Is the AI Bubble Like the Dot-Com Bubble? A Balanced Comparison

Mar 18, 2026

The Problem: I See the Same Pattern Again

I’ve been watching the AI investment frenzy for the past two years. Billions flooding into startups with minimal revenue. Headlines screaming about AGI within months. Engineers commanding million-dollar packages. And I keep thinking: I’ve seen this movie before.

In 1999, I watched the dot-com bubble inflate. Pets.com raised $82.5 million and collapsed in 9 months. Webvan spent $1 billion on infrastructure before making a profit. The Nasdaq peaked at 5,048 in March 2000, then lost 78% by October 2002.

Today, OpenAI is valued at $80+ billion. NVIDIA briefly hit a $3 trillion market cap on AI chip demand. Hundreds of AI startups launch monthly with no clear path to profitability.

The parallels are striking. But I also see differences that matter. Let me break down what’s similar, what’s different, and what this means for investors and professionals.

What Makes a Tech Bubble?

A tech bubble forms when four conditions align:

1. Speculative investment → Funding outpaces actual product viability
2. Valuation disconnect → Prices detach from revenue/profit fundamentals
3. FOMO-driven decisions → Fear of missing out drives irrational funding
4. Copycat saturation → Many companies pursue identical strategies

Both the dot-com and AI bubbles check all four boxes. But the underlying dynamics differ in ways that change the outcome.

The Dot-Com Bubble: What Actually Happened

Let me recap the dot-com era quickly, since many readers weren’t investing then.

Timeline:

1995-1999: Internet startups proliferate, valuations soar
March 2000: Nasdaq peaks at 5,048
October 2002: Nasdaq bottoms at 1,139 (78% decline)
~48% of dot-com companies survived

What drove the bubble:

“New economy” rhetoric claimed traditional valuation metrics no longer applied
Companies valued on “eyeballs” and “mindshare” instead of revenue
Low interest rates flooded capital into speculative ventures
Media hype created a feedback loop of ever-higher expectations

The aftermath:

Most companies failed (Pets.com, Webvan, eToys)
Survivors became giants (Amazon, eBay, Priceline)
Infrastructure investments proved valuable (fiber optics, server capacity)
The internet itself transformed every industry

I lost money in that crash. But I also learned something: bubbles pop, but the technology often endures.

The AI Investment Wave: What I’m Seeing Now

The current AI wave has eerily similar dynamics.

OpenAI: $80B+ valuation (2024)
Anthropic: $7.3B+ raised
NVIDIA: $3T+ peak market cap
AI startups: Hundreds launched monthly
Talent: AI engineers offered $1M+ packages

What’s similar:

Massive speculative funding for unproven business models
“AGI is coming” rhetoric replacing “new economy” rhetoric
Companies valued on “AI potential” instead of current revenue
Media hype cycle driving investment decisions

What I keep hearing from practitioners:

“LLMs are useful but I have to babysit them all the time”
“Most AI wrappers have no moat”
“Where are the backlog items being burned down?”
“The bubble will pop”

A Reddit discussion I followed captured the skepticism well. One engineer pointed out that software from major AI-using companies is “buggier than ever.” Another noted the lack of concrete examples where AI has truly disrupted a profession.

What’s Different This Time

Here’s where the comparison breaks down in important ways.

1. Infrastructure Maturity

Dot-com era:
- Internet infrastructure was nascent
- Bandwidth was expensive and limited
- Server costs were prohibitive
- Payment systems were unreliable

AI era:
- Cloud infrastructure is mature (AWS, Azure, GCP)
- Compute is expensive but accessible
- APIs and SDKs are standardized
- Enterprise integrations exist

In 1999, building an internet company meant building infrastructure. Today, building an AI company means plugging into existing infrastructure.

2. Immediate Utility

In the dot-com era, most websites provided minimal actual value. You could order pet food online, but why? You could get groceries delivered, but at what cost?

Today’s AI tools provide immediate, measurable value:

GitHub Copilot increases coding speed by 40-60% (studies vary)
ChatGPT reached 100M users in 2 months because it was immediately useful
Enterprise AI tools have paying customers from day one

This matters. Immediate utility creates sustainable demand that speculative hype cannot.

3. Incumbent Response

Dot-com era:
- Traditional companies ignored the web
- Newspapers dismissed online news
- Retailers resisted e-commerce
- Banks avoided online banking

AI era:
- Google: "AI-first" company
- Microsoft: Integrated AI across products
- Meta: Massive AI research investment
- Apple: AI features in every device

The incumbents learned from the dot-com era. They’re not ignoring AI—they’re racing to lead it.

4. Revenue Models

Many dot-com companies had no clear path to revenue. “We’ll figure out monetization later” was a common pitch.

Many AI companies already have revenue:

OpenAI: $2B+ annual revenue
Anthropic: Growing enterprise customer base
AI coding tools: Subscription models with proven conversion

This doesn’t mean valuations are reasonable. But it does mean more companies will survive a correction.

