Tech companies that heavily relied on internal AI usage are now taking steps to control costs due to the rising expenses associated with intensive AI utilization. Uber, for instance, revealed that it had exhausted its entire AI budget for 2026 within the first four months of the year, leading the company’s COO to express concerns about the escalating internal AI expenditures. Similarly, OpenAI CEO Sam Altman highlighted that AI costs have emerged as a significant issue for their clients.
This trend is not limited to industry giants, as leaders from Canadian startups also indicated grappling with the escalating internal AI costs. The focus now shifts towards cost tracking and strategic AI utilization, raising questions about the impact on the high valuations of AI companies if tech firms start curbing their spending.
The surge in expenses is primarily attributed to the use of “tokens,” which are data units necessary for inputting prompts and receiving outputs from AI systems. The substantial volume of tokens being utilized, driven by the practice known as “tokenmaxxing,” directly correlates with the expenses incurred by users interacting with AI technology.
While the cost of real-world AI applications, termed inferences, has generally decreased, tech companies are increasingly leveraging AI for intricate tasks such as coding and complex reasoning processes. This shift differs significantly from basic AI interactions like seeking recipe suggestions from ChatGPT, as noted by cognitive scientist and AI researcher Gary Marcus.
Previously, many tech companies encouraged extensive AI experimentation among employees, with a focus on maximizing token usage. However, faced with the financial implications of high token consumption, some businesses are reevaluating their expenditure strategies. For instance, Uber recently implemented a monthly cap of $1,500 per employee per coding tool to manage costs more effectively.
To navigate the balance between innovation, cost control, and tangible benefits, companies are now exploring AI “tokenomics” – a strategic approach involving a deeper understanding of token costs and the strategic deployment of AI resources. Businesses are advised to conduct micro-sized experiments to identify AI’s utility as a tool, focusing on speed and cost-effectiveness compared to human capabilities across various organizational functions.
As the AI sector undergoes a period of reflection, companies are assessing the ROI of complex AI applications amid mounting costs. The industry faces a dilemma of balancing the need to recover token expenses with maintaining market share in a fiercely competitive landscape. Moves such as revising pricing structures, as seen with companies like Anthropic and GitHub Co-Pilot, reflect the evolving dynamics of AI economics and the ongoing quest for a sustainable cost model that aligns with business objectives.