/* MINDFIRE BLOG */

Artificial intelligence is arguably the most hyped technology in decades. But hype isn’t the same as reality. In the debate over whether AI is in a bubble, there is both a pessimistic and a more optimistic view. Cory Doctorow warns that the economic foundations of AI are shaky and potentially disastrous, and Jeff Bezos, who agrees there’s a bubble — but insists it’s the good kind and heralds long-term benefits.

Let’s explore both sides with concrete examples.

The Pessimistic View: AI as “Funny Money” and Fragile Economics

Cory Doctorow’s article “The real (economic) AI apocalypse is nigh” argues that today’s AI bubble isn’t built on sound economics but on massive investor excitement and weak financial fundamentals. He paints a picture of an industry where capital flows in without sustainable revenue models or profit paths — classic signs of a bubble that could end badly.

Doctorow argues many AI firms “keep the lights on by soaking up hundreds of billions of dollars in other people’s money and then lighting it on fire.” He argues it’s a dependency on the continual infusion of new capital, hardly a sustainable business model.

Doctorow points out that, unlike past emerging technologies, each generation of AI technology costs more to build and operate than the last, with little evidence that those costs decline or that scaling will improve profitability.

He also points out that some AI data center companies are reportedly collateralizing loans with tens of thousands of GPUs, despite these chips losing value quickly — a bizarre financial setup that signals desperation rather than strength.

Even scarier is where Doctorow highlights practices where the same money is booked as an investment by one company and revenue by another — such as Microsoft “investing” in OpenAI by providing server access, then counting that as revenue. That’s not real earnings; it’s accounting sleight-of-hand.

Finally, he quotes a venture capitalist suggesting AI firms would need to sell hundreds of billions worth of services just to break even, a goal so large it’s hard to take seriously given current revenue levels. Is there really this much demand for AI in real life?

Doctorow's view is that this whole thing is just hype run out of control, leading to bad decisions—such as firing people over AI that can't do the job. In Doctorow's opinion, this is just the market gone mad.

The Optimistic Take: Jeff Bezos on the “Industrial Bubble”

But not everyone sees this as pure madness. Jeff Bezos — longtime tech leader and founder of Amazon — also acknowledges we’re in bubble territory, but his perspective is more nuanced.

Bezos said at Italian Tech Week 2025 that AI’s rapid growth and inflated valuations fit the classic pattern of a bubble, where “every experiment and every company gets funded — the good ideas and the bad ones alike.”

Bezos compared the AI bubble to past tech booms particularly the dot-com bubble. During the dot-com bust, Amazon’s stock famously plummeted from roughly $113 to around $6 a share during the early 2000s bust, even as its business fundamentals were improving. This shows how markets can wildly misprice companies in a bubble.

Bezos also points out that during the dot-com bubble everyone was chasing the Internet. Fiber-optic companies went bankrupt laying fiber, but the fiber they laid was bought out by other companies and is still around. The overall benefit to society was real.

This actually makes sense. Everyone can sense that AI is a transformative technology just like the Internet was. Back then everyone was trying to figure out how to use (and preferably dominate) the Internet because they knew there was massive profits on the line. So, from an investor's perspective, it made sense to chase these profits even if you risked losing everything. And for the companies that won this mad dash—Amazon, Google, Netflix, Facebook, etc—the rewards were well worth the risk.

Notice how this doesn't deny the problems Doctorow highlights, but it explains it differently. Instead of trying to explain it as market madness, it explains it as a correct understanding that huge profits are on the line. The problem is that no one knows at this point who the big winners are going to be.

This is what Bezos calls an industrial bubble: capital may flow to losers, but the underlying technology creates lasting value. Compare this to the bad kind of bubble, such as the 2008 housing bubble. That bubble was all bad because it really was just market madness with on actual paradigm-changing technology at its center.

Bezos points out that during hype cycles, investors often can’t distinguish good ideas from bad ones — meaning capital pours into projects that may never have viable products. But in the process, real innovation also gets funded.

Importantly, Bezos argues that just because valuations are frothy doesn’t mean the technology itself isn’t transformative. The dot-com bubble was a bubble, but the Internet really did transform everything. So, Bezos maintains that AI will change every industry and that the benefits to society will be “gigantic.”

So What Should We Make of It?

🚨 The Bubble Side

• AI has massive investment without clear returns.
• Financial engineering and accounting quirks mask true economics.
• If capital stops flowing, many companies might fail spectacularly.

🌟 The Breakthrough Side

• Innovation doesn’t happen without experimentation and funding, even if it’s wasteful at times.
• Past industrial bubbles (Internet) left real infrastructure and products behind.
• AI’s underlying technical progress isn’t a mirage — it’s real and pushing boundaries.

So I Should Be Afraid of the AI Bubble?

Surprisingly, no — though, you probably should be concerned. Take a look at this chart of the Shiller PE ratio and our three bubbles:

A Shiller PE is a price-to-earnings ratio, except smoothed out over 10 years. It's a great way to fundamentally value the market historically that smooths out all the local ups and downs. During the dot-com bubble the Shiller PE got up to 44 against a historical average of closer to 20. We are currently at 40. Even the 2008 housing bubble only got up to 28. So there is a lot of potential downside in this AI bubble. The stock market is clearly overvalued in large part due to the AI bubble.

Conclusion: Bubble or Breakthrough?

The answer might be both.

AI today shows many signs of a bubble: unsustainable valuations, weak unit economics, and overfunded startups with no path to profit. That’s worth worry — especially for investors and workers whose livelihoods depend on sound business fundamentals.

But if we look at history, bubbles aren’t uniformly bad. As Jeff Bezos emphasizes, when the dust settles, the infrastructure and breakthroughs left behind can help reshape industries for decades.

Whether you come down on the doom side or the long-term value side, one thing is clear: AI’s story is just beginning, and its economic impact will be debated for years to come.

In the meantime, one value that can be certain from AI is its use in projects and software that have an immediate purpose and measurable benefit or profitability. Mindfire Tech is committed to providing state-of-the-art software solutions and we believe AI can be a great boon to those who know how to use it properly. If you are interested in learning more about how AI could be used in your company or in your future projects, please don't hesitate to reach out for a free consultation.

Sources:
- Cory Doctorow, The real (economic) AI apocalypse is nigh — Medium. :contentReference[oaicite:18]{index=18}
- Jeff Bezos on the AI bubble, Italian Tech Week 2025 — various reporting. :contentReference[oaicite:19]{index=19}