Ask a financial services company where it uses advanced iterations of artificial intelligence in its daily operations the most, and you’ll get a quick answer: the back-office function of data and technology. Nearly all, or 95%, of banks and insurers are broadly deploying or more intensively embedding new AI tools in that process. (“New AI tools” means technology centered on large language models and agentic AI released since November 2022, when ChatGPT’s first model emerged and almost immediately reached more than 1 million users.)
Now ask a company in media and advertising or healthcare and medical services the same question. You’ll hear a more muted—but only slightly—response. More than eight in 10 companies in these two industry groups report those same high levels of AI use for data and technology.
Next, turn the question to how companies are using new AI for other business processes. The answers are very different. New AI tools for the growth and revenue function? Among financial services firms and insurers, only three in 10 are broadly deploying or directly embedding the technology. The same is true for nearly as many media and advertising firms, but with a difference. Nearly one in 10 have embedded AI tools in these two functions, almost double the rate of financial services companies and insurers. Meanwhile, new AI usage in other key functions—supply chains and fulfillment, risk and compliance, and HR and the workforce—is nestled in either preliminary exploratory phases or limited development.
In other words, how different industries use the newest forms of AI depends on two factors: which business function it assigns the technology to work on, and which industry it’s in.
National survey evidence shows generative AI, a less sophisticated tool than fully autonomous agentic AI, being adopted as rapidly as personal computers were in the 1980s, with potential but uncertain productivity growth implications. Yet its diffusion across companies varies widely. And still, amid that variation, enterprises consistently award high marks to AI’s value. Nearly all companies across the three industry groupings say AI is performing quite well in the functions where it’s deployed and rate their return on investment over the past 12 months as positive.
Almost all firms say they’re now getting at least some financial return on their uses of new AI, but almost none say the investment has fully paid off. Just 5% to 10% say they’re currently getting a full return. At least half of firms in every industry group date full payback at five to six years.
These are some of the findings in “How Services Enterprises Are Putting New Forms of AI to Work,” a PYMNTS Intelligence exclusive report, based on a June 2026 survey of 60 senior technology executives, primary AI decision-makers at U.S. companies with at least $1 billion in annual revenue.
This is what we learned.
Key Findings
New AI usage is scaled in two enterprise business functions and lagging or stalled in the rest.
New AI is broadly deployed or embedded in data and technology processes (81%–95% of firms). For payments and finance processes, most financial services and healthcare firms have scaled new AI, while among media firms, limited deployment remains the most common stage. Meanwhile, AI has entered growth and revenue, corporate and strategy, and product and customer experience functions without spreading beyond limited deployment, while supply chain, risk and HR remain in evaluation, with most companies across industries not having moved past the exploration stage.
Firms rate new AI tools a success where they run them.
Eighty-six percent to 95% of firms call new AI very or extremely effective in the functions where they use it. Measured against their own expectations, it performed better than expected more often than as expected on all 20 business goals tracked, with business expansion, staffing mix, digital experience optimization, and customer support and service consistent soft spots. In a separate question about the last 12 months, 90% to 100% of firms in every sub-industry rated the ROI of new AI as somewhat or very positive.
Returns read positive, payback sits years out and investment budgets rise anyway.
Ninety percent to 100% of enterprises say they’re now getting some kind of return on their uses of new AI, but almost none say the investment has fully paid off. Only 5% to 10% say they’re currently getting a full return. At least half of firms in every industry group date full payback at five to six years.
Still, budgets are rising: At least eight in 10 enterprises across all surveyed industry groups expect to increase their AI investments next year, and none expect a cut.
New AI Deployment Varies by Business Function
New AI usage is scaled in two enterprise business functions and lagging or stalled in the rest.
Broad or embedded new AI is the norm in data and technology (81% to 95% of firms) and follows unevenly in payments and finance.
PYMNTS Intelligence finds that for data and technology processes, 95% of surveyed financial services firms, 84% of healthcare firms and 81% of media firms report broad or embedded deployment of new AI tools—what we call a deep tier. For payments and finance functions, nine in 10 financial services firms and 63% of healthcare firms have scaled new AI tools versus 43% of media firms, for which limited deployment is still the most common stage.
Significantly, the deep tier runs through functions where technical teams own the pipelines and outcomes are auditable.
Inside these functions, specific activities show where new enterprise AI is concentrated. Among firms scaled in data and technology, the most common are security monitoring (77%), infrastructure optimization (68%), data ingestion and cleansing (68%) and AI governance tooling (63%). In payments and finance, the leaders are treasury and liquidity management, accounts payable and receivable, automation, and pricing optimization.
Corporate and customer-facing: the business functions where new AI has entered but not yet scaled
In growth, corporate and customer-facing functions, limited deployment of new AI tools is the most typical stage (43% to 63%).
This middle-tier usage of functions is defined by the presence of AI without scale. In growth and revenue, corporate and strategy, and product and customer experience processes, most firms have moved past evaluation. But the most common stage in every sub-industry is limited deployment, meaning new AI is used in isolated or specific spots (by 43% to 63% of firms per function).
Even the strongest showing in this middle tier is weak. Healthcare firms reach broad or embedded deployment for growth and revenue processes more often, and that’s still just by 37% of these companies. In product and customer experience, financial services does worse: Only 15% have broad or embedded deployment.
The specific activities firms most often cite within the growth and revenue functions are customer segmentation and churn prediction. For corporate and strategy functions, market and competitive intelligence and executive briefing preparation top the list. For product and customer experience functions, customer service chatbots and in-product recommendations lead. The base sizes here are small, so these indications are named rather than quantified.
