AI Token Costs Disrupt Business Strategies

Companies Facing Pressure from AI Token Costs

In the business world, AI tokens have become a major concern for leaders like Marty Kausas, CEO of BIPylon. Faced with a potential bill of $1.4 million, Kausas decided to limit token spending for certain non-technical employees. This decision is part of a strategy to prevent costs from soaring beyond forecasts. The company, which employs nearly 150 people, found that exceeding this threshold would triple its expenses. The vice president of finance at Pylon is now tasked with determining the caps to be implemented, marking the end of an era of unlimited spending.

This phenomenon is not isolated. More and more companies are realizing that the excessive use of AI tools, once encouraged, now needs to be controlled to avoid unnecessary expenses. Leaders like Sam Altman of OpenAI have noted a radical shift in the perception of AI budgets. While companies were previously satisfied with their spending, they now view these costs as a major issue.

A Transformation in Software Engineering

The series "The Great Coding Reset" explores how AI is profoundly transforming software engineering. Engineers, once focused on software development, now have to justify the use of the computing resources necessary for their work. This new dynamic forces managers to advocate for their teams' token needs, sometimes presenting them as high-potential projects, reminiscent of the show "Shark Tank." To attract AI talent, recruiters are now promising token allocations, highlighting the growing importance of these resources in recruitment.

The competition for computing resources is intensifying, involving all levels of the company, from executives to junior developers. This struggle for tokens could redefine power dynamics within teams.

The Effects of the Token Frenzy

Max Kan, a tokenomics analyst at SemiAnalysis, has long advocated for high token budgets, believing they could double employee productivity at a relatively low cost. However, the transition from a period of generous spending to budget tightening has raised concerns among engineers. Previously, companies encouraged intensive token use, creating pressure to consume more AI.

An analysis by AlphaSense revealed that the term "tokens" was mentioned in 129 earnings calls in the second quarter of 2026, compared to 57 in the previous quarter. This shift in discourse has led companies like Coinbase and Walmart to impose strict limits on token usage, while Amazon has shut down its internal token leaderboard.

Max Kan continues to advocate for generous allocations per engineer, but he worries about the impact on worker morale, who may feel caught between conflicting expectations from their employers.

Towards More Consistent AI Policies

Companies, from financial giants like JPMorgan to media conglomerates like Disney, are striving to develop consistent AI policies. Some, like Pega, have always been skeptical of the token frenzy. Ken Stillwell, CFO of Pega, criticized this trend as a "self-serving narrative" from AI companies. Pega has not set digital caps but has limited requests exceeding a certain amount.

As the discourse around tokens evolves, Stillwell feels vindicated in his cautious stance, emphasizing the importance of a measured approach to AI.

Continued Increase in AI Spending

According to Ramp's AI Index, AI spending averaged $66.29 per employee in May, up from $58.84 in April. Ara Kharazian, chief economist, anticipates a continued increase in these expenditures, although signs of tightening are emerging, particularly through the increased use of model routers to optimize costs.

Some companies, like the startup MindFort, are reconsidering their workforce in terms of a tokens-to-people ratio. Brandon Veiseh, its CEO, stated that the company would have needed 20 employees before AI to reach its current scale. Tokens have helped bridge this gap, but Veiseh insists on the need to carefully weigh the return on investment.

While token costs are expected to decrease with increased competition among AI companies, compromises on resource usage will remain inevitable. Companies must therefore navigate carefully between innovation and cost control.

Fierce Competition for Tokens

As engineers realize they may have to fight for their tokens, tensions within teams could escalate. Some compare this situation to a survival competition, evoking "The Hunger Games." During job interviews, developers are increasingly asking about token allocations, a sign that these resources are becoming a crucial criterion for candidates.

Max Christoff, CTO of Everlaw, has proposed establishing token caps while allowing engineers to negotiate larger budgets. He compares this to using cellular data before unlimited plans, highlighting the importance of balanced resource management.

Russ Franklin, founder of Larridin, predicts that companies will impose restrictions on access to AI models, comparing it to booking economy flights versus renting private jets. Access to advanced AI models could become a privilege reserved for a select few.

Engineers have good reasons to fight for their tokens, as limited access to AI could harm their long-term careers. Brock Simon, a former AI consultant at Bain & Company, observed that some companies are delaying the adoption of technology, which could disadvantage their employees in the job market.

The Consequences of the Token Frenzy

Max Kan holds the official title of tokenomics analyst. Within the data provider SemiAnalysis, Kan helps build token models for hedge funds and hyperscalers. When I called him in May, he was optimistic about the impact that high token budgets could have on the workforce. "It's essentially true for everyone that if you have an employee making $100,000 a year, you can probably make them twice as productive with $10,000 in tokens," he said.