The AI Value Pyramid

Not all AI companies are equal. I think of the market in tiers:

Tier 1 (Foundation Models):
  OpenAI, Anthropic, Google, Meta
  - Massive compute requirements
  - Billions in investment needed
  - Clear business models emerging
  Risk: Medium-High (capital intensive)

Tier 2 (Infrastructure):
  NVIDIA, AWS, Azure, GCP
  - Selling picks and shovels
  - Sustainable demand
  Risk: Lower (established revenue)

Tier 3 (Enterprise Applications):
  Cursor, Notion AI, Jasper
  - Clear productivity value
  - Paying customers
  Risk: Medium (competition intense)

Tier 4 (AI Wrappers):
  Thousands of startups
  - Thin differentiation
  - Vulnerable to model updates
  Risk: Very High (most will fail)

The risk profile varies dramatically by tier. Tier 4 companies are most likely to fail in a correction.

What I’m Telling Investors

If you’re evaluating AI investments, here are the red and green flags I look for.

Red flags:

AI wrappers with no proprietary data or models
Companies that can’t explain their differentiation
Startups dependent on a single LLM provider
Valuations based on “potential” rather than revenue
Founders who talk more about AGI than customers

Green flags:

Proprietary data advantages
Clear enterprise use cases with paying customers
Infrastructure plays (chips, cloud, tools)
Domain-specific applications with expert oversight
Founders who talk about unit economics

The question isn’t whether a correction is coming. It’s whether the company can survive and thrive after the correction.

What I’m Telling Job Seekers

Career decisions in AI require similar discernment.

High Sustainability:
- AI infrastructure and tooling
- AI safety and alignment research
- AI integration for traditional industries
- Companies with clear revenue models

Medium Sustainability:
- AI product management
- AI training and fine-tuning
- AI consulting for enterprises
- Established tech companies adding AI

Higher Risk:
- AI startups without clear moats
- Pure "prompt engineering" roles
- Companies dependent on API wrappers
- Ventures with no revenue path

I’ve seen engineers jump to AI startups for 50% raises, only to be laid off 6 months later when funding dried up. The compensation premium is real, but so is the risk.

Common Mistakes I’m Seeing

Mistake 1: Equating “Bubble” with “Worthless”

The dot-com bubble popped. But Amazon became the world’s largest retailer. Google became the dominant search engine. Salesforce pioneered cloud computing.

The lesson: Bubbles pop, but the technology often transforms industries. The companies with real value survive and thrive.

Mistake 2: Ignoring the Winners of Tomorrow

During the dot-com crash:

Amazon stock fell 90% but fundamentals were solid
Google incorporated in 1998, raised $25M in 1999
Salesforce founded in 1999, survived and thrived

Great companies can be built during bubbles. The challenge is identifying them.

Mistake 3: Assuming All AI is Equal

The AI hype cycle is real:

Peak of Inflated Expectations (Current):
- "AI will replace all programmers"
- "AGI within 2 years"
- "Every app needs AI integration"

Trough of Disillusionment (Coming):
- AI assistants that hallucinate
- Startups that can't afford compute costs
- Enterprise AI projects that fail to deliver ROI

Slope of Enlightenment (What Survives):
- AI coding assistants (Copilot, Cursor)
- AI-powered search and research
- Domain-specific AI applications
- AI-enhanced creative tools

The trough is coming. But what emerges from it will be genuinely useful.

Mistake 4: Overestimating Short-Term Impact

From what I’m hearing from practitioners:

LLMs require constant supervision
Software quality hasn’t dramatically improved across the industry
Many “AI-powered” features are barely functional

Adoption takes time. The productivity gains are real but unevenly distributed.

The Reality Check

I don’t know when the AI correction will happen or how severe it will be. But I’m confident about a few things:

A correction is inevitable. The math doesn’t work for most current valuations.
The technology will endure. LLMs provide genuine productivity gains for many tasks.
Most AI startups will fail. That’s normal in any new technology wave.
The best companies will emerge stronger. Just like Amazon and Google after the dot-com crash.
Infrastructure players are safest. The “picks and shovels” strategy works.

What I’m Doing Differently

I’m not avoiding AI investments or careers. But I’m being more selective:

For investments:

Focus on infrastructure and tools
Look for proprietary data advantages
Verify revenue, not just “potential”
Avoid undifferentiated AI wrappers

For my career:

Building skills that complement AI, not compete with it
Learning how to evaluate AI tools critically
Maintaining technical depth that AI cannot easily replace

For my business decisions:

Asking: What specific problem does this solve?
Measuring: Can we calculate ROI?
Planning: What if the AI provider raises prices or shuts down?
Building: Do we have the talent to maintain this?

The Bottom Line

The AI bubble and dot-com bubble share striking similarities in speculative behavior and overvaluation. But critical differences exist in technology maturity and immediate practical utility.

A market correction is likely—perhaps inevitable. But the underlying AI technology will endure and transform industries, just as the internet did after the dot-com crash.

The key is distinguishing between hype and value:

LLMs provide real productivity gains despite overinflated claims
Proprietary data and paying customers matter more than “AI” buzzwords
Many AI startups will fail, but the best will emerge stronger

I’m not predicting doom. I’m predicting consolidation. And I’m positioning for what comes after.

Final Words + More Resources

My intention with this article was to help others share my knowledge and experience. If you want to contact me, you can contact by email: Email me

Here are also the most important links from this article along with some further resources that will help you in this scope:

👨‍💻 Reddit: Are people massively underestimating what's coming?
👨‍💻 Dot-com bubble historical data
👨‍💻 AI investment trends 2024

Oh, and if you found these resources useful, don’t forget to support me by starring the repo on GitHub!