Unpacking the 18x Claim
The explosive 18-fold figure, highlighted by Gartner at its IT Symposium in Kochi, doesn't mean your next AI software subscription will be 18 times more expensive. Rather, it reflects the total investment required to generate measurable business value
from AI. As companies move from small experiments to enterprise-wide deployment, the associated costs of turning a clever model into a reliable business process are escalating dramatically. This includes everything from infrastructure and implementation to employee training and ongoing oversight. While over half of functional leaders report seeing some positive outcomes from AI, the price of that success is climbing steeply, forcing a reckoning between the initial hype and the reality of commercial returns.
The Hidden Iceberg of AI Costs
The initial cost of an AI model or API is just the tip of the iceberg. The real expenses are buried in the total cost of ownership (TCO). A primary driver is 'inference cost' — the computing power needed every time the AI performs a task. As AI gets more sophisticated, moving from simple chatbots to complex, multi-step 'agentic' workflows, these costs can increase fivefold or more. Then there are the people costs. Finding and retaining specialist talent like AI engineers and data scientists is a major expense. Beyond that, significant investment is needed for data preparation — cleaning and structuring your company's data so the AI can use it effectively. Other hidden costs include infrastructure upgrades, cloud services, energy consumption, and the essential but often-overlooked work of change management and employee training.
The India Equation
For Indian businesses, the AI cost equation has unique variables. While the country boasts a large pool of engineering talent at competitive rates, this advantage is offset by other challenges. The hardware needed for AI, particularly high-end GPUs, is often more expensive due to import dependencies and supply chain complexities. Furthermore, while India's data centre capacity is growing, the power-hungry nature of AI workloads puts pressure on energy infrastructure, where industrial tariffs can be high. Reports suggest that for many Indian firms, the savings from AI-driven productivity are being outweighed by rising spending on cloud subscriptions and infrastructure. This has created a scenario where AI is boosting efficiency but not necessarily profits, forcing a strategic rethink on how and where to deploy it.
Smarter Spending: Taming the AI Cost Beast
The rising costs don't mean businesses should abandon AI. Instead, they demand a more strategic and disciplined approach. Experts advise starting with a clear return on investment (ROI) in mind, focusing on projects that solve specific business problems rather than adopting AI for its own sake. One key strategy is 'inference tiering,' which involves using the most cost-effective model for each task. Not every query requires a powerful, expensive frontier model; simpler, cheaper models can often handle routine jobs perfectly well. Establishing strong governance from the start is also crucial. This includes setting budgets, tracking usage across departments, and implementing controls to prevent runaway spending on cloud resources and API calls. Ultimately, managing AI costs requires treating it as a core business function, not just a technical challenge.
















