Skip to content

Is AI just a bubble – or are the numbers telling us something else?

I keep seeing two stories in my feed: 1️⃣ We’re in an “everything bubble” fuelled by cheap money and debt. 2️⃣ AI is the new internet – huge adoption and revenue, not just hype.

3 min read

Originally posted on LinkedIn · 8 reactions · 3 comments · View original →

  • Is AI just a bubble – or are the numbers telling us something else?

I keep seeing two stories in my feed: 1️⃣ We’re in an “everything bubble” fuelled by cheap money and debt. 2️⃣ AI is the new internet – huge adoption and revenue, not just hype.

They’re really answering different questions:

  • “Is AI itself fake or overhyped?”
  • “Is the whole financial system over-leveraged?” Both have a piece of the truth.
  1. What each narrative is really saying

🧩 AI-is-real story

This camp says there are (at least) two kinds of bubbles:

  • Credit/speculation bubbles (1929, 2008): lots of leverage, little lasting value.
  • Infrastructure bubbles (rail, fibre, internet): prices overshoot, but the infrastructure powers decades of growth.

On this view, AI looks like the infrastructure type:

  • Trillions going into data centres, chips and power – hard assets, not just paper
  • Real demand: hundreds of millions of people using AI tools

Model companies with multi-billion revenue run-rates:

  • ChatGPT has scaled to ~800M weekly active users in under three years, and OpenAI is talking about $10–13B in annualised revenue.
  • Anthropic went from roughly $1B to over $5B run-rate in about eight months and is guiding towards ~$9B ARR, mostly enterprise.

Hype can inflate a valuation, but hype alone doesn’t create tens of billions in recurring revenue and that level of usage.

Conclusion: valuations may be frothy, but the tech, infra and usage are very real.

💣 Doomer / everything-bubble story

This side says:

  • Years of cheap money and stimulus have pulled all asset prices up
  • Debt and interest costs are rising fast
  • Central banks are stuck: raise rates and you stress the system; cut rates and you inflate bubbles further
  • When confidence snaps, everything gets repriced at once – AI included

Short version: AI-is-real = sector fundamentals Doomer = macro plumbing

  1. Who’s “right”?

I don’t think the macro doom story actually debunks AI.

It mostly says: “Even if AI is real, prices can still be too high because the whole system runs on cheap debt. When that breaks, AI gets hit too.”

The AI-is-real side says: “Yes, but demand, infra and revenue are already here. This looks more like the internet build-out than the 2008 subprime.”

Both can be true:

  • We can have transformative technology
  • And over-extended financial conditions The doom story mostly attacks price and funding, not whether AI creates value.
  1. My view

If I have to compress it to one line: We’re in an AI investment bubble in parts of the market, but not an AI reality bubble.

--> On a 10–20 year view, AI is closer to electrification or the internet.

--> On a 1–5 year view, there are bubble-like pockets: concentrated gains, aggressive capex, “must own AI” FOMO.

So I’ve stopped trying to pick a camp. I’m holding both truths at once: respect the macro risk, and take the AI demand and revenue seriously.