Summary of the Core Content
This is a cautionary tale written in a science fiction narrative style, set in the year 2050, reflecting on past events. It imagines that twenty to thirty years from now, another wave of AI innovation (specifically, brain-inspired intelligence) has emerged. A group of people visits the ruins of computing facilities left behind after the previous AI bubble burst, examining the fates of four typical failed intelligent computing centers. The story starkly reveals the recurring cycle that humanity always follows when faced with new technological trends: first, there is an exuberant belief that this time will be different; then a rush to invest; followed by widespread failure and the creation of new buzzwords to start the cycle all over again. The message is a reminder to all local governments, investors, and ordinary participants in the AI industry today: history does not repeat itself exactly, but the same patterns persist.
---
Simplified Explanation of Key Points
1. The Failure of Intelligent Computing Centers
The article describes how, between 2024 and 2026, more than 30 cities across the country competed to build intelligent computing centers. However, 90% of the computing power remained unused from the day they were activated. This is exactly the same logic as the rush to build industrial parks and cultural tourism facilities in the past, which resulted in ghost cities with no inhabitants. Local governments wanted to capitalize on the trend and gain policy benefits without truly understanding the local demand for AI services. They followed the crowd, sometimes even using refurbished, low-end graphics cards to inflate the reported computing capacity, just as some developers cut corners when building buildings. When the bubble burst, speculators who bought the cards sold them and disappeared, leaving behind server manufacturers and operators with unpaid debts. The empty data centers stood for decades in the wilderness, with slogans like “Computing power empowering all industries” still on the walls, a reminder of the former frenzy.
2. The High Cost of Failure
The only way to “stop the loss gracefully” for these failed centers was to convert them into “digital economy housing.” Government departments, universities, and public hospitals, which were less likely to flee, were forced to buy the facilities at extremely low prices, not even getting enough funding to upgrade their hardware. This is similar to how the government bought back unsold real estate during the housing crisis and converted it into affordable housing. The quick profiteers from the boom had already taken their profits, and if the gaps were not filled, it would either lead to the loss of state assets or cause instability due to supplier complaints. In the end, it was the public’s money that had to cover the losses.
3. The Destruction of Old Data Centers
In the more developed regions, people were willing to spend money to demolish old data centers that were not yet completely obsolete. This was not just a waste of resources; it was also about freeing up valuable land and energy consumption quotas. During the AI boom, these old centers, which did not meet energy efficiency standards, were holding onto these quotas and preventing new development. Demolishing them allowed for the construction of new brain-inspired intelligence centers, generating more profits. In other words, urban development priorities also shift with the times—quotas allocated to large models in the past are now reallocated to new technologies.
4. The AI Arms Race
The article mentions a company named CloseAI, which once planned to invest $100 billion in the world’s largest AI park. But once the bubble burst, it abandoned all its long-term plans and focused on stable, profitable businesses such as search, advertising, and cloud services. It sold off its excess computing power to competitors. The survivors in the AI industry were those with strong cash flow, not necessarily the most technologically advanced companies. The so-called “computing power changing the world” turned out to be a hot topic in a bear market; giants with sufficient cash could buy up underutilized resources and wait for the next wave of demand to rise in value.
5. The Real Purpose of Technological Bubbles
All technological bubbles, no matter how different they seem, are essentially about cashing in on anticipated future demand. Investors in 2050 claim that brain-inspired intelligence will solve the energy problems of large models and that AGI (Artificial General Intelligence) is guaranteed to succeed. This is similar to the hype around the internet in 2000, when everyone believed it would transform all industries. The reality is that people are betting on future needs that may not materialize for decades. The resulting waste can only be borne by future generations. The officials and investors from 2050 are essentially the same as those from 2025 and 2000, just with different labels and new narratives.