3.5 Years After ChatGPT Took Off, Is the AI Bubble Really Nearing Its End?

According to Gelonghui, an economist argues that the current AI boom, measured from ChatGPT's debut in November 2022, has now lasted about 3 and a half years. The same article makes a blunt call: the AI bubble will 100% burst, and it will most likely end in a very painful way. That sounds severe, but the piece frames it less as a radical prediction than as a familiar stage in a major technology cycle.

The figures cited in the report explain why that view feels plausible. The AI frenzy has pushed Nvidia to roughly $5 trillion in market value, lifted multiple AI companies to trillion-dollar valuations, and even rising memory prices have been tied to AI demand. Once one theme is driving chips, computing power, large models, and upstream and downstream hardware at the same time, capital is no longer pricing businesses on current earnings alone. It is paying in advance for a future narrative.

The scale of spending may be the most striking part. The article says a small group of US tech giants, including Google, Microsoft, and Amazon, are already investing $700 billion to $800 billion a year in AI. It adds that this spending could account for 75% of US GDP growth, calling it an event on the level of America's national trajectory. For players, the knock-on effects are not hard to recognize: over the past 2 years, GPU prices, server costs, and cloud computing discussions have all been touched by AI to some degree.

The economist's core point is not that AI lacks value. It is that a "great technology" often arrives with a "great bubble." The article explicitly compares the current AI cycle to what followed the 2000 internet bubble: first valuations detach from reality, then the shakeout comes, and only after that do the companies with viable business models remain standing. In that reading, a bubble bursting is part of the process, not the end of the story.

What makes this phase risky is how concentrated global capital has become around the same idea. Across stock markets in the United States, Japan, South Korea, Europe, and China, many of the hottest names are tied to AI chips, large models, computing power, and related hardware. If a business cannot tell an AI story at all, it may find investors less interested from the start, with financing and valuation both marked down. The problem is not necessarily that the market has chosen the wrong direction. It is that everyone is using the same direction to price assets.

From a games-media perspective, this reads less like an emotional warning and more like an industry consumer alert. In the short term, companies on the AI chain may not be suitable for normal value-for-money logic; volatility could stay high, and pullbacks could be brutal. Over the long run, the names that matter most will still be the ones that turn technology into stable products and real demand. Put simply, the collapse of an AI bubble would not necessarily mean the AI revolution has failed. It would look more like a massive reset before the next version goes live, with the survivors deciding whether this cycle ultimately justified its cost.