Open LinkedIn today and you will find two opposing camps. One believes artificial intelligence will replace millions of jobs, mint trillion-dollar companies, and reshape civilisation. The other argues AI is simply another hype cycle destined to end like the dot-com bubble. History suggests both are right.
Because every technological revolution follows the same pattern. Not identical technologies — identical human behaviour. For more than five hundred years, every transformational invention has produced the same sequence: excitement, then fear, then enormous wealth, then spectacular failure, and ultimately a completely new economy. AI is not the exception to that sequence. It is the latest instance of it.
This briefing does something the daily AI discourse rarely bothers to do: it puts the current moment on a timeline that starts in 1450. Once you can see eleven revolutions stacked against a single curve, the argument about whether AI is "real" or "a bubble" dissolves. It is both, in sequence — exactly as railways were, exactly as the internet was. The useful question is not if but where on the curve we are standing.
The five stages every revolution repeats.
Looking back across centuries, technological revolutions almost always move through five predictable phases. The labels change with the fashion of the era, but the shape of the curve does not.
Usually ignored, usually underestimated, usually the work of researchers rather than corporations. The printing press, the steam engine, electricity, the transistor, the internet, and large language models all began here — as curiosities before they were industries.
Money floods into the market. Thousands of new businesses appear. Every entrepreneur wants exposure, every investor wants allocation, and every board suddenly discovers it has a strategy for the new technology.
Infrastructure is massively overbuilt. Too many companies chase the same problem. Valuations disconnect from fundamentals and the narrative becomes "this changes everything." Ironically, that statement is usually correct. The timing is not.
Weak companies disappear, capital dries up, and markets panic. Journalists announce that the bubble has burst. They have said it before — many times — and they have been early on the death of the technology every time.
Years later, the technology simply becomes part of everyday life. Railways, electricity, the internet, cloud computing, GPS, smartphones. Nobody calls them disruptive. They are simply infrastructure — the ground the next revolution is built on.
500 years, eleven revolutions.
Before we walk them one by one, here is the whole sweep on a single line. Each crest is one revolution running through its own five-stage cycle; the line never returns to zero because each wave settles into infrastructure that the next one is built upon. The eleventh crest — AI — is still rising.
Now the same five-stage curve, applied to each revolution in turn — with the year and the defining event at every phase. Read them as a sequence, not a list. The pattern is the point.
The Printing PressInformation goes from scarce to abundant.
Gutenberg's press democratised information for the first time. Books became dramatically cheaper, ideas travelled across Europe faster than ever, and the Renaissance, the Reformation and modern science all accelerated in its wake. Even then, critics warned it would spread misinformation and undermine established authority. Sound familiar?
Navigation & Global TradeThe world gets smaller.
Better ships, better navigation and better maps produced the first truly global economy. Trade routes reshaped Europe, empires expanded, and entire industries grew up around maritime commerce. Technology did not merely improve transportation — it rewired global economics.
The Industrial RevolutionMachines multiply human power.
Steam engines changed everything. Factories replaced workshops, mechanisation replaced manual labour, and productivity and urbanisation exploded together. Millions feared machines would destroy employment; instead they changed it — and history's first great automation debate began.
RailwaysThe internet of the 19th century.
Railways connected cities, slashed transport costs and created national markets, triggering a wave of investment unlike anything before it. Then the railway bubble burst and thousands of investors lost fortunes. The tracks remained — indispensable infrastructure for the century that followed.
ElectricityPower on demand changes everything.
Electricity transformed factories, homes, healthcare, manufacturing and communication in turn. Investment surged and speculation followed the same familiar path. Once again, the technology survived long after many of its earliest businesses did not.
AutomobilesPersonal mobility redefines society.
Cars reshaped cities, created suburbs and spawned entirely new industries — oil, insurance, highways, logistics and tourism among them. The automobile was never merely transportation. It redesigned how society was built and where people lived.
The TransistorThe building block of the digital age.
Arguably the most important invention of the twentieth century. Without it there are no computers, no smartphones, no cloud, no AI. The transistor became the invisible foundation beneath every digital business that followed.
Personal ComputingComputing leaves the lab.
Computers left the laboratory, entered the office and then the home. Software became an industry, productivity accelerated, and a new generation of monopolies — Microsoft, Apple, Oracle, Adobe — was born. The digital economy had arrived.
The InternetThe world gets connected.
Nothing illustrates the cycle better. Thousands of startups launched, capital became almost unlimited, and companies with no revenue reached billion-dollar valuations — until the dot-com crash. Many concluded the internet had failed. Instead Amazon survived, Google emerged, the cloud followed, and the internet became civilisation's nervous system.
Mobile & CloudComputing anywhere, on demand.
Smartphones put supercomputers into billions of pockets while cloud computing removed the infrastructure barrier to building software. Products became subscriptions, every industry became digital, and Fortune 500 incumbents were disrupted by companies that had barely existed a decade earlier.
AI & RoboticsThe automation of intelligence begins.
