There is an infographic that has been widely shared in enterprise leadership circles that maps 90 years of AI history — from Turing’s foundational work in the 1930s and 1940s, through the six documented hype-and-winter cycles that followed, culminating in the current AI moment as the seventh cycle.
The infographic is factually accurate. Every one of the previous six cycles happened. Each was preceded by dramatic hype about AI’s imminent capabilities. Each collapsed into a “winter” of reduced funding, discredited claims, and disillusioned researchers.
The implicit conclusion the infographic invites is that the current moment is cycle seven. That the boom is the setup for the collapse. That responsible executives should treat AI hype in 2026 with the same skepticism that would have been appropriate during the previous six cycles.
This conclusion, in my direct assessment, is structurally wrong.
The current moment is not a cycle. It is a floor.
Cycle versus floor
A cycle is a pattern of hype and disillusionment that returns to approximately the same baseline. Cycles have amplitude but no trend. The 1970s AI winter left researchers at approximately the same capability level as 1960s AI researchers. The 1980s expert-systems winter left researchers at approximately the same capability level as pre-boom.
A floor is different. A floor is a permanent step-change in capability. After the floor is established, subsequent cycles happen around a new baseline, not the old one. The transistor established a floor. The internet established a floor. The smartphone established a floor.
Cycles are common. Floors are rare. Most technology news is cycle news, which is why the infographic’s framing feels intuitive.
The specific question, for every executive right now, is whether the current AI moment is producing a floor or a cycle. If it is producing a cycle, wait it out. If it is producing a floor, position for the new baseline before it is fully established.
Three diagnostic tests
Three specific tests distinguish floors from cycles. All three are, in the current AI moment, pointing away from cycle and toward floor.
Test one: the cost curve.
Cycles have flat or declining cost efficiency. The capability produced by 1970s AI at a given cost was approximately the same as the capability produced by 1960s AI at the same cost — the winter came because the promised capability improvements didn’t arrive.
Floors have exponentially improving cost efficiency. Transistor computing got exponentially cheaper per unit of capability every 18-24 months for decades. Internet bandwidth got exponentially cheaper per unit of data transferred. Smartphone compute got exponentially cheaper per unit of processing.
AI compute is currently getting exponentially cheaper per unit of capability. The DeepSeek $5.6M training run in January 2026 is a specific data point in this curve. The curve is not flattening. It is accelerating.
Test two: the deployment pattern.
Cycles have narrow, elite deployment. 1980s expert systems were deployed in a small number of specialized enterprise applications. When those applications underperformed, deployment collapsed.
Floors have broad, distributed deployment. The transistor was deployed in every electronic device. The internet was deployed in every home and business. The smartphone was deployed in every pocket.
Current AI deployment is broad and distributed. Every knowledge worker with an internet connection has access. Every enterprise is running pilots. The deployment pattern is a floor pattern, not a cycle pattern.
Test three: the research trajectory.
Cycles have plateauing research output. In previous AI winters, the underlying research stopped producing capability improvements before the funding cuts arrived. The winter was a lagged response to a research plateau.
Floors have accelerating research output. Transistor research continued producing capability improvements for decades after the initial breakthrough. Internet research produced continuous protocol and infrastructure improvements. Smartphone research produced continuous chip and interface improvements.
Current AI research is accelerating, not plateauing. New capabilities, new architectures, and new deployment patterns are emerging on a monthly basis. There is no plateau in the underlying research trajectory.
What this means for executives
All three tests point away from cycle and toward floor. Which means the current AI moment is not the seventh iteration of a historical pattern. It is the establishment of a new baseline.
The executive posture appropriate for a cycle — wait it out, protect existing capital investments, avoid overcommitting — is exactly wrong for a floor.
The executive posture appropriate for a floor — position early, absorb the new capability into core operations, plan for a fundamentally different competitive baseline — is what a small number of executives are actually doing right now. Most are not.
Three practical questions
One: which of the three tests would you argue against? If you believe the current AI moment is a cycle, name specifically which of the three — cost curve, deployment pattern, research trajectory — you think is a cycle pattern rather than a floor pattern. The specificity forces the argument.
Two: what would your investment thesis look like if you accepted the floor framing? Higher, faster, more integrated. Most executive AI investment theses are calibrated to cycle assumptions. Recalibrate to floor assumptions and observe the change.
Three: who in your organization is arguing for the floor framing? They are, structurally, correct. Are they being heard?
The closing thought
The infographic is a compelling piece of pattern-matching. Six previous cycles. Each preceded by hype. Each followed by disillusionment. It is emotionally satisfying to conclude that the seventh cycle is underway.
Emotional satisfaction is not evidence. The three diagnostic tests are evidence. All three point at a floor.
The executives who understand this in 2026 will position early. The executives who rely on the infographic will position late. The gap between the two groups will be visible by 2029.
The world has changed. The leaders who notice will be the ones the next decade is built around.
The post the AI Boom Is Just the Latest of Six 90-Year Hype Cycles first appeared on Agile Agilist | SAFe® Gold Partner for Agile Transformation, Innovation & Leadership Training.