The Alchemy of Anxiety

Recently everyone has been obsessed with AI demos.

ChatGPT writing poems.
AI girlfriends.
AI agents booking restaurants.
Some startup founder in San Francisco claiming his product will “replace humanity.”

But honestly, if you actually look at where the money is going, you realize very quickly this is no longer some cute software revolution.

This is a construction project.

A gigantic one.

The people making the real money right now are not sitting in coffee shops talking about prompts.

They’re building data centers next to power plants.

They’re fighting over transformers.

They’re buying land near cheap electricity.

They’re trying to secure H100 shipments like it’s black market military equipment.

That’s the part most people still don’t understand.

AI is starting to look less like Silicon Valley software culture and more like Cold War industrial policy.

And the scary part is:

I don’t think this thing is even close to peaking yet.

---

The atmosphere right now honestly reminds me of 2020 when people were panic-buying toilet paper.

Except now the panic-buying is happening at the corporate level.

Meta is spending tens of billions on AI infrastructure.

Microsoft keeps scaling compute.

Amazon is terrified of falling behind.

Google suddenly looks like a company trying to defend its throne while pretending everything is under control.

Nobody wants to be the company that missed the next platform shift.

And once fear enters the equation, discipline usually disappears.

That’s when capital cycles stop behaving rationally.

The funny thing is that Wall Street still tries to explain all this with neat little PowerPoint slides.

“AI monetization.”

“Long-term productivity gains.”

“Digital transformation.”

Come on.

A lot of this is just pure fear.

Nobody actually knows what the AI economy looks like ten years from now.

But everybody knows one thing:

if your competitors build the infrastructure first, you may never catch up again.

So the spending continues.

And the numbers are getting absurd.

---

People keep calling Jensen Huang an “arms dealer.”

I actually think that description is too flattering.

He’s closer to a landlord.

NVIDIA built a toll booth around the AI economy and now the entire tech industry has to drive through it.

CUDA is the real trap.

That’s the genius of it.

The chips matter, sure. But the ecosystem matters more.

Every CEO gets on stage talking about how AI will transform the future, and then quietly sends billions of dollars straight into NVIDIA’s bank account.

Everyone else is working for the AI economy.

Jensen is collecting rent.

And honestly, the market still underestimates how difficult it is to escape NVIDIA’s grip.

People outside semiconductors always think these things are easy to copy.

They’re not.

This industry is brutal.

One tiny mistake destroys billions.

Yield problems. Packaging bottlenecks. Thermal issues. Supply chain delays. Networking constraints.

The deeper you go into semiconductors, the uglier it gets.

That’s why I laugh a little whenever people casually say:

“Competition will catch up soon.”

Maybe eventually.

But the real world moves slower than Twitter thinks.

---

What’s happening around TSMC is even crazier.

At this point Taiwan isn’t just important economically.

It has become one of the world’s most valuable strategic chips on the geopolitical poker table.

And everybody knows it.

That’s why governments suddenly care about fabs now.

A few years ago, nobody outside the industry cared about advanced packaging or EUV machines.

Now politicians talk about semiconductors like they’re talking about oil reserves during the 1970s.

Because deep down, the world realized something uncomfortable:

modern civilization is weirdly fragile.

Way more fragile than people thought.

You remove enough advanced chips from the system and suddenly:

• cloud infrastructure slows down
• military systems get affected
• AI scaling hits a wall
• industrial automation stalls
• entire supply chains start choking

People used to think semiconductors were just another tech sector.

Now they’re starting to realize semiconductors are closer to oxygen.

And once governments realize something is strategically essential, markets stop behaving normally.

That’s when the real distortions begin.

---

I also think people are getting a little drunk again.

You can feel it.

Every company suddenly claims to be an AI company.

Every VC deck now has glowing neural network graphics.

Every mediocre SaaS business adds “AI-powered” somewhere on the homepage and hopes investors get excited.

Honestly, some of it is complete nonsense.

But bubbles are weird like that.

The dangerous thing about bubbles is that they usually form around something real.

Railroads were real.

The internet was real.

Smartphones were real.

And AI is probably real too.

That’s why these cycles become so psychologically dangerous.

Because even if the long-term story is correct, you can still get financially destroyed halfway through.

I lived through enough market cycles to know how this usually goes.

At first everybody feels skeptical.

Then prices go higher.

Then people start feeling smart.

Then leverage appears.

Then discipline disappears.

And eventually you reach the phase where nobody even cares about valuation anymore.

We’re not fully there yet.

But we’re definitely moving in that direction.

You can already feel the temperature changing.

Honestly, walking through parts of San Francisco lately reminds me a little too much of old dot-com stories.

Everyone suddenly has an AI startup.

Everyone suddenly has “revolutionary infrastructure.”

Everyone suddenly believes exponential growth will solve every business model problem.

That usually ends badly for somebody.

The problem is:

nobody knows who dies until after the cycle breaks.

---

And here’s the part almost nobody talks about yet:

electricity.

AI isn’t just eating semiconductors.

It’s eating power.

Massive amounts of it.

At some point the bottleneck may stop being GPUs and start becoming energy itself.

That’s why data centers are quietly turning into strategic infrastructure.

That’s why nuclear energy is suddenly becoming fashionable again.

That’s why power grids matter again.

People still think AI is mostly about software engineers writing code.

I don’t think that’s true anymore.

This thing is becoming physical.

Concrete. Steel. Cooling systems. Transmission lines. Industrial power management.

The AI boom is slowly turning into an industrial boom.

And industrial booms can get messy fast.

Especially once governments, geopolitics, and national security start mixing into the trade.

---

The thing that frustrates me the most is that retail investors still spend way too much time staring at meaningless noise.

Fed headlines.

Daily candles.

Twitter arguments.

Some random influencer predicting Bitcoin at $1 million.

Most people are staring at the scoreboard while completely ignoring the machinery underneath the stadium.

The real story is that the world is entering another giant global spending cycle.

Money is flowing into:

• AI infrastructure
• semiconductors
• energy systems
• defense-adjacent technology
• sovereign compute
• industrial capacity

And once governments start participating in a cycle, things can stay irrational far longer than people expect.

That’s the dangerous part.

People keep assuming this thing will cool down naturally.

Maybe.

But huge infrastructure cycles rarely end quietly.

Especially when national pride, technological dominance, and trillions of dollars get mixed together.

That’s when markets become dangerous.

And addictive.

Very addictive.