What Happens When The Bubble Pops

We've covered the economics. The circular money. The AGI gamble. The threat from cheaper local models.

Now the question everyone's asking: What happens when the bubble pops?

There are two scenarios. In both cases, neither are pretty.

Scenario 1: The Loud Pop

OpenAI and other major AI companies can't find a path to profitability. Investors lose patience. Funding dries up.

Suddenly, OpenAI is in trouble.

And remember that "if we fail, you fail" strategy we talked about?

Microsoft has billions tied to OpenAI. NVIDIA's record sales depend on AI companies buying chips. AMD, Oracle, and others are all interconnected.

If OpenAI goes down, it doesn't go down alone.

Imagine Microsoft, NVIDIA, and AMD stock all taking major hits at basically the exact same time.

These aren't small companies. They're probably in your retirement fund. Your 401k. Your index funds.

The ripple effects could last years.

Scenario 2: The Slow Deflation

The companies find a way to cut losses. Scale back. Find some equilibrium between cost and revenue.

The bubble doesn't pop. It just deflates.

Less dramatic for sure. But still significant.

For us, this means investment slows. Hiring freezes. Layoffs. AI-focused startups slowly shut down.

AI doesn't go away. But the gold rush is over.

What Happened With Dot-Com

The Nasdaq dropped 80% over two years.

Not in one day. Two years of slow, grinding decline.

60% of companies were gone. 200,000 people laid off. $5 to $7 trillion in value just evaporated.

It wasn't instant. It was a slow bleed. Companies failing one by one. Funding rounds that never came. Acquisitions at pennies on the dollar.

Which Scenario Is More Likely?

Honestly? Probably somewhere in between.

Some companies will fail and make major headlines. Others will just fizzle out quietly.

The biggest companies will survive but at reduced valuations.

The market will consolidate. Just like streaming. Just like dot-com.

A few winners. Lots of losers. And everyone who built on the losing platforms scrambling to migrate.

The Timeline

Could be months. Could be years.

The big players have deep pockets. Microsoft, Google, Meta can burn cash longer than most.

But the math doesn't change. At some point, the bill comes due.

What I'm Watching For

The first sign will be pricing changes. When companies start raising API costs significantly, that's when you know the subsidies are ending.

Then layoffs. OpenAI, Anthropic, and others cutting staff to extend runway.

Then consolidation. Smaller players getting acquired or shutting down. Features getting cut. Services getting merged.

It won't be one dramatic moment. It'll be a series of adjustments as reality catches up with hype.

What This Means For Developers

If you're building systems that depend on AI APIs, you need contingency plans.

What happens if your primary API provider raises prices 5x? Can your business model handle that?

What happens if they shut down? How quickly can you migrate to an alternative?

What happens if the features you depend on get deprecated during consolidation?

These aren't hypothetical concerns anymore. This is what happens in every bubble.

What This Means For Businesses

If you're considering major AI investments, understand the risk.

The technology works. But the companies providing it might not survive in their current form.

Build systems that aren't dependent on a single provider. Have migration paths. Don't bet your entire operation on a company burning billions annually with no path to profitability.

The Part That Actually Worries Me

It's not that AI will go away. The technology is real and useful.

What worries me is the disruption during consolidation.

Features breaking. APIs changing. Prices spiking. Companies shutting down.

If you've built critical business systems on these platforms, that disruption is expensive and painful.

And unlike dot-com where websites were relatively easy to migrate, AI systems are deeply integrated. Moving from one provider to another isn't trivial.

What I'm Doing

I'm building with AI but not depending on it exclusively.

For critical client systems, I'm using AI to enhance processes, not replace them entirely. If the AI fails or gets too expensive, the core process still works.

I'm also testing local models more aggressively. Building contingency plans. Making sure I'm not locked into any single provider.

That's the smart approach in a bubble. Use the tools while they're cheap. But don't build your foundation on sand.

The Bottom Line

The AI bubble is going to pop. The only question is how and when.

Some companies will survive. Most won't. The survivors will own the consolidated market.

If you're building with these tools, plan accordingly. Use them. Get value from them. But don't bet everything on companies that are burning billions with no clear path to profitability.

The music is going to stop eventually. Make sure you're not the one left without a chair.

Next up in Part 7: The replaceable developer problem. What this means for people who build with AI every day.