Six months of AI investment. Pilot projects are scattered across departments. ROI numbers that are… underwhelming. CFOs are asking pointed questions. Headlines declaring “AI bubble bursting.” VPs of Innovation are updating their resumes.
This is exactly how transformative technology adoption works. We’ve run this script before. Everyone who panics during Act Two misses the entire plot.
The 1995 Parallel
February 1995. Two competing narratives about the internet’s future.
The utopians: “Everything will be digital within two years. Newspapers are dead. Retail is finished. The information superhighway changes work forever — immediately.” Investors poured billions into companies with “.com” in their names. The short-term promise was a revolutionary transformation by 1997.
The skeptics: Newsweek published Clifford Stoll’s rebuttal: “No online database will replace your daily newspaper.” He was a credentialed astronomer and internet pioneer, dismissing the long-term impact entirely.
Both were wrong about timing.
By 2000, the dot-com crash had proven the utopians had overestimated the short-term transformation. Newspapers didn’t die in two years. E-commerce was clunky. Remote work remained niche. The immediate ROI was disappointing.
By 2025, the skeptics’ predictions look absurd. The five largest companies — Apple, Microsoft, Alphabet, Amazon, Meta — are internet-native or internet-transformed. Google’s market cap exceeds $2 trillion. Newspapers didn’t disappear, but print advertising revenue dropped 80% over 20 years.
Everything promised in 1995 eventually arrived. The utopians were right about the destination, wrong about the timeline. The skeptics were right about the timeline, wrong about the destination.
The Trough We’re In Right Now
The current AI disappointment statistics are brutal — and completely predictable:
In July, MIT researchers found that 95% of businesses that tried implementing bespoke AI systems hadn’t scaled them beyond pilot stage after six months. The study also revealed that around 90% of companies had an “AI shadow economy” where employees used personal chatbot accounts outside official pilots — suggesting the problem isn’t AI capability but organizational readiness.
This isn’t failure. This is Amara’s Law in action: we overestimate the impact of technology in the short term and underestimate it in the long run.
This is the trough of disillusionment — where most transformations lose participants. Companies chasing quarterly wins abandon projects. Executives expecting immediate ROI cut funding. Teams are pivoting to the next trend.
Every previous technology transformation followed the same pattern: winners don’t chase demos. They build foundations during the trough.
What 2040 Probably Looks Like
If the internet pattern repeats — and all evidence suggests it will — the largest companies by market value in 2040 will be AI-native or AI-transformed. Not because AI is magic. Because foundational technology shifts create competitive moats that compound over decades.
The winners won’t be the ones with the flashiest 2025 pilots. They’ll be the ones who used the 2025–2030 trough to build:
Data architecture that AI can actually use. Not data warehouses designed for quarterly reports. Real-time, accessible, clean data pipelines that enable learning systems to improve.
Organizational structures that match AI capabilities. Distributed decision-making. Autonomous teams. Rapid experimentation cycles. The same organizational evolution that the internet required.
Infrastructure investments that look wasteful today. The equivalent of laying fiber-optic cable in 1998, when dial-up modems were still widely used. Building capacity for a future that isn’t obvious yet.
The Strategic Opportunity Hidden in Disappointment
Competitors are reading the same disappointing statistics. Abandoning projects. Cutting budgets. Declaring AI “overhyped.”
This creates space.
No space for more demos or pilots. Space to build differently. While competitors optimize for quarterly AI wins, organizations can optimize for capability development.
The question shifts from “What’s our AI ROI this quarter?” to “What organizational capabilities do we need when AI actually works?” Because it will work. Just not on a quarterly timeline.
The infrastructure investments that matter now:
Decision rights. AI-powered analytics provides teams with real-time data. Do they have the authority to act on what they see? Or are they producing prettier dashboards for executives who still make the actual decisions?
Funding models. AI enables rapid experimentation and learning cycles. Are you funding annual projects or continuous learning?
Measurement systems. AI can track Current Value, Unrealized Value, Ability to Innovate, and Time to Market in real-time. Are you measuring what matters or what’s easy to measure?
These aren’t AI problems. They’re organizational architecture problems that AI makes visible. The trough is the perfect time to fix them — while competitors are distracted by disappointing pilots.
Punctuated Equilibrium
Evolutionary biologists Stephen Jay Gould and Niles Eldredge identified a pattern in the fossil record: long periods of relative stability interrupted by rapid bursts of evolutionary change. Technology adoption follows the same dynamics.
Pre-internet equilibrium lasted decades. Organizations optimized for physical distribution, paper-based workflows, and local markets. Stable. Predictable. Efficient within the constraints.
First punctuation: The dot-com hype (1995–2000). Massive capital deployment. Revolutionary claims. Investors overestimated the speed of transformation. The crash proved them wrong about timing.
New equilibrium: The trough (2001–2010). Appeared stable to observers. “The internet was overhyped.” But companies like Amazon, Google, Netflix, and Facebook weren’t resting — they were building infrastructure. Data centers. Algorithms. Logistics networks. Payment systems. The foundations looked wasteful in 2005. They enabled trillion-dollar market caps by 2020.
Second punctuation: Sudden dominance (2010–2025). The five largest companies became internet-native or internet-transformed. Not gradually. Explosively. Organizations with infrastructure captured disproportionate value once the critical mass was reached.
AI follows identical mechanics. We’re in the new equilibrium phase — the trough where 95% of pilots fail to scale. What appears to be stability is actually infrastructure accumulation. Organizations building decision rights, adaptive governance, continuous funding systems, and AI-ready data architecture during this phase position themselves for the second punctuation.
When will that punctuation event occur? Unknown. History suggests a 5–15-year lag from initial hype. But the timing question misses the strategic point: value accrues to those who build during equilibrium, not those who react during punctuation.
Competitors abandoning projects during the trough? They’re optimizing for the current equilibrium. Organizations building capabilities are preparing for the inevitable punctuation. The question isn’t whether transformation arrives. It’s whether you’ll have the infrastructure to capture value when it does.
Ralph Jocham is Europe’s first Professional Scrum Trainer, co-author of “Professional Product Owner,” and contributor to the Scrum Guide Expansion Pack. As an ICF ACC certified coach, he works with organizations to build Product Operating Models where strategic clarity, operational excellence, and adaptive learning create measurable competitive advantage. Learn more at effective agile.