Major tech companies like Meta, Amazon, Microsoft, and Alphabet are pouring billions into artificial intelligence hardware, but a new report reveals a costly paradox: their investments are rendered obsolete within three years. According to research from Research Affiliates, the hyperscalers' AI equipment loses economic value far faster than its physical lifespan, creating financial strain for these tech giants.
The AI Arms Race and Hardware Turnover
Chris Brightman, CEO of Research Affiliates, compares the AI hardware market to a supermarket with rapid turnover. 'They’re more like supermarkets than traditional tech or industrial enterprises,' Brightman explains. 'Their turnover isn’t in the likes of grocery items. It’s the stuff that generates their large language models, vector search, and other products.'
'The economic life of AI hardware is a lot shorter than its accounting life,' Brightman writes. 'Right now, each is using AI to maintain crucial dominance in their field, but the immense spending needed to maintain those moats could generate puny returns going forward.'
Energy Constraints and Innovation Drive Obsolescence
The primary driver of this rapid obsolescence is the relentless pace of innovation by companies like Nvidia and AMD. Each year, new hardware offers significant increases in computing power per watt, forcing hyperscalers to continually upgrade their equipment to stay competitive. This cycle is exacerbated by energy constraints, as data centers seek to maximize efficiency while managing skyrocketing electricity costs.
Historically, industries like steel and railroads relied on infrastructure with multi-decade lifespans. In contrast, AI hardware depreciates fully within three years, as its revenue-generating potential no longer covers acquisition, operating, and capital costs.
Financial Implications for Big Tech
The financial toll of this rapid turnover is staggering. Nvidia’s H100 GPUs, for example, deliver a 137% return on investment in their second year but suffer a 34% loss by year four. With AI capital expenditures projected to reach $650 billion this year—equal to 2% of GDP—the impact on profitability could be severe. As Brightman notes, this relentless spending cycle threatens the long-term financial health of Big Tech companies heavily invested in AI.