has allowed unrestricted spending on Anthropic tools, this approach isn’t sustainable in the long run. "We can't tell our investors that we've given away our profits to help another company go public," he said. The shift in corporate attitudes comes as AI providers have adjusted their pricing structures. Between February and June, companies like OpenAI, Anthropic, and GitHub began charging customers based on token usage—an indicator of AI workload—rather than a flat fee. This change has prompted organizations such as Walmart and Amazon to impose internal usage restrictions and rethink their AI strategies. Uber, for instance, has already exhausted its AI budget for the year, revealing that excessive spending on tokens hasn’t led to proportional benefits. Niranjan Krishnan, head of AI solutions at FPT Americas, remarked, "The novelty of AI has worn off, giving way to a focus on practical utility." A survey conducted by Wakefield Research found that 79% of executives expressed concern about potential cuts to their AI budgets, as spending often does not correlate with new revenue streams. Developers are also feeling the pressure; changes in GitHub's pricing model have led to unexpected costs, with some tasks potentially exceeding $100 under the new usage-based system. Coinbase's Witoff highlighted the significant costs associated with previously routine tasks, such as comprehensive code analysis, which could now result in bills ranging from $50,000 to $100,000 per run. This has led companies to implement mechanisms that encourage more mindful usage of AI resources. LogicMonitor and other firms are also placing limits on token usage, with a focus on value-driven outcomes. Karthik Sj, head of AI at LogicMonitor, indicated that firms are now more judicious in selecting AI tools, opting for cost-effective solutions unless a substantial improvement is evident. OpenAI's CEO, Sam Altman, acknowledged the sudden shift in budget discussions surrounding AI. As the demand for advanced AI models continues to rise, companies are feeling the financial strain. Innovations in chip technology may have reduced some operational costs, but the surge in AI adoption means that providers can no longer afford to subsidize heavy users. The recent pricing changes have not gone unnoticed by developers, many of whom have voiced their concerns about escalating expenses. GitHub's chief product officer admitted that the previous billing model was unsustainable and that the company needed to adapt. In response to these challenges, businesses are exploring alternatives. Some firms have started utilizing less advanced AI models to handle basic tasks, while startups offering affordable AI solutions have seen a surge in demand. Ahmad Awais, founder of Command Code, reported a significant increase in customers, driven by the search for budget-friendly options. As the AI landscape continues to evolve, companies are grappling with finding the right balance between innovation and fiscal responsibility. With ongoing changes in pricing structures and usage patterns, organizations are forging ahead with a more cautious, strategic approach to AI adoption.