Is AI a Bubble? Why I Think It’s Bigger

Every time I mention AI stocks, someone tells me it’s a bubble.

I understand the skepticism. Valuations are high. The hype is real. History is full of technology bubbles that ended badly for late investors. The dot-com crash is still fresh enough in collective memory to make people nervous when they hear about trillion-dollar valuations and exponential revenue growth.

But here’s my honest take: I think AI is bigger than the Industrial Revolution. Not the AI stocks — some of those are probably overvalued. The technology itself. Here’s why.

What the Industrial Revolution Actually Did

The Industrial Revolution, which began in Britain in the late 18th century and spread globally through the 19th, was fundamentally about replacing human muscle with machines.

The steam engine could do the work of dozens of horses. The power loom could weave fabric faster than any human hand. The railroad could move goods and people at speeds previously unimaginable. Factories concentrated production in ways that multiplied output per worker by orders of magnitude.

The result was the most dramatic improvement in human living standards in history. Life expectancy rose. Poverty fell. Cities grew. The global economy expanded in ways that would have seemed impossible to anyone living in 1750.

But the Industrial Revolution only replaced one thing: physical labor. It gave humans mechanical leverage over the physical world. The thinking still had to be done by people.

What AI Is Actually Doing

AI is replacing something different: cognitive labor. The thinking, analyzing, writing, diagnosing, designing, coding, translating — the work that has historically required human minds.

And when you combine AI with robotics — physical machines guided by artificial intelligence — you get something that has never existed before in human history: a system that can replace both physical and cognitive labor simultaneously.

A factory robot from 2010 could weld the same seam in the same position, perfectly, ten thousand times. But it couldn’t adapt. It couldn’t notice a problem and adjust. It couldn’t figure out a better approach.

An AI-guided robot in 2026 can do all of those things. Tesla’s Optimus robot is already performing tasks in Tesla’s factories — not pre-programmed movements, but adaptive responses to real-world conditions. Boston Dynamics’ Atlas can navigate complex environments. Figure AI’s robots are working in BMW plants.

The combination — intelligence plus physical capability — is qualitatively different from anything the Industrial Revolution produced. It’s not just faster or stronger. It’s capable of learning, adapting, and improving in ways that no mechanical system could.

Why “It’s a Bubble” Gets the Analysis Wrong

The bubble argument usually goes like this: AI valuations are high, the hype is unsustainable, and when the hype fades, the stocks will crash like dot-com stocks did in 2000.

There are two problems with this argument.

First, it conflates the technology with the stock prices. The dot-com bubble was a bubble in stock prices, not in the internet itself. The internet was real and transformative — it just took longer than investors in 1999 expected, and many of the specific companies that were overvalued went bankrupt. But Amazon, Google, and others survived and eventually justified extraordinary valuations. The technology was right. Most of the stock prices were wrong.

The same distinction applies to AI. Some AI company valuations are probably stretched. The technology itself is not a bubble — it’s a genuine capability revolution with real, measurable effects on productivity, revenue, and economic output.

Second, the numbers are real in a way that dot-com numbers weren’t. In 1999, many high-flying internet companies had minimal revenue and no profits. They were valued on speculation about future potential.

NVIDIA generated $60 billion in profit last year. Broadcom reported $10.8 billion in AI chip revenue in a single quarter, up 143% year over year. Dell reported $16.1 billion in AI server revenue in one quarter, up 757% year over year. These are extraordinary numbers, but they’re real revenue from real customers making real capital expenditures.

When hyperscalers — Microsoft, Google, Amazon, Meta — are each spending tens of billions of dollars per year on AI infrastructure, that’s not speculation. That’s the largest coordinated capital investment in history, made by the most sophisticated technology companies in the world, betting their competitive positions on AI being transformative.

The Scale of What’s Coming

I want to be careful not to sound like someone selling AI stocks. I’m not. I’m a regular dad trying to understand the world well enough to make reasonable financial decisions. And here’s what I actually believe:

The penetration of AI into the global economy is still in the very early stages.

