AI spending pressure rises as firms burn through ‘full-year budgets’ in months

The global artificial intelligence boom is entering a new phase where rising costs are becoming a major concern for both corporations and users, as AI companies shift from heavy investor funding to revenue-driven models.

Earlier, AI firms spent billions on training and deploying advanced models, largely supported by investors and venture capital. However, with funding conditions tightening, companies are increasingly recovering costs directly from customers, placing pressure on businesses that rely heavily on AI tools.

Corporations feel the cost pressure

Large corporations are now among the biggest consumers of AI services, and many are reporting sharply rising operational expenses.

Uber’s Chief Technology Officer Praveen Neppalli Naga recently said the company exhausted its full-year AI budget within just a few months of 2026, driven largely by the increased use of AI coding tools such as Anthropic’s Claude Code.

Similarly, Amos Bar-Joseph, CEO of getswan.com, revealed in a LinkedIn post that his company’s AI bill reached around $113,000 (approximately ₹1.06 crore) in a single month for a four-person team, highlighting how quickly costs can scale even for smaller organisations.

Even within the AI ecosystem itself, concerns are being raised. Bryan Catanzaro, vice president of applied deep learning at Nvidia, noted that in many cases “the cost of compute is far beyond the cost of employees,” underscoring how infrastructure expenses are reshaping business economics.

According to Gartner, global IT spending is projected to reach $6.31 trillion in 2026, marking a 13.5% increase from 2025. This growth is being driven by AI deployment, cloud infrastructure, and subscription-based software models.

Companies are now under increasing pressure from investors and shareholders to demonstrate clear returns on AI investments, particularly as operating costs continue to rise.

Users also affected

The financial strain is not limited to corporations. Individual users of AI platforms have also reported faster exhaustion of usage limits, particularly under paid subscription plans.

Some users of Anthropic’s Claude have complained about reduced access due to token limitations, while advanced AI features are increasingly being placed behind paywalls.

A survey by Epoch AI also found that premium AI tools are more commonly used by higher-income users, while lower-income users tend to rely on less advanced alternatives such as Meta AI.

Despite rising costs, AI companies are actively working to improve efficiency. Google has introduced new TPUs designed to reduce energy consumption for AI training and deployment.

OpenAI has also launched GPT-5.5, which is designed to complete tasks using fewer tokens, improving efficiency while maintaining capability.

As competition intensifies, the industry is expected to focus on reducing the cost of AI usage through more efficient models and infrastructure improvements.