The State of AI in 2025: From Pilots to Transformation—What Separates the High Performers?

Three years since the introduction of generative AI tools signaled a new era in artificial intelligence, AI tools are now commonplace within organizations. Almost all survey respondents indicate that their organizations are using AI. In fact, 88 percent of organizations report regular AI use in at least one business function. However, despite this widespread adoption, the path toward realizing material enterprise-level benefits remains uneven. Our latest data reveals a landscape defined by both wider use—including a growing proliferation of agentic AI—and **stubborn growing pains**, as the transition from pilots to scaled impact remains a work in progress for most organizations. ## The Scaling Challenge and the Rise of AI Agents While AI use continues to broaden, with respondents increasingly reporting use in multiple business functions—more than two-thirds now say their organizations are using AI in more than one function—most are still in the early stages of scaling AI and capturing enterprise-level value. **Nearly two-thirds of respondents** say their organizations have not yet begun scaling AI across the enterprise. Only about one-third report that their companies have begun to scale their AI programs. This lack of depth is why only 39 percent of respondents report seeing EBIT (Earnings Before Interest and Taxes) impact at the enterprise level, although positive leading indicators, such as use-case level cost and revenue benefits, are being reported. **Key points on scaling:** * **Size matters:** Larger companies (those with more than $5 billion in revenue) are nearly half (47%) more likely to have reached the scaling phase compared with smaller organizations (29% of those with less than $100 million in revenues). * **Innovation boost:** A majority of respondents (64%) say that AI is enabling their innovation. * **Cost vs. Revenue:** Cost benefits from individual AI use cases are most often seen in **software engineering, manufacturing, and IT**. Revenue increases are most commonly reported in **marketing and sales, strategy and corporate finance, and product and service development**. In addition to general AI adoption, organizations are displaying high curiosity and exploration regarding **AI agents**—systems based on foundation models capable of planning and executing multiple steps in a workflow. Sixty-two percent of survey respondents say their organizations are at least experimenting with AI agents. However, the enterprise use of agents is not yet widespread: only 23 percent report scaling an agentic AI system somewhere in their enterprises, usually in only one or two functions. Looking at individual functions, agent use is most commonly reported in **IT and knowledge management**. By industry, use is most widely reported in the technology, media and telecommunications, and healthcare sectors. ## The High Performer Playbook: Transformation, Not Just Efficiency Organizations that achieve the greatest success with AI—defined as AI high performers who report more than 5 percent of EBIT impact and "significant value" from AI use—share a common playbook. While 80 percent of all respondents say their companies set efficiency as an objective of their AI initiatives, the high performers often set **growth or innovation as additional objectives**. In fact, high performers are more than three times (3.6x) more likely than others to say their organization intends to use AI to bring about **transformative change** to their businesses. This transformative ambition helps them realize value by fundamentally reimagining their businesses. Crucially, redesigning workflows is a key success factor: * High performers are nearly three times (2.8x) as likely as others to report they have **fundamentally redesigned individual workflows** in their deployment of AI. * They are also much more likely to be scaling their use of AI agents across most business functions compared to their peers. Moreover, high performance is linked to strong organizational practices: 1. **Senior Leadership Commitment:** High performers are three times more likely to strongly agree that senior leaders at their organizations demonstrate ownership of and commitment to their AI initiatives, including role modeling the use of AI. 2. **Human-in-the-Loop Processes:** High performers are more likely to have defined processes to determine how and when model outputs need **human validation** to ensure accuracy. 3. **Investment:** High-performing organizations are investing significantly more. More than one-third of high performers commit over 20 percent of their digital budgets to AI technologies, nearly five times (4.9x) the rate of other respondents. ## Workforce Impact and Emerging Risks As organizations expand their use of AI, expectations regarding the effect on the workforce in the coming year vary. Overall, respondents have differing expectations for AI's impact on the total workforce size: **32 percent expect decreases** (3% or more), **43 percent expect no change**, and **13 percent predict an increase** of that magnitude in the next year. Respondents at larger organizations are more likely to expect an AI-related reduction in workforce size. Despite potential decreases, organizations are actively hiring for AI-related roles, with **software engineers and data engineers** being the most in demand. Larger companies are twice as likely to hire roles that integrate, model, and industrialize data. In parallel with adoption, efforts to mitigate AI risks are becoming more common. The share of respondents reporting mitigation efforts for risks such as regulatory compliance and personal privacy has grown since 2022, with organizations now acting to manage an average of four AI-related risks, up from two in 2022. **Inaccuracy** is the AI-related risk that organizations most often report having experienced (nearly one-third of all respondents) and are working to mitigate. Interestingly, high performers are more likely than their peers to report negative consequences—particularly related to intellectual property infringement and regulatory compliance—but they also mitigate these risks at a higher rate. The message is clear: while AI's full promise remains ahead, achieving significant, enterprise-wide value requires moving beyond an incremental, efficiency-first mindset. The highest-performing organizations are realizing results by treating AI as a catalyst for transformation, driving innovation, and systematically redesigning the way work gets done.