AI Spending Soars, But Are Businesses Seeing Real Value?

AI investment is on the rise, but according to KPMG, the gap between spending and tangible value is widening. The numbers are eye-opening. Organizations worldwide plan to shell out an average of $186 million on AI over the next year. Yet, only 11% have reached a stage where AI deployment is driving broad business outcomes.

Big Spending, Little Return?

64% of respondents claim AI is delivering meaningful results. But here's the kicker: "meaningful" often just means incremental gains rather than transformative efficiency. The pitch deck says one thing. The product says another.

KPMG distinguishes "AI leaders", those effectively scaling AI, from everyone else. Among these leaders, 82% see real business value from AI, while their peers lag at 62%. That's a significant gap. It's not just about better tools, it's about smart deployment.

The Real Cost of AI

So, what does $186 million buy? The bulk goes to model licensing, infrastructure, and risk management. But here's the rub: many organizations overlook the operational infrastructure needed to truly benefit from AI. The friction costs, engineering hours, system integration headaches, regulatory compliance, often catch companies off guard and exceed initial budgets.

Take vector database integration as an example. It's important for retrieving data in real time, yet maintaining this infrastructure adds complexity and ongoing costs. When neglected, AI performance dips, leaving companies scratching their heads.

Trust and Governance

One of KPMG's key insights is the link between AI maturity and risk confidence. Where 49% of AI leaders express confidence in managing AI risks, only 20% of those still experimenting do. Governance isn't just a compliance checkbox. it's what enables faster, more confident deployment.

Mature organizations build governance into their deployment pipelines, embedding trust and allowing them to scale. As Steve Chase from KPMG puts it, "there's no agentic future without trust and no trust without governance." He's got a point. If governance isn't part of your launching strategy, you're just setting yourself up for delays and risks.

Ultimately, the question for laggards isn't whether to ramp up AI deployment, it's how to do it without digging a deeper hole of integration debt and unmet governance needs.