This middle tier of new AI use is where the next wave of scaling decisions sits: The tools are in the door, and the question is which isolated deployments will expand.
Most enterprises are still in the early stages of AI development for supply chain, risk and compliance and HR functions.
For supply chain and fulfillment, risk and compliance, and HR and workforce processes, most enterprises haven’t moved new AI past exploration (53% to 67%).
Most large companies remain at the exploration stage (53% to 67%), and broad or embedded deployment reaches at most 10% of firms (financial services in risk and compliance), with zero scaled deployment in five of the nine cells.
The places where the handful of scaled firms do deploy are order fulfillment and returns optimization in supply chain, credit and third-party risk assessment in risk and compliance, and workforce and headcount planning in HR. (The base sizes are too small to quantify.)
These are also the functions where stakes and oversight demands run high, which is consistent with firms keeping the newest generation of AI in evaluation the longest for its application to high-stakes processes.
Enterprises Say New AI Is Delivering Results
Firms rate new AI tools a success where they run them.
So far, enterprises are more than happy with how their use of the technology is unfolding.
Asked to rate the effectiveness of new forms of AI in the functions where they use them, 95% of financial services and insurance firms, 95% of healthcare and medical firms and 86% of media and advertising firms choose very or extremely effective. The scale included two options to offer negative ratings of not at all effective or slightly effective, and no firm in the sample chose either.
The difference between sub-industries is one of intensity rather than direction. Half of financial services firms and insurers (50%) pick the top of the scale, extremely effective, versus 11% of healthcare and medical firms and 10% of media and advertising firms, whose ratings concentrate at a lower but still enthusiastic very effective. That intensity gap tracks financial services’ deeper deployment. It’s the only industry group where a large share of firms embed new AI into day-to-day operations (50% in data and technology).
These are self-assessments by the executives who own the programs, so they read as a measure of internal conviction rather than verified performance.
New AI is outperforming enterprise expectations.
New AI is beating expectations rather than merely meeting them: enterprises rate AI’s performance as better than expected on 20 goals, topped by competitive position (87%) and business expansion, staffing mix, digital experience optimization, and customer support and service at the bottom (65%).
Compared with what they expected, firms say new forms of AI have performed “somewhat or much better than expected” more often than as expected on every one of the 20 goals measured. The strongest beats sit at the strategic level: competitive position (87%) and market adaptability (85%), with speed to market, production costs and data security all at 80%.
Enterprises’ ratings of “worse than expected” are rare and concentrated in staffing, customer experience and business expansion goals. Digital experience optimization is the one goal where roughly one in five firms in each sub-industry report performance below expectations, and staffing mix draws 24% worse than expected among media firms, a directional read. The consistency of the beat signals that the expectations firms set for this technology are being cleared, at least in the judgment of the executives running it.
Long-Term Confidence Keeps AI Investment Growing
Returns read positive, a full payback sits years out and investment budgets rise.
Most firms say the real payback on their AI investments will come in five to six years.
That the long clock coexists with positive report cards underscores a tension.
In a separate question whose data is not shown in the accompanying figure, firms rated the overall ROI their business has seen from new forms of AI in the last 12 months. Ninety percent to 100% in every sub-industry answered somewhat or very positive. Yet on the question of when investments will deliver a return, only 5% to 10% chose “already delivering ROI,” and the most common answer everywhere is five to six years (50%, 63% and 62%).
The likely reading is that firms are seeing incremental returns now, while dating full payback later, much as a newly planted fruit tree bears some fruit each year long before it has covered the cost of planting and tending. The two questions read against each other, so we offer this as interpretation rather than as a finding.
Budgets keep climbing despite a long payback clock.
While the payback clock may be down the road, enterprises overwhelmingly plan to increase their AI investment budgets. At least eight in 10 enterprises expect to boost new-AI spend next year, and none expect to slow down from their current levels.
No firm in the sample expects to decrease its budget for new forms of AI. Advertising and media is the most aggressive industry, with 24% expecting a spending increase of more than 15%.
Strategic and competitive positioning drives new AI spending.
While different industries have different rationales for boosting their new AI spending, the shared driver for seven in 10 enterprises collectively centers on strategy and competitive position. Financial discipline splits the sectors, led by healthcare (63%), which cites monetary ROI as the main reason, above financial services (40%) and media and advertising (24%). Meanwhile, advertising and media leans on risk reduction (76%). Media also shows the largest share of enterprises funding new AI pilots without a formal ROI requirement. Firms could cite more than one justification for investing in new forms of AI.
This data measures why firms say they invest, not which capabilities they’re building. All figures are directional at these sample sizes.
Read More
PYMNTS Intelligence is the leading provider of information on the consumer trends driving innovation in consumer finance, digital payments and financial inclusion. To stay up to date, subscribe to our newsletters and read our in-depth reports.
Methodology
“The Enterprise AI Payback Curve: Adoption Accelerates as Returns Take Shape” is the latest installment in the Enterprise AI Benchmark Report series. PYMNTS Intelligence surveyed 60 verified senior technology executives who are primary decision-makers about AI usage in June 2026. Companies have at least $1 billion in annual revenue. Industries surveyed were financial services and insurance, healthcare and medical, and media and advertising. “New forms of AI” refers to tools released since November 2022 that rely on large language models and agentic AI. The deployment stage is rated per function on a four-stage scale of exploring, limited deployment, broad deployment and embedded. Business functions covered were corporate and strategy, payments and finance, growth and revenue, product and customer experience, supply chain and fulfillment, risk and compliance, data and technology, and HR and workforce.