Consider what’s already happening. Radiologists are using AI to detect cancers more accurately than human specialists alone. Law firms are using AI to review contracts in minutes that previously took weeks. Software engineers are writing code significantly faster with AI assistance. Customer service operations are handling millions of interactions with AI agents.

Now consider what hasn’t happened yet. Most factories still don’t have AI-guided robotics. Most hospitals haven’t integrated AI diagnostics into routine care. Most schools haven’t transformed how they deliver education. Most construction, agriculture, transportation, and logistics operations are still largely pre-AI.

The sectors that have been transformed so far — data centers, software development, certain knowledge work — represent a fraction of the global economy. The sectors that haven’t been transformed yet are much larger.

AI + Robots: The Combination That Changes Everything

If AI alone is transformative, AI combined with robotics is something else entirely.

There are currently about 8 billion humans on Earth. The global workforce — the people doing economic work — is roughly 3.5 billion. Every one of those people has a finite amount of energy, requires sleep, gets sick, ages, and eventually dies.

AI-guided robots don’t have any of those constraints. They work 24 hours a day. They don’t get tired. They improve with software updates. They can be manufactured at scale.

Elon Musk has said he expects Tesla to eventually produce more Optimus robots than cars — potentially millions per year. He’s described it as potentially the most valuable product Tesla makes. Whether his specific predictions prove accurate, the direction is clear: the era of AI-guided physical robots operating in the real economy is beginning, not approaching.

The economic implications are difficult to fully comprehend. Every analysis of AI’s economic impact so far has been based primarily on AI as software — AI that assists knowledge workers and automates cognitive tasks. The addition of physical capability — robots that can actually do physical work in the real world — multiplies the potential impact by an order of magnitude.

What Could Go Wrong

I want to be honest about the risks, because anyone who tells you this transition is guaranteed or smooth is selling something.

The timeline could be much longer than expected. Technology transitions almost always take longer than optimists predict and shorter than skeptics predict. The practical deployment of AI and robotics across the global economy will face regulatory, social, infrastructure, and technical barriers that aren’t fully visible yet.

The distributional effects could be severe. The Industrial Revolution improved aggregate human welfare over time, but the transition was brutal for many people — displaced workers, disrupted communities, enormous social upheaval. AI’s displacement of cognitive and physical labor could create similar disruptions at enormous scale, with no guarantee that the benefits will be widely shared.

Specific companies and investments could still disappoint. Even if the overall AI transformation unfolds as I expect, identifying which companies will capture most of the value — and which will be displaced by better competitors — is genuinely difficult. The history of technology investing is full of people who correctly identified transformative technologies but lost money because they backed the wrong companies.

My Investment Approach Given This View

Given my belief that AI is a generational transformation rather than a speculative bubble, you might expect me to be heavily concentrated in AI stocks.

I’m not. Here’s why.

I believe the transformation is real. I’m much less certain about which specific companies will capture most of the value, and at what price that value is already reflected in current stock prices.

My approach is to own the transformation through broad S&P 500 index funds — which automatically include NVIDIA, Microsoft, Broadcom, Apple, Amazon, and every other major AI infrastructure company as they grow to become significant parts of the economy. As AI companies grow, their weight in the index grows, and my exposure grows with it. Without having to pick winners.

That’s not the highest-conviction approach. Someone who correctly identified NVIDIA at $100 and held it to $1,000 did far better than any index fund. But that requires being right about the specific company, the specific timing, and having the conviction to hold through the inevitable corrections. Most investors — including me — are better served by the index.

The AI transformation is real. The Industrial Revolution comparison is apt. The investment approach that captures it most reliably for most people is simpler than it sounds: own the market, stay invested, let time do its work.

Related: What is NVIDIA? covers the company most directly powering the AI buildout. And How to Start Investing With $100 explains the practical first steps for anyone who wants to participate in this transformation through the stock market.