Chapter 6 — Cisco & the Dot-Com Bubble

What Cisco teaches us about bubbles, infrastructure, hype cycles, and why NVIDIA mirrors the same historical pattern.

At the height of the dot-com boom in 1999, Cisco briefly became the most valuable company in the world. It wasn’t a consumer darling like Amazon or a flashy startup like Pets.com. Cisco didn’t sell to ordinary people at all — it sold the invisible infrastructure the internet needed to function: routers, switches, networking gear, the plumbing of the online world.

To many investors in 1999, Cisco felt unstoppable.
To many skeptics in 2001, Cisco felt like a fraud.

Both groups were mistaken.

Cisco was neither a miracle nor a mirage.
It was simply a company perfectly positioned for a technology wave the world misunderstood.

And that makes Cisco the perfect analogy for today’s AI giants — especially NVIDIA.


What Cisco Did Right

Cisco didn’t chase consumer hype.
Cisco didