Unverified50% confidenceOpinionExact time
AI accelerator cards face a 2–3 year technology iteration cycle, and their physical depreciation rate far exceeds that of optical fiber (physical lifespan exceeding 25 years), meaning the residual asset value of the AI infrastructure bubble will be lower than that of the fiber bubble.
1
Sources
50%
Confidence
Long-term
Relevance
7/19/2026
First Seen
Sources
Related Claims
UnverifiedThe return cycle on AI infrastructure investment may be as lengthy as that of fiber optics, and the market's preset assumptions about AI profitability timelines may carry a systemic optimism bias.74% similarVerifiedThe AI Big Crunch signals that the AI industry will undergo a rational process of bubble elimination and value return, rather than a failure of the technology itself.71% similarUnverifiedThe maturation speed of AI infrastructure, including declining computing costs, improving model capabilities, and maturing development tools, is faster than that of the cloud computing era.70% similarUnverifiedThe current capability iteration cycle for large AI models has shortened to 3-6 months, compared to Moore's Law's roughly 18-month cycle in traditional hardware68% similarUnverifiedSome researchers estimate that the knowledge half-life of applied AI technology is approximately 1 to 3 years, while the half-life of foundational mathematics and statistical theory is as long as several decades.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/556650API
curl https://kongchang.com/api/v1/knowledge/claims/556650MCP
get_claim(id=556650)