Over the past few years, the rapid advancement and adoption of “artificial intelligence” have driven a profound economic and technological boom. Organizations across the globe have eagerly integrated AI into their workflows, captivated by the promise of unprecedented efficiency and innovation. However, as the initial waves of excitement begin to settle, a more complex reality is emerging. The foundational costs of developing, running, and scaling AI models are increasing significantly, prompting a vital reassessment of how businesses deploy this technology.

The trajectory of the AI industry is currently following a familiar economic pattern often seen in the technology sector: the transition from a growth-focused “launch and scale” phase to a period of aggressive monetization. Early on, service providers frequently absorb heavy losses to build their user base and establish market dominance. Now, those same providers must demonstrate a return on their staggering capital expenditures.

To put the scale of these investments into perspective, the largest technology providers collectively spent roughly $155 billion on AI infrastructure in 2025 alone. This financial pressure is translating directly into an “AI tax” on enterprise software. According to a recent analysis in Forbes, relying on data from the procurement platform Tropic, software subscription costs rose between 20% and 37% by the end of 2025, largely driven by AI-related pricing adjustments. The era of heavily subsidized AI access is giving way to a market where providers are leveraging their pricing power.

The physical requirements of artificial intelligence – specifically, the data centers, energy consumption, and specialized hardware required for compute – are immense and increasingly constrained. This hardware bottleneck is directly impacting cloud computing costs.

For instance, Amazon Web Services (AWS) recently implemented its second price increase this year for its EC2 Capacity Blocks for Machine Learning service. Following an approximate 15% increase in January 2026, AWS announced a further 20% hike effective July 2026, as reported by VAR India. These increases reflect the mounting pressure of GPU hardware shortages and rising memory costs. Unlike consumer-level inflation, these infrastructure cost increases directly affect the foundational layer upon which countless enterprise applications are built, effectively raising the baseline cost of innovation across the entire ecosystem.

As AI usage matures, businesses are encountering unanticipated financial hurdles. Moving beyond pilot programs into full-scale production has revealed that the consumption-based pricing models of AI can quickly drain IT budgets.

A stark example of this occurred at Uber, which reportedly exhausted its entire 2026 annual budget for AI coding tools in just four months. With monthly API costs ranging from $500 to $2,000 per engineer, the financial burden scaled far more rapidly than anticipated.

This challenge is expected to intensify with the rise of “agentic AI” – systems designed to handle complex, multi-step tasks autonomously. According to a recent report covered by CFOtech UK, which cited Gartner analysis, agentic AI models can require 5 to 30 times more tokens per task than standard chatbot interactions. Consequently, organizations that based their cost projections on early, simpler generative AI pilots are finding that real-world deployment is vastly more expensive.

“Stop measuring GenAI success by how much your people are using it. Governance, budgets, and usage controls need to be in place before you scale. And critically, before any organisation moves from pilot to production, especially into agentic AI, they need to stress-test what costs might look like at full deployment.”

— Mark Brown, SAS

The escalating costs of AI are not just a microeconomic concern for individual organizations; they pose broader macroeconomic questions. The Bank for International Settlements (BIS) recently cautioned that the rapid surge in AI-related investments raises valid questions regarding long-term sustainability. In their assessment, highlighted by Cointelegraph, the BIS warned:

“Should inflation rise significantly or AI-led investment turn to a bust, the macroeconomic consequences could be amplified by existing financial vulnerabilities.”

They further cautioned that if central banks tighten policy, or if ROI does not materialize, it could precipitate a “sharp pullback in AI} asset prices after a prolonged period of exuberant risk-taking.”

This cautious outlook is mirrored in enterprise sentiment. Research commissioned by SAS found that among organizations in the UK and Ireland that have fully integrated generative AI, 45% reported a below-expected return on investment. As the cost of implementation rises, the margin for error shrinks. Business leaders are being forced to shift their perspective from broad adoption to rigorous, ROI-justified deployment.

The current AI boom is not necessarily ending, but it is undeniably maturing. The rising costs of AI—driven by massive infrastructure demands, hardware shortages, and the natural evolution toward profitability by service providers—are enforcing a necessary discipline on the market. For the boom to remain sustainable, organizations must move away from speculative adoption and toward highly targeted, efficient deployments. The future of AI in the enterprise will belong to those who can balance the transformative potential of the technology with the sober realities of its underlying economics.

Likely there is a massive market for reasonable AI consulting. :)

But as per usual – the choice is yours!