Why do new technologies swing from excitement to disappointment? Because expectations rise much faster than the technology can mature: early demos suggest everything will change at once, money and attention flood in, real-world limits appear, and interest collapses before the useful core quietly settles into everyday use. Enterprise blockchain, the metaverse and self-driving cars all followed this path, and parts of AI are on it now.
Understanding the pattern helps you decide when to experiment, when to wait and when to ignore the noise, whether you run a business, plan a career or simply want to avoid buying the wrong thing at the wrong time.
What is the technology hype cycle?
The best-known description of this pattern is the hype cycle popularised by the research firm Gartner. It has five phases: a trigger, such as a striking demo or research result; a peak of inflated expectations, when coverage and investment surge; a trough of disillusionment, when early projects disappoint; a slope of enlightenment, as people learn what the technology is actually good for; and a plateau of productivity, when it becomes ordinary infrastructure.
The model is a description, not a law. Some technologies never climb out of the trough; others skip the drama altogether. Critics also point out that it is easy to fit almost any history to the curve after the fact. Still, the shape captures something real about how human attention and money behave around new tools.
The key insight is that both the peak and the trough are distortions. At the peak, people overestimate how fast and how broadly things will change. In the trough, they underestimate what has quietly started working. An observation usually attributed to the futurist Roy Amara puts it neatly: we tend to overestimate the effect of a technology in the short run and underestimate it in the long run.
For practical decisions, the most useful question is not “is this hyped?” but “which specific use of this technology has already proved itself, for whom, and at what cost?” Even at the height of a hype wave there are usually a few such uses, and they tend to be less exciting than the headlines.
Recent hype waves: what was promised and what remained
Looking back at earlier waves shows a consistent gap between the promise at the peak and what eventually proved useful.
| Technology | Promised at the peak | What happened | What survived |
|---|---|---|---|
| The dot-com boom | Every business would move online at once; profits did not matter. | Many companies collapsed when the bubble burst around the turn of the millennium. | E-commerce, search and online advertising, which grew into the backbone of the digital economy on a slower, more uneven path. |
| Enterprise blockchain | Distributed ledgers would replace databases across banking, logistics and government. | Many corporate pilots were quietly shelved because a shared database did the job more cheaply. | Narrow uses where several parties distrust each other, plus the infrastructure behind digital assets and stablecoins. |
| The metaverse | People would work, shop and socialise in shared 3D worlds; companies bought virtual land. | Most virtual worlds stayed small and large investments were scaled back. | VR and AR for training, gaming, design and some industrial work, with hardware still improving. |
| Self-driving cars | Fully autonomous cars would be common within a few years. | Edge cases, safety and regulation proved far harder; several projects closed. | Driver-assistance features in ordinary cars and limited robotaxi services in a small number of cities. |
| Generative AI (ongoing) | AI will replace most office work and autonomous agents will run companies. | Very wide daily use for writing, coding and search, alongside errors, high costs and uneven returns. | Too early to say; drafting, summarising, coding help and customer support already show durable value. |
The lesson is not that hyped technologies fail. It is that the useful part is usually narrower, slower and more boring than the peak promised, and then larger than the trough assumed.
Signs a technology is near the peak of hype
Everyone is rebranding. Companies with no relevant product add the buzzword to their name, slogans or job titles.
Budgets arrive before use cases. Organisations announce strategies, labs or spending before they can describe a specific problem the technology will solve.
Demos outnumber deployments. Coverage is full of impressive prototypes, but you struggle to find customers using it daily for real work.
Courses and consultants multiply. A sudden wave of expensive courses warning that you will be left behind without them is a classic signal.
Timelines are vague but confident. Statements like “within a few years everything will change”, without saying who, how or at what cost.
Critics are treated as dinosaurs. Reasonable questions about cost, safety or demand are dismissed as failing to understand the future.
Speculative assets appear. Virtual land, tokens or shares tied to the buzzword trade at prices disconnected from any actual use.
Your relatives ask you about it. When the topic dominates family dinners and taxi conversations, expectations are usually running ahead of reality.
Where is AI on the curve, and what should you do about it?
Generative AI differs from the metaverse in one important way: very large numbers of people use it every day for real tasks such as drafting text, translating, summarising and writing code. That points to a durable core, whatever happens to the excitement around it.
At the same time, several popular claims show classic peak signs: that autonomous agents will soon run whole departments, that most jobs will vanish within a few years, or that any company without an AI strategy is doomed. Many organisations are also finding that pilots are easy while reliable, measurable returns are hard.
A practical approach for people and businesses in Vietnam
For individuals and small businesses in Vietnam, the practical approach is to separate the layers. Adopt now what is proven and cheap to try, such as AI for drafting, translation and summarising with a human checking the result. Experiment carefully with what is promising but uneven, such as agents that take actions on your behalf. Wait on anything that requires large upfront spending justified only by forecasts.
Hype cycles also create opportunities in the trough. When attention moves on, prices fall, tools stabilise, and the people who kept learning quietly have an advantage. Staying curious without being swept along is the most valuable stance through the whole curve.
Frequently asked questions
What is the Gartner hype cycle?
It is a widely used model describing five phases of public expectations around new technology: trigger, peak of inflated expectations, trough of disillusionment, slope of enlightenment and plateau of productivity. It describes a common pattern rather than predicting any specific technology.
Is AI a bubble like the metaverse?
Some expectations and valuations may be inflated, but unlike the metaverse, generative AI already has very wide daily use for real tasks. A cooling of the excitement would not erase that useful core.
Did blockchain fail?
The idea that blockchain would replace most databases did not hold up, and many corporate pilots ended. Narrower uses remain, especially where parties need a shared record without trusting a single operator, and in digital asset infrastructure.
When should a small business adopt a new technology?
When it solves a specific problem, there is evidence that similar businesses benefit, and a trial can be stopped cheaply. Being first is rarely worth a large upfront spend.
What is Amara’s law?
It is the observation, usually attributed to Roy Amara, that we overestimate a technology’s effect in the short run and underestimate it in the long run. It is a useful reminder to stay patient rather than either excited or dismissive.