Kan worries about what engineers might think as they transition from a token frenzy to budget tightening. Janice Chung for BI reported that it was the days of the token frenzy when companies sent their engineers diving into pools of tokens like Scrooge McDuck. Companies encouraged token rankings, where those at the bottom felt pressure to use more AI, and leaders across various sectors kept talking about it.

The word "tokens" was used in 129 earnings calls in the second quarter of 2026, compared to 57 calls in the previous quarter, according to an analysis conducted for Business Insider by the business intelligence platform AlphaSense.

Within weeks, the tightening began. Companies started cutting their AI budgets and setting token limits. Coinbase set a cap; Walmart did too. Amazon shut down its internal token leaderboard.

Kan still advocates for large allocations per engineer — and wonders what workers will think of this rapid shift in discourse. He worries that engineers might think, "My boss is pushing me hard to do one thing, then I did that thing, and now I'm being yelled at because I did that thing well."

Leaders Developing Consistent AI Policies

Leaders from various sectors — from financial giants like JPMorgan to media conglomerates like Disney — are working to develop consistent and effective AI policies. Some companies have always been against the token frenzy. The enterprise software company Pega is one of them. When I spoke on the phone in May with its CFO and COO, Ken Stillwell described this trend as an "incredibly self-serving narrative" from AI companies. His company has not set digital caps for tokens but has limited requests that exceed a certain amount.

When we spoke a month later, as the discourse had evolved, Stillwell felt vindicated. "We're quite happy to be one of many talking about this," he said.

AI Spending Continues to Rise

Tech and media companies spent an average of $66.29 per employee on AI in May, up from $58.84 in April, according to Ramp's AI Index. Ara Kharazian, its chief economist, told Business Insider that he expects this figure to continue to rise, but he noted early signs of tightening, such as the increased use of model routers, which can help better manage costs.

Some companies are not yet cutting their AI budgets, but they are critically reflecting on their workforce. For example, MindFort, an AI startup backed by Y Combinator, has six employees. Its CEO, Brandon Veiseh, stated that the company would have needed 20 employees before AI to reach its current scale. Where did those funds go? Tokens.

Brandon Veiseh is focused on the return on investment of AI spending in his company. Morgan Lieberman for BI reported that he stated, "We need to weigh our tokens-to-people ratio." "It's not something we consider particularly comfortable or pleasant to say."

While token costs are expected to decrease as AI companies like Google increasingly compete on price by offering smaller, more efficient models, these types of compromises are unlikely to disappear. Often, the cheaper a resource is, the more it is consumed.

For now, companies are thinking more carefully and sometimes taking strategic steps back — but they hesitate to act too quickly. Kausas, the CEO of Pylon, stated that he wants to prioritize ensuring a return on investment — and avoid engineer dissatisfaction. "If we told engineers they weren't allowed to use AI products, they wouldn't work here," he said. "It would feel like being in the Stone Age."

The Dawn of the Token Hunger Games?

As engineers increasingly learn that they may have to fight for their token allocation, conflicts within teams could grow. Some have compared this to a survival-of-the-fittest scenario. "Coding is now cockroach protein bars and we're all fighting for crumbs," said one coder on X, likening the dynamic to "The Hunger Games."

Developers are also asking more questions about tokens during job interviews. Kausas noted that candidates have asked him about budgets. AI expert and former AWS employee, Allie K. Miller, has heard of interviewees inquiring about details: "What level of model will I have access to? Do you have partnerships with AI labs that give us relatively early access?"

This is a sign of a new era where tokens — or at least the number that workers desire — are not guaranteed. Max Christoff, the CTO of legal tech company Everlaw, has advocated for giving engineers token caps while allowing them to negotiate larger budgets. He compared this to using cellular data before unlimited plans. Sometimes you have to spend a lot for data, but other times, you scroll mindlessly, not realizing how much you're wasting. Christoff wanted all of the former and none of the latter.

"We want to facilitate the demand for more if you can actually use it," said Christoff.

If a company does not set token caps, it may also impose restrictions on models. Russ Franklin, the founder of Larridin, a platform for tracking AI usage, was emphatic. "Of course, they will limit who can use these tools. It's not even a question," he said. Franklin compared the allocation of access to models to a company-paid trip. Many can book an economy flight, but few — if any — can rent a private jet, he said. Access to cutting-edge AI models may be akin to a private jet: so costly that only a few stars can afford it.

Engineers have good reasons to fight for their tokens. Limited access to AI could harm their long-term careers, making them less qualified or less attractive in future job searches. Brock Simon advised companies on AI for Bain & Company before founding his own startup, Native. He observed that some companies were slow to adopt technology or restricted access to specific tools and agents, leaving their employees behind.

"It has really harmed the careers of some people," he said.