Will AI’s Overbuild Become a Public Backbone—or Private Scrap?

Unlike the dotcom era’s open, long-lived infrastructure, today’s AI buildout is proprietary and short-lived, centered on vendor-tied GPUs and specialized data centers. If the boom stalls, we risk being left with hardware and facilities that are hard to repurpose. Some benefits could still emerge—lower prices, second-hand markets, and grid/network upgrades—but realizing broad gains requires opening the ecosystem.

Key Points

• The dotcom overbuild created open, durable infrastructure (fiber and open protocols) that powered decades of innovation.
• Today’s AI investment is largely proprietary and vertically integrated, centered on short-lived GPUs and specialized AI data centers tied to a few vendors.
• If demand falters, the industry could be left with hard-to-repurpose hardware and facilities—“silent cathedrals of compute.”
• A surplus could still lower costs, spur experimentation, create a second-hand GPU market, and leave useful power/network upgrades and lasting operational expertise.
• Openness is pivotal: without shared standards and interoperability, any overbuild will remain a private surplus rather than a broadly beneficial public good.

Sentiment

The community is predominantly skeptical of the AI investment boom and broadly supportive of the article's cautionary thesis. While many acknowledge that LLMs are useful tools with real applications, the dominant sentiment is that the scale of spending is unjustified by current capabilities, the proprietary nature of the infrastructure makes the dotcom comparison unfavorable for AI, and the hype echoes previous technology bubbles. A vocal minority of AI enthusiasts pushes back with arguments about generalized learning algorithms and transformative potential, but they are consistently outnumbered and challenged to provide concrete evidence.

In Agreement

• Unlike fiber optics and open internet protocols, AI GPUs are proprietary, vendor-locked, and short-lived, making them poor candidates for durable shared infrastructure.
• Purpose-built AI data centers are designed around specific GPU configurations and cannot be easily repurposed for general computing, unlike the internet backbone.
• The AI buildout lacks open standards and interoperability, meaning any surplus remains a private asset controlled by a few corporations rather than a public platform.
• The current AI investment scale dwarfs the dotcom era, amplifying the potential economic fallout when financial reality catches up to hype.
• The 1980s expert systems era shows a precedent where AI hype collapsed and what survived was simply relabeled as automation, not as revolutionary intelligence.

Opposed

• Like cell phone technology, each generation of AI hardware becomes obsolete but cumulative investment drives the technology toward being plentiful and cheap for everyone.
• Even if GPUs are not repurposable, the surrounding infrastructure — power grids, networking, cooling, physical buildings — has lasting value and can be reused.
• LLMs already deliver real practical value: compressing months-long policy analysis into weeks, enabling rapid prototyping through vibecoding, and serving as always-available research assistants.
• AI is fundamentally different from crypto because it has demonstrable everyday utility — people actually use ChatGPT and Claude for productive work, unlike blockchain which remained hypothetical.
• A financial bubble and genuinely transformative technology are not mutually exclusive — the internet was world-changing and the dotcom bubble still happened.