cc.web.local

Rising AI costs push enterprises towards multi-model strategies

Over the past 12 months, 59 per cent of executives reported that their organisation’s AI-related expenditure surpassed expectations.

The global rush to integrate artificial intelligence into enterprise workflows has hit a stark financial reality.

According to a recent survey, nearly six in ten organisations are facing significantly higher AI costs than initially expected. This unbudgeted expenditure has pushed AI costs to the forefront of executive concerns, prompting a rapid pivot towards multi-model strategies and stricter financial oversight.

A snapshot report published by the Capgemini Research Institute surveyed 500 CXO-level executives, revealing a widespread struggle to balance technology consumption with verifiable business returns. With 53 per cent of business leaders now actively concerned about deployment costs, the era of unquestioned investment is giving way to a sharper focus on economic outcomes.

The hidden costs of enterprise AI

The financial toll of scaling artificial intelligence is becoming increasingly clear. Over the past 12 months, 59 per cent of executives reported that their organisation’s AI-related expenditure surpassed expectations. The scale of this overspending is substantial:

  • Affected organisations overshot their budgets by an average of 24 per cent.

  • Roughly one in five organisations exceeded their initial budgets by more than 30 per cent.

A fundamental lack of visibility is driving these budget overruns. Half of the executives surveyed admitted that spending is frequently approved without adequate insight into the total financial commitment. While direct token costs are easily tracked, the broader picture remains obscured. Only a third of leaders can confidently estimate indirect costs, which include extensive human oversight, system rework, and ongoing governance.

For many, this lack of foresight has led to tangible financial distress. Two-thirds of the organisations that exceeded their budgets faced negative consequences, including reduced discretionary spending and lower free cash flow. Consequently, Chief Financial Officers are now demanding greater involvement in approving major investments.

Pivoting to a multi-model strategy

In response to spiralling expenditures, 53 per cent of organisations have introduced stricter expenditure controls, ranging from tightening specific project budgets to implementing enterprise-wide spending limits. However, only a fifth of businesses have fundamentally re-evaluated their tools, processes, or model routing.

Looking ahead, a more structural shift is imminent. Over the next 12 months, more than three in five executives expect their organisation to adopt a multi-model strategy, dynamically selecting models based on cost, capability, risk, and latency. Portfolio optimisation plans include:

  • Leveraging lower-cost proprietary models for standard operations.

  • Deploying small or highly specialised models tailored for specific tasks.

  • Expanding the use of open-weight frameworks, a strategy cited by nearly three in five businesses.

The emergence of AI Yield

Despite high confidence in verifying value, most organisations still struggle to directly match their spending with successful business outcomes. Currently, only one in five calculates "AI Yield"—a critical metric capturing the verified value of an outcome after accounting for the full safe-delivery cost.

When businesses do successfully measure this value, they primarily point to enhanced employee productivity, released capacity, and operating-cost reductions. Conversely, less than a third attribute revenue growth directly to their investments. Demonstrating enough value to secure ongoing funding remains the largest hurdle for 57 per cent of organisations.

This challenge is driving a shift in economic metrics. While 60 per cent of businesses currently measure total expenditure against their budget, only 44 per cent plan to prioritise this metric moving forward. Instead, operational efficiency metrics—such as the first-pass success rate and the cost per completed workflow—are set to become the standard.

Managing portfolios for economic outcomes

To successfully scale, businesses must change how they view these investments. Volker Darius, CEO of Capgemini Invent, highlighted the necessity of measuring quality and business value in tandem with direct expenditure.

"As AI moves deeper into business workflows, organizations need to stop treating its cost as a technology or procurement issue and start managing AI as a portfolio of economic outcomes," said Darius. "Cost discipline alone will not create a profitable AI estate, just as value creation without cost control will not be sustainable. Organizations need to measure cost, quality and business value together, use the most appropriate model for each task, and account for the human oversight required to deliver outcomes safely. By continuously measuring AI Yield, organizations can make more informed capital-allocation decisions and scale AI with confidence."