Behind millions of dollars of funding in AI sit enterprises with just a 5% average utilisation rate. Inference cost plus cost of ownership also rose to 41% from 34%
Summary
Enterprises that rushed to buy massive GPU fleets for AI now face low utilization rates (5%) and rising costs (inference cost plus cost of ownership rose to 41% from 34%), highlighting significant infrastructure inefficiencies in AI deployment.
Similar Articles
Why compute might get 10x more expensive in coming years (8 minute read)
The article analyzes factors that could drive AI compute costs up 10x in coming years, including rising lab revenue, margins, and spot prices, while suggesting that inference spending may signal stalled progress.
@rohanpaul_ai: Gartner forecasts that AI providers will spend nearly $1T on infrastructure in 2026 and another $6T from 2027 to 2030, …
Gartner forecasts AI providers will spend nearly $1T on infrastructure in 2026 and $6T more from 2027 to 2030, warning that over $3.5T of that could face write-downs as revenue fails to cover costs. Asset-light AI-native providers and exposed neoclouds are highlighted as key risks, with AI-optimized IaaS growth deemed too slow to recoup investments before current chips are replaced.
AI Is Too Expensive
The article argues that AI is too expensive to be economically viable for most companies, with hyperscalers spending trillions on data centers but failing to generate proportionate AI revenue. It suggests only hardware suppliers like NVIDIA benefit from the current AI bubble.
Running AI agents will cost 5x more by 2028
Gartner predicts that AI inference costs per agentic workflow will increase more than fivefold by 2028, driven by efficiency gains that enable more powerful models and applications, paradoxically raising overall costs.
Are inference providers able to make any margins?
The post discusses the low profit margins of AI inference providers due to high GPU costs and competitive pricing, suggesting the business model faces challenges despite market potential.