Companies are deploying artificial intelligence at record rates, but the technology is not yet translating into measurable financial gains for most organizations. While 44% of companies have reached the AI scaling phase in 2026, up from 38% the previous year, nearly the same percentage of respondents as last year, 37%, reported that AI positively contributed to their organization's earnings before interest and taxes (EBIT), according to McKinsey's recent State of AI in 2026 report.

The disconnect reveals a critical challenge facing enterprise leaders: adoption and scaling are not the same as execution and profitability. A new class of AI tools called "judgment models" may offer a path forward, but first, companies need to understand why their current AI investments are failing to move the needle on the bottom line.

Why Is Enterprise AI Adoption Outpacing Financial Results?

The numbers paint a puzzling picture. McKinsey's survey of more than 1,700 people globally found that 88% of organizations are now using AI in at least one business function, up from 78% in 2024. In customer service specifically, 98% of enterprise contact centers now use AI, according to USAN's 2026 research. Yet only 12% of those organizations have a fully optimized AI strategy, exposing a wide gap between adoption and real execution.

The problem is not that companies lack AI tools. It is that they lack clarity on where AI delivers the most value. McKinsey found that "organizations are scaling AI in the functions with the greatest potential value for their industries," but this varies dramatically by sector. Technology companies report using AI agents in software engineering far more commonly than other industries, while consumer goods and retail companies are deploying agents in sales and marketing, and advanced manufacturing companies are using them in supply chain and inventory management.

The gap between large and small enterprises is also widening. Companies with more than $1 billion in annual revenue are deploying agentic tools at significantly higher rates. Year over year, the percentage of companies earning more than $1 billion who said they were "at least scaling" AI agents grew 13 percentage points, to 40%. Those making less than $1 billion who noted "at least scaling" AI grew only one percentage point, to 22%.

What Are Companies Actually Measuring for AI ROI?

When companies do report positive AI impact, the gains are often narrow and function-specific rather than enterprise-wide. Respondents cited cost reductions in specific business functions over the past 12 months, most frequently from AI use in supply chain management, service operations, and manufacturing. In customer service, businesses see an average return on investment of $3.50 for every $1 invested in AI customer service, while top performers report returns up to 8 times that figure, showing a steep gap between average and excellent implementations.

The customer service sector offers a concrete example of AI's mixed impact. A peer-reviewed study by Brynjolfsson and colleagues found that agents using generative AI resolved 15% more issues per hour on average, with gains reaching 34% among the least experienced agents. Yet 93% of consumers still prefer human support, 50% would cancel a fully AI-driven service, and 42% would pay extra for human access, according to Kinsta's 2025 survey of 1,011 U.S. consumers. This suggests that while AI improves agent productivity, it may not improve customer satisfaction or loyalty.

How Can Companies Bridge the Adoption-to-Earnings Gap?

One emerging approach involves a new category of AI models designed to make fast, inexpensive judgments rather than generate text. A model called Jev, discussed in The AI Daily Brief, is built to help with business automation, allow agents to check their work, and coordinate decisions across teams. This "judgment model" approach could reshape how companies think about AI deployment, moving away from broad text generation toward targeted decision-making.

McKinsey's research suggests that the highest-impact AI users treat AI like a reasoning partner, and those skills can be taught at scale, according to research from KPMG and the University of Texas at Austin cited in The AI Daily Brief. This implies that the problem is not the technology itself, but how organizations train their teams to use it.

For companies looking to move from piloting to scaling AI, McKinsey found that chatbots have been widely scaled and can be an accessible starting point. However, successful deployments look different across industries, and organizations need to identify the functions with the greatest potential value for their specific business model.

Steps to Improve AI ROI and Execution

• Audit Your AI Strategy: Only 12% of enterprise contact centers have a fully optimized AI strategy, according to USAN's 2026 research. Conduct a comprehensive review of where AI is deployed, what it is actually doing, and whether it aligns with your highest-value business functions.
• Train Teams to Use AI as a Reasoning Partner: Research from KPMG and the University of Texas at Austin shows that the highest-impact AI users treat AI like a reasoning partner, and those skills can be taught at scale. Invest in training programs that teach employees how to collaborate with AI tools rather than simply using them as automation.
• Focus on Function-Specific Value: Rather than deploying AI broadly, identify the specific business functions where AI delivers the most value for your industry. Technology companies should focus on software engineering, retail on sales and marketing, and manufacturing on supply chain management.
• Measure Narrow Wins First: Start with functions where AI has already proven cost-saving potential, such as supply chain management, service operations, and manufacturing. Build momentum with measurable, function-specific gains before attempting enterprise-wide transformation.
• Consider Judgment Models for Decision-Making: Explore newer AI approaches, such as judgment models, that are designed to make fast, inexpensive decisions and coordinate across teams, rather than relying solely on text-generation models.

What Does the Future Hold for Enterprise AI Profitability?

The outlook suggests that AI's financial impact will improve, but not without deliberate strategy. McKinsey found that 44% of companies have reached the AI scaling phase, and almost every sector represented expected to see their companies' AI investment grow in the next year, with the highest anticipated increases in pharma and medical products, insurance, and financial institutions.

However, cost pressures are beginning to constrain AI use. "AI-related operating costs are beginning to constrain AI use for about one in five organizations," according to McKinsey. This suggests that companies will need to become more disciplined about where they deploy AI and how they measure its return. As McKinsey senior fellow Michael Chui noted, even as per-token costs have declined, the number of tokens consumed and generated has increased even faster, and because these use cases have demonstrated value, organizations are planning to invest more, even as they develop new disciplines for optimizing ROI from their AI expenditures.

The gap between AI adoption and AI profitability is not a technology problem; it is an execution problem. Companies that treat AI as a strategic tool aligned with their highest-value business functions, train their teams to use it effectively, and measure results carefully will likely see the earnings impact that has eluded most organizations so far. For now, the majority of companies are still in the early stages of that journey.