With the advent of AI most view that cloud spending will increase. However AI might also optimize existing system (including itself) and there is heavy research into reducing the cost (in terms of compute and energy) of existing AI systems.

On the other side, even if the spending is reduced it could cause a Jevons Paradox in terms of more compute being used but will that compute be local or cloud-based?

This market will resolve Yes if services like AWS, Azure, GCP, etc continue to collectively grow or even stay at the same level of spending according to sources.

This market will resolve No if services like AWS, Azure, GCP, etc start to shrink in terms of revenue/net sales according to sources. Specifically they have to collectively be shrinking for three quarters in a row. This is a imperfect measure because a market down turn could also trigger this event but will be used for this market.

The key cloud cloud measures will be AWS, Azure, GCP quarterly earnings reports. I might add Cloudflare or other cloud providers later.

Resolution criteria

This market resolves YES if AWS, Azure, and GCP collectively maintain or grow their combined revenue/net sales through end of 2027. This market resolves NO if these three providers' combined revenue shrinks for three consecutive quarters at any point before the end of 2027.

Resolution will be determined by official quarterly earnings reports from Amazon, Microsoft, and Google. For AWS, use Amazon's quarterly earnings disclosures. For Azure, use Microsoft's Intelligent Cloud segment revenue. For GCP, use Google Cloud revenue.

Background

Global enterprise spending on cloud infrastructure has grown from $58 billion in Q3 2022 to $99 billion in Q2 2025, driven significantly by AI adoption. AWS remains the revenue leader at ~$126.6 billion annualized, but its growth cadence is ~18%, well below Microsoft Azure and Google Cloud Platform, both running high-30s. Enterprise cloud costs rose an average of 30% in the last year, with spending on AI applications and generative AI cited as top drivers for growing cloud spend by half of respondents.

However, efficiency improvements in AI are emerging. DeepSeek's V3 model achieved substantial improvements in training and reasoning efficiency, reducing training costs by approximately 18 times and inferencing costs by about 36 times, compared with GPT-4o. Efficiency gains may not substantially impact overall compute power demand over the long term, aligning with the concept of Jevons Paradox, which posits that improvements in efficiency can lead to increased overall demand.