For all their talk about AI, few companies are quantifying gains

The artificial intelligence industry is grappling with a significant issue - many companies are failing to demonstrate quantifiable gains from implementing AI tools. While anecdotal evidence exists, there is a notable lack of concrete data showing AI-driven profitability or cost savings. Barclays strategist Venu Krishna highlighted this gap in a recent note, stating that only 12% of companies willing to discuss AI on earnings calls could provide quantitative evidence of AI-related value capture.

Tech, financial, and healthcare companies accounted for 68% of these instances. The average reported efficiency gain stands at 56%, yet this figure has remained largely unchanged over the past three years of large language model advancements, raising doubts about its accuracy. Meta's AI tools reportedly boosted Facebook ad clicks by 8.3% and conversions by 15.7%, while online pet care retailer Chewy aims for $50 million in cost savings by 2027 through AI deployment.

Hilton expects AI to expand margins by 75-100 basis points for hotel owners. Despite these examples, the percentage of S&P 500 firms quantifying AI productivity gains remains significantly lower than those simply claiming AI presence. This disparity underscores the gap between companies' desire to appear AI-enabled and their actual ability to leverage AI effectively.

As hyperscalers like Meta, Microsoft, and Amazon invest billions in AI infrastructure, the need for tangible financial returns from AI adoption becomes increasingly pressing, particularly as capital expenditures on data center infrastructure are projected to surge from $800 billion annually in 2026 to $1.8 trillion by 2050. PwC's global infrastructure leader, Clara Cutajar, emphasized that AI infrastructure investment is set to accelerate, impacting various sectors including technology, energy, real estate, and supply chains.