Key Takeaways:

• Generative AI has entered the Trough of Disillusionment as real-world limitations temper expectations.
• AI agents and AI-ready data sit at the Peak of Inflated Expectations, attracting enterprise attention despite maturity gaps.
• ModelOps, causal AI, knowledge graphs, and AI risk management (TRiSM) are gaining traction toward practical enterprise value.
• Many organizations are prioritizing data readiness, with 57% reporting they’re not prepared to support AI effectively.
• Despite GenAI’s hype cooling, global IT spending is projected to grow 7.9% in 2025, driven in part by AI infrastructure.

Gartner’s latest Hype Cycle for Artificial Intelligence offers a clearer picture of where enterprise focus is heading—and where it’s fading. The report signals that the once-unquestioned excitement around generative AI is beginning to settle. Tools like large language models are still widely used but now sit in the Trough of Disillusionment, a stage that reflects growing awareness of their cost, complexity, and inconsistent ROI.

According to Gartner analysts, many generative AI projects have underdelivered, with organizations reporting a median spend of $1.9 million per initiative and widespread dissatisfaction with results. Business leaders are no longer content with novelty—they’re demanding real impact.

Meanwhile, AI agents—autonomous systems capable of perceiving, reasoning, and acting without direct user input—remain near the Peak of Inflated Expectations. Their promise to offload decision-making and streamline operations has captured C-suite attention, but real-world maturity lags behind the hype.

Another concept rising fast is AI-ready data. Enterprises are realizing that without structured, high-quality, and contextually relevant data, no AI model—no matter how advanced—can produce reliable insights. Gartner’s data shows that 57% of organizations are not yet prepared to support AI with the necessary data foundations, making data readiness a top priority going forward.

The Slope of Enlightenment—the stage where technologies start delivering practical value—is now populated with tools that focus on governance, infrastructure, and scalability. These include ModelOps (which enables lifecycle management of AI models), causal AI (which moves beyond correlation to understand causation), knowledge graphs (used for structured reasoning and context), and AI Trust, Risk and Security Management (TRiSM), which addresses the growing concern around AI risk and compliance.

What’s becoming clear is that organizations are moving from a tool-first approach to a systems-level strategy. Instead of adopting flashy AI products in isolation, companies are aligning AI with operational goals, data strategy, and risk management frameworks.

This maturing view of AI is also influencing IT budgets. While enthusiasm for generative AI has cooled, investment in AI-related infrastructure remains strong. Gartner projects global IT spending will rise 7.9% in 2025 to $5.43 trillion, in part due to enterprise investments in AI software and embedded intelligence across business applications.

Software with built-in generative features is expected to outpace traditional software growth through at least 2026. But rather than being standalone GenAI tools, these applications are increasingly embedded into core platforms—CRMs, ERPs, collaboration tools—where they can drive measurable productivity.

The 2025 Hype Cycle suggests that the AI industry is at a critical inflection point. Enterprise buyers are becoming more disciplined, evaluating AI initiatives not by their technical novelty but by their ability to deliver repeatable results at scale. Technologies that support reliability, control, and transparency are quickly rising in prominence, while those rooted in hype alone are starting to fade.

As organizations look beyond pilots and demos, success will increasingly hinge on foundational capabilities: data quality, model governance, deployment infrastructure, and integration with business workflows. The hype may be slowing, but the hard work of real enterprise AI is just beginning.