Now we arrive at AI. Unlike previous revolutions, it is not simply another software category — it is the automation of knowledge work itself. For centuries we automated muscle, manufacturing, transport and information. Now we are beginning to automate reasoning, and on the weight of capital already committed, we are standing in the overinvestment phase. The correction and infrastructure stages shown here are forecast, not history.
Notice what the correction never does: it never sends the technology back. The railway bubble ruined thousands of investors and left behind a national rail network. The dot-com crash erased a decade of paper valuations and left behind Amazon, Google, and the cloud. The bust clears the speculation; the infrastructure remains. That is the mechanism, repeated eleven times.
Why this revolution may be different.
For all the reassurance the pattern offers, one distinction deserves to be taken seriously. Previous revolutions changed how humans worked. AI changes who — or what — performs the work. That is a different kind of shift.
For centuries humanity automated muscle, then manufacturing, then transportation, then communication, then the storage and retrieval of information. In every case, software and machines helped people do the work. What is beginning now is the automation of reasoning itself: systems that do not assist the knowledge worker so much as perform the knowledge work, with a person supervising rather than operating. Instead of software helping people, people increasingly supervise software.
That distinction matters because it makes almost every business function addressable by an intelligent agent — sales, marketing, finance, legal, operations, research, customer success, engineering. The previous revolutions redistributed labour between humans. This one proposes to redistribute it between humans and machines. The economic implications are unlike anything in the earlier chapters, even if the market behaviour around them will rhyme perfectly.
Every prior revolution
Changed how we workMuscle, manufacturing, transport, communication, information. Software extended human capability. The human stayed in the loop as the operator.
The AI revolution
Changes who does the workReasoning itself becomes executable. The human moves from operator to supervisor. Every business function becomes something an agent can run.
The coming shakeout.
Will there be an AI correction? Almost certainly. History effectively demands one, and the preconditions are already visible. Capital has arrived faster than enterprise adoption. Too many startups are solving identical problems. Another foundation model appears roughly every week. Valuations assume flawless execution across a decade, and markets rarely reward that assumption for long.
So the honest forecast is not comfortable, but it is unremarkable in historical terms. Some AI companies will disappear. Others will consolidate. A great many products that are pitched today as companies will turn out to be features of someone else's platform. The strongest platforms will remain and compound. This is not pessimism about AI — it is precisely what stage four looks like from the inside, and it has preceded every period of durable value creation on this list.
The real winners.
History offers one consistent lesson about who captures the value. The winners are rarely the companies with the loudest marketing during the boom. They are the organisations quietly building enduring infrastructure while everyone else debates valuations.
During the railway boom, the tracks mattered. During electrification, the grid mattered. During the internet, the cloud mattered. During AI, the durable infrastructure may not be chips or models alone — those layers are being commoditised in real time by the very overinvestment now under way. The more defensible layer may be the intelligence systems that let an organisation sense its market, understand its buyers, orchestrate revenue, and make decisions in real time. The models are becoming a utility. What you build on top of the utility is where ownership accrues.
The GTM Bench perspective.
At GTM Bench Review we read AI not as another software trend but as the next layer of economic infrastructure. Enterprise software digitised processes. The internet connected the world. Cloud computing made software infinitely scalable. AI is digitising expertise itself — and that is a change of the same order as the five turns of the wheel that came before it.
The companies that define this era will not simply ship better chatbots or larger models. They will build the operating systems for autonomous organisations — systems where human talent and AI agents collaborate to sense demand, orchestrate revenue, and make decisions at a tempo no purely human team can match. This connects directly to the thesis this Review has developed across recent issues: the seven-layer AI industrial stack and the unowned industry layer that sits above the hyperscalers. The 500-year pattern tells you a correction is coming. The stack tells you where the value will settle once it does.
Every technological revolution has created extraordinary fortunes, but those fortunes were rarely made by timing the peak. They were made by recognising which innovations would become indispensable infrastructure after the excitement faded. The question facing today's leaders is therefore not whether AI is in a boom — it almost certainly is. The more useful question is this: when historians write about the AI revolution fifty years from now, which companies will they credit with laying the foundations of the new economy, and what are you building today that will still matter when today's hype cycle has become tomorrow's history?
The operator's takeaway
Over five centuries, humanity has reinvented how we communicate, travel, manufacture, compute, and connect. Each revolution sparked speculation, created bubbles, and eventually reshaped civilisation in ways few could foresee at the peak. AI is the latest chapter in that story — and plausibly the most consequential, because for the first time we are not merely inventing tools that extend human capability. We are building systems that replicate elements of human cognition.
If that trajectory holds, future historians may divide the timeline into before AI and after AI. The task for today's leaders is not simply to adopt the technology, but to build organisations capable of thriving through the entire cycle — the exuberance, the correction, and the long infrastructure phase that has followed every great revolution before this one. Time the peak and you will be wrong. Build for the infrastructure phase and you will still be standing when it arrives.