How Much Is AI Really Costing Your Company in 2026?

In Brief:

• AI spending is outpacing hiring savings. Companies delaying new hires in favor of AI tools are finding that token costs are climbing faster than the salaries they’re avoiding, eroding the expected cost benefit.
• The shift to agentic workflows is the real cost driver. A basic AI query cost about 4 cents in 2023; a multi-step agentic workflow now runs around $1.20, roughly 30 times more, even as per-token prices have fallen.
• Usage volume, not price, is the hidden risk. Per-developer AI consumption has increased by a factor of 18.6 in nine months, and average business spend on AI tokens is now 13 times higher than January 2025 levels.
• Rising AI costs directly hit profit margins and valuation. Since AI spend lands on the expense line, unchecked growth can compress margins, a metric CFOs actively optimize and investors watch closely.
• Companies need to act before 2027 budget planning. A year-to-date AI spend audit, forecasted costs, per-developer token limits, and an ROI/staffing cost comparison are recommended steps to get ahead of the fall budgeting cycle.

Depending on who is doing the talking, artificial intelligence (AI) is a revolutionary technology that will foundationally change the way business is done in nearly every industry with unprecedented speed and agentic decision-making — or it is an existential threat to the American worker whose skills will soon be irrelevant in an AI-driven marketplace.

But there is another impact that AI is already having in companies that have come to rely on it for engineering and building their software systems, one that will shake these companies’ CFO and finance departments to the core as soon as this fall, when 2027 budgets are being hashed out.

The Cost of AI

Companies, particularly startups in the technology arena, are utilizing on-demand cloud-based AI tools for increasingly more functions such as engineering, operations, finance, and human resources. AI can help people evaluate options much faster and product teams are using AI to validate hypotheses, challenge concepts and synthesize large amounts of data.

For young companies especially, the lure of using AI tools to accomplish functions that they would normally have to hire staff to do is appealing. No staff = no expense.

But it’s not that simple. They are indeed saving on workforce expenses by delaying the hiring of new employees. But at the same time, they are spending more on AI tools. And the cost of using AI tools is going through the roof.

For many companies, spending on AI is accelerating so fast, it’s starting to outpace the cost of hiring new employees. AI is still a young technology, and it is maturing at breakneck speed, so its promises are great. But many companies are barely out of the AI starting gate and, consequently, have no ROI track record to prove the value of the AI tools they are using. As a result, the increasing costs of AI are outweighing the benefits.

The Downside for Young Companies

For startups and many young companies, the downside of pushing a lot of dollars into AI has to do with the geography of where those dollars land on the financial statement. Accounting and financial results are still important, even in the AI era. Rising costs for AI mean rising expense lines in the financial statement. That translates into lower profit margins and, if the trend gets out of hand, lower valuation for the company.

The profit margin is a metric that drives business and drives valuation. It’s a metric for which CFOs are always trying to optimize.

Most early-stage companies across industries are heavily using AI, in part to forestall hiring people. Early-stage companies view the AI model as more efficient than staffing up and then risking the need for layoffs if they hit a slowdown in the marketplace.

The primary sources for AI tools are recognizable to nearly everyone — Claude, Anthropic, Open AI and Grok, as well as a few others.

Why Are AI Costs Rising?

Excuse us while we nerd out here and explain how AI costs are rising.

To integrate AI tools that reside in the cloud with local software, developers use a Large Language Model Application Programming Interface (LLM API), which acts as a bridge between your software and a language model (AI). It allows applications to send text inputs, along with optional settings, to a model, which then generates outputs.

The unit of cost for accessing an AI interface is called a “token.” Each token represents approximately four characters. So if the output that resulted from a query contained 20,000 characters, the cost would be approximately 5,000 tokens.

While per-token prices have fallen dramatically in the past four years, the volume of tokens used by developers has exploded. GPT-4-level output now costs about 40 cents per million tokens. Sounds cheap, until you consider that per-developer AI consumption rose by a factor of 18.6 in a recent nine-month period, and the average business now spends 13 times more on AI tokens than in January 2025.

The reason for the explosion in token usage is the shift from simple chatbots to agentic systems. In 2023, a basic AI query cost about four cents. An agentic workflow query in 2026 costs approximately $1.20, around 30 times more. Again, these numbers sound cheap. But when thousands of developers and analysts are running multi-step agents that loop through dozens of actions per task, volume explodes faster than prices drop.

Many large companies are capping their developers’ use of AI tools to manage costs, but still will likely have to reallocate more money to AI when the 2027 budget planning season hits.

How Can Startups and Middle-Market Companies Control Their AI Spend?

With AI costs rising so precipitously, there is no time to waste in getting it under control in your company. Some key measures include:

• Perform a year-to-date audit of your AI spending — now. Be sure to compare your actual spend to your budgeted amount.
• Create a forecast of what you expect to spend on AI for the rest of the year.
• If necessary, impose token-level limits on what each developer can spend, and enforce them.
• Analyze your ROI on AI spending. What are you actually getting for your investment in AI? What are the positive, actionable results?
• If you are using AI to forestall hiring staff, create an analysis comparing the real costs of both options.

We’re Here to Help

AI costs are rising precipitously and the impact on a company’s bottom line — and hence, valuation — could be dire if not reined in now.

If you would like help analyzing your AI spend and putting in place controls that can help your company stay on solid financial footing, contact your JLK Rosenberger team member, call 949-860-9902, or click here to contact us. We look forward to speaking with you soon.