Is AI Really Worth It? Here’s How Businesses Can Find Out

Artificial intelligence has become the headlining act in boardrooms. From fraud detection to personalised marketing and assistant tools for staff, it’s changing how companies compete. Yet a paradox runs through this progress: the smarter our systems become, the more complex it is to see what they cost – and whether they’re worth it.

Boston Consulting Group’s latest survey of 1,000 executives found three quarters of the firms that responded still struggle to turn pilots into value at scale. The problem isn’t a lack of ideas – it’s a lack of clarity regarding outcomes.

AI spending is rarely one neat line in a budget. Model training devours computing power and requires specialised infrastructure like GPUs; running AI in production for customers brings unpredictable usage fees; data must be stored, cleaned and secured; licences multiply; and teams of engineers are needed to keep everything working.

Those costs are scattered across invoices and departments, often hard to spot inside the wider IT budget, or unnoticed in the budgets of the business units. So while boards celebrate AI’s promise, few can say with confidence what it really costs – or whether it’s paying off.

“In most board reports, AI spending and value is hidden and leaders can’t tell what is actually from AI,” says Greg Holmes, field CTO for EMEA at Apptio, an IBM company. “What’s needed is a single, shared view of AI spending that finance, technology and business teams can all understand.”

It sets out the basics so leaders have one version of the truth they can trust

The cost of not knowing

Without that clarity, familiar mistakes creep in. A retailer launches AI-driven recommendations but never separates the cost of running them from its wider cloud bill, so no one knows if the system makes or loses money. A bank bundles all inference costs into a single innovation budget, leaving it unclear how much went to experiments and how much to customer service. In both cases, good intentions are undone by muddled bookkeeping. Utilisation of AI resources is segmented and not clear, so it is difficult to make strategic investments.

Without a shared view of the numbers, people can talk at cross purposes. Technology chiefs may argue for investment while finance teams question rising cloud costs and product managers claim successes that are not obvious from the accounts. Decisions can become reactive, forecasts can slip, budgets can unravel and value can disappear into the inference ether. Promising projects may stall for lack of resources while others that should be stopped limp on.

A clearer way to govern AI

The answer is simple: treat AI like any other major investment and give it a common language of cost and value. That means bringing together the costs of cloud services, software licences, data pipelines, storage and the people who build and maintain them, then viewing those costs through what the business actually uses, then correlating this with operational data around how the technology is used.

Additionally, by bringing in business data to demonstrate the strategic business value being created, we can see how the spend is actually realising value. When you can see what an AI system costs per customer or per claim, you can compare projects on equal terms and back the ones that deliver.

This is the thinking behind Apptio’s Technology Business Management, a straightforward way of linking what companies spend on technology to what they get out of it. In practice, it facilitates keeping control of cloud technology costs so the computing power behind AI doesn’t come as a surprise; tracking the full cost of technology so each business unit knows what it consumes; and tying budgets to results so funding can be quickly directed to the projects that prove their worth.

“Think of it as a layer of financial intelligence for the business,” says Holmes. “It combines financial, operational and organisational data to help clearly surface the basics: what’s being used, what it costs and what it delivers, so leaders have one version of the truth they can trust.”

The aim isn’t to stifle innovation. Quite the opposite: when companies see where money goes and what value it brings, they can grow what works and stop what doesn’t.

At MassMutual, the US insurer, cost-transparency practices helped leaders understand the total cost of ownership for applications and services during a major divestiture. That insight enabled $60 million in direct expenses to be eliminated on day one of the deal closing, and gave teams the clarity to rationalise systems and focus on what mattered most.

BMO, one of North America’s largest banks, took a similar approach to accelerate its digital first strategy. By standardising the way technology costs were categorised and improving cost attribution by 70 per cent, leaders gained a clearer view of the actual costs of each service and application. That insight helped reduce waste and fund new initiatives, creating the governance and data foundation needed to adopt emerging technologies like AI with confidence.

Different sectors, same story: once leaders see the business value of their investments in innovation, they can steer it with confidence.

None of this calls for complex maths – just discipline and a few good habits. Drawing a clear line between the cost of training models and running them day to day can help prevent month-end shocks. Agreeing what success looks like before money is spent, linking each AI project to a clear goal – faster fulfilment, lower customer churn or higher sales – and knowing who is accountable all help keep ambition grounded. Understanding the cost of a single transaction or prediction lets enthusiasm be weighed against value, while treating funding as a living process helps ensure money flows to projects showing real progress, not those kept alive by optimism.

The winners will be those who can explain, predict and align spending on AI

Behind this lies a simple truth. The most successful firms invest less in algorithms and more in the people and processes around them. Good management, clear ownership and shared visibility do far more for performance than another model tweak.

“AI spending will keep growing,” says Holmes. “The winners will be those who can explain it, predict it and align it – in the language the board understands. If they can’t do that, they’ll scale risk, not value.”

From pilots to profit

It is clear for now, most company AI projects don’t make it out of the pilot stage and into production. They can build prototypes quickly but struggle to prove lasting value. The answer isn’t another lab or proof of concept; it’s a clear view that lets the business decide where to place its bets.

With that clarity, technology leaders can stop debating invoices and start talking about results. Finance can move from being the department of “no” to the arbiter of value, and boards can judge AI projects in business terms rather than technical promise.

Complexity may be inevitable, but chaos is not. In a year of tight budgets and high expectations, cost transparency is what turns AI from a fashionable strategy line into a genuine source of competitive strength.