Introduction: From Hype to Doubt

Gartner’s Hype Cycle has long provided a useful framework for understanding how technologies move from exuberant expectation to more grounded application. In 2025, Generative AI appears poised to enter the Trough of Disillusionment—a stage where inflated promises collide with practical limitations, investor impatience, and regulatory scrutiny. The critical question is whether this represents a temporary cooling, or the onset of a deeper, second “AI Winter.”

Creative Destruction and AI’s Place in the Cycle

The Nobel Prize in Economics recently awarded to Philippe Aghion and Peter Howitt underscores a central theme: innovation thrives on creative destruction. Disruptive technologies replace older ones, driving growth. Yet, their model also cautions that not every innovation wave achieves transformative scale.

Applied to AI, the comparison with past revolutions such as electricity or the steam engine seems premature. Instead, parallels to the dot-com bubble are stronger: even as many internet firms collapsed, the infrastructure endured and later underpinned global digitalization. AI, by contrast, demands extraordinary computational resources, generates inconsistent returns, and struggles to scale in ways that justify the current surge in capital investment.

Gartner’s Warning Signs

According to Gartner’s projections, GenAI is following a predictable trajectory:

• Overinvestment and Hype: Massive inflows of capital into startups, with valuations of Big Tech swelling on AI promises.
• Early Disillusionment: Enterprises are discovering that GenAI deployments are costlier, less reliable, and less scalable than promised.
• Energy and Resource Constraints: The ballooning size of models collides with sustainability concerns, sharpening regulatory focus.
• Investor Skepticism: Without demonstrable ROI, funding streams risk drying up, echoing past technology contractions.

These dynamics suggest Gartner’s Trough of Disillusionment is less a theoretical stage and more an imminent market correction.

Strategic and Governance Implications

The analysis points toward several implications for leaders, aligning with Gartner’s own cautions:

• Enterprise Strategy: Expect a pivot from “AI everywhere” to AI where it pays off. Automation and narrowly scoped, explainable AI tools may regain favor over ambitious but unreliable GenAI pilots.
• Policy and Regulation: Governments may treat AI less as an economic growth engine and more as a risk technology, with proposals such as compute licensing to rein in energy use.
• Research: The pendulum may swing back toward smaller, interpretable, neuromorphic systems, reflecting demand for transparency, reliability, and efficiency.
• Education & Workforce: Universities may reallocate funding away from costly LLM labs and toward AI governance, ethics, and computational sustainability.

Is an “AI Winter” Coming?

The key difference between a trough and a true “AI Winter” lies in resilience. The internet survived its bust because its core utility was undeniable. AI’s future depends on whether its current infrastructure proves equally indispensable. If adoption contracts but foundational AI capabilities remain embedded in workflows, this will be a trough. If disillusionment leads to widespread abandonment of investment and research, it may feel more like a second winter.

Either way, Gartner’s projection is clear: the peak of inflated expectations is behind us. The next stage will separate durable, economically grounded AI applications from speculative excess.

Authored by Alex Wyka, EA Principals Senior Consultant and Principal