Annual AI Budgets Exhausted in Just Two Months: Here's Why

Jakarta, CNBC Indonesia - AI researcher and OpenAI co-founder Ilya Sutskever once remarked, “the future will remain bright for AI, but it will be more beneficial if that future is also bright for humanity.” These words are increasingly relevant today, as technology giants race to create the most advanced AI systems to win over investors. Ironically, AI is bringing new harms to human life across various sectors, from environmental pollution due to the proliferation of AI data centre infrastructure, to water and electricity crises caused by the high consumption required to run these facilities. Furthermore, a global memory chip shortage driven by AI data centres has already pushed up consumer electronics prices, not to mention the rampant spread of disinformation and online fraud facilitated by AI tools. Waves of layoffs and anxiety among the younger generation about human jobs potentially being replaced by AI have also become a spectre. In May 2026, the technology sector in the United States recorded the highest number of job cuts in nearly two years. According to data from outplacement firm Challenger, Gray & Christmas, tech companies announced plans to eliminate 38,242 positions during May, the largest figure since August 2024. Many tech giants have announced layoffs, using efficiency gains from AI adoption as a scapegoat. A recent CNBC International report suggests the widespread layoff phenomenon is also likely driven by companies’ ambitions to allocate spending wildly towards AI development, forcing them to sacrifice employees. The rising cost of using AI systems means corporate leadership must question exactly where they are spending the money needed to keep the business running. CNBC International spoke with two AI company CEOs who are busy implementing AI for their employees, and both highlighted that AI costs are rising sharply, with AI prices surging to their highest levels. The core of the problem is the price of AI tokens, the basic input units processed by large language model (LLM) systems, the core technology at the heart of systems like ChatGPT. Initially, many AI creators allowed users to access their models for free, with limits on the number of interactions. However, in an effort to monetise these tools, AI creators are charging increasingly higher fees for access to AI tokens from their advanced models. Annual Budgets Exhausted in Mere Months Concurrently, aggressive AI users seeking the productivity gains promised by the technology are consuming more and more tokens to try to solve their business problems. Some users are even engaging in ‘tokenmaxxing’, a meme-like trend where frontline workers frantically spend as many AI tokens as possible to appear productive. In a recent interview, OpenAI CEO Sam Altman acknowledged that the cost issue is becoming increasingly critical. “People are saying, ‘my company spent my entire 2026 budget in Q1. Can you make this more efficient?’” he said. The high cost of adopting AI has become a problem that suddenly emerged in 2026, as worker dependence on AI grew significantly. From the start of 2026, it transformed from “a problem that never existed into a major one,” Tom’s Hardware reported. Glean CEO Arvind Jain echoed similar sentiments. Based on testimonials from companies using Glean, “their AI budgets are exhausted in one or two months, and these are annual budgets,” he stated. AI Gets Smarter, and More Expensive The problem, according to Jain, is that even as AI usage increases, “AI costs have not fallen as buyers expected.” Instead, with each release of a new, more advanced AI model, the cost is roughly twice as expensive per token compared to the previous model. Jain noted that this is an unsustainable trend in the long term, which is why AI adoption is increasingly draining corporate funds. Should companies spend more on AI, or more on their AI users, namely the traditional workforce? Business Insider reported that the founder and CEO of AI recruitment startup Mercor discussed this issue in a recent podcast. Brendan Foody admitted that “we currently spend more on tokens for our internal agents than on the number of employees.” He actually sees no end to this trend, noting he is betting that “in five years, the average company will spend more on computation than on employees.” Jain is experiencing a similar problem. He explained that this is the first time he can recall “that technology costs as much as labour, and ultimately companies have to make a decision: choose technology or labour.” His concern is that the future expansion of AI budgets could impact spending on workforce growth. All of this may push companies to re-evaluate how they implement AI, including how much they need to spend on it.