These high marginal costs are a direct consequence of unprecedented levels of data center spending. In 2025, companies spent nearly $450B+ on data center spending to generate just $65B in AI software/services revenue.
This is almost certainly wrong. OpenAI themselves generates $20b and growing at 4x per year. Anthropic generates $9b.
Google, Amazon, Microsoft do not break down their AI-specific revenue. Many companies also bolt on AI services to their current services and product like Adobe, Salesforce, Figma, etc. There are also startups, small businesses that are generating meaningful AI revenue. Tons of AI apps advertising on iOS and Instagram so they must be making money too. There is no possible way to calculate these revenues.
From everything we know, every big tech can't keep up with demand.
You’re going to see us continue to be very aggressive investing in capacity because we see the demand. As fast as we’re adding capacity right now, we’re monetizing it.
- Amazon earnings call
Azure AI services revenue was generally in line with expectations, and this quarter, demand again exceeded supply across workloads, even as we brought more capacity online.
- Microsoft
Google and Meta are saying the same thing. AI products and services must be generating much more revenue than claimed by Gartner. Not everyone has billions of VC money to sell at a loss.
That's a publicly reported number and widely cited. It includes all revenue from the labs and hyperscalers. Menlo Ventures themselves reported that all enterprise AI spend in 2025 totaled $37B.
Yes, hyperscalers claim there's a lot of demand. A lot of that demand isn't actually priced, e.g. demand for Nano Banana.
Google, Amazon, Microsoft do not break down their AI-specific revenue. Many companies also bolt on AI services to their current services and product like Adobe, Salesforce, Figma, etc. There are also startups, small businesses that are generating meaningful AI revenue. Tons of AI apps advertising on iOS and Instagram so they must be making money too. There is no possible way to calculate these revenues.
From everything we know, every big tech can't keep up with demand.
- Amazon earnings call - MicrosoftGoogle and Meta are saying the same thing. AI products and services must be generating much more revenue than claimed by Gartner. Not everyone has billions of VC money to sell at a loss.
That's a publicly reported number and widely cited. It includes all revenue from the labs and hyperscalers. Menlo Ventures themselves reported that all enterprise AI spend in 2025 totaled $37B.
Yes, hyperscalers claim there's a lot of demand. A lot of that demand isn't actually priced, e.g. demand for Nano Banana.
Microsoft, Google, Amazon, and software companies like Adobe, Figma, Salesforce do not break down AI-only revenue.
How can they know?
How do they know how much revenue some viral AI app on iOS is generating?
All bubbles do not leave behind infrastructure. Some just create shortages, spike prices and clear the way for the next technology cycle.