How did researchers avoid counting the same spending twice?

They tracked the dollar spent by an end customer as their primary figure. If a customer pays $1 to Anthropic and Anthropic pays Amazon 50 cents to serve it, they report the de-duplicated amount of $1 to avoid inflating the market size by counting the same revenue at multiple levels of the supply chain.

Why is this harder to measure than the supply side of AI?

Much AI revenue flows from private companies such as OpenAI, Anthropic, Cursor, and ElevenLabs that do not legally need to disclose financial details, while public hyperscalers like Amazon, Google, and Microsoft do not consistently disclose AI segment revenues. Researchers had to examine public statements, leaks, and self-reports to build a proprietary model.

What does demand elasticity tell us about the market as token prices fall?

The analysis shows that every 10% price cut leads to 12–18% more tokens in use, meaning total spending still rises even when per-token prices decline, suggesting lower prices expand the market rather than shrink revenue.

Back to articles

AI economy hits $110B in revenue, growing 3× faster than mobile or Internet waves

Hacker News23h ago5 min read

Key takeaway

Researchers have released the first comprehensive bottom-up measure of global AI spending, finding the industry generated $110 billion（約18兆円） in revenue over the past 12 months and is growing at a $175 billion（約28兆円） annualized run rate. This is roughly three times faster than previous IT waves like mobile or the Internet. The study reveals that companies are still in early scaling stages but intend to invest more heavily, and it shows revenues just about cover the capital costs required to build the necessary infrastructure.

Summaries like this, in your inbox every morning.

3 Key Points

What happened
Researchers released a bottom-up analysis of global AI spending, finding the ecosystem generated $110 billion（約18兆円） in revenue over the past 12 months and is growing at an annualized run rate exceeding $175 billion（約28兆円）. The study de-duplicates spending across the supply chain to avoid double-counting value flowing through intermediaries like Amazon, Google, and Microsoft.
Why it matters
These are the first comprehensive demand-side figures for AI spending across consumers and enterprises. Companies are still in early scaling stages, but executives across industries—from finance to pharma—say they intend to invest more heavily in AI going forward. The growth rate is roughly three times faster than previous IT waves like mobile or the Internet, suggesting AI is transforming business investment at an unprecedented pace.
What to watch
Whether revenues can sustain the infrastructure buildout—the analysis shows AI revenues from hyperscalers just about cover depreciation costs when compute assets are modeled over 6 years. The report also found that every 10% price cut in tokens leads to 12–18% more tokens in use, meaning lower prices may expand market size rather than shrink it.

FAQ

How did researchers avoid counting the same spending twice?: They tracked the dollar spent by an end customer as their primary figure. If a customer pays $1 to Anthropic and Anthropic pays Amazon 50 cents to serve it, they report the de-duplicated amount of $1 to avoid inflating the market size by counting the same revenue at multiple levels of the supply chain.
Why is this harder to measure than the supply side of AI?: Much AI revenue flows from private companies such as OpenAI, Anthropic, Cursor, and ElevenLabs that do not legally need to disclose financial details, while public hyperscalers like Amazon, Google, and Microsoft do not consistently disclose AI segment revenues. Researchers had to examine public statements, leaks, and self-reports to build a proprietary model.
What does demand elasticity tell us about the market as token prices fall?: The analysis shows that every 10% price cut leads to 12–18% more tokens in use, meaning total spending still rises even when per-token prices decline, suggesting lower prices expand the market rather than shrink revenue.

Discussion

No discussion yet for this article

Stay ahead with AI news

Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.

Get Started Free

Free · takes 30 seconds · unsubscribe anytime

1 minute a day. The AI essentials.

200+ sources · Email / LINE / Slack

Get it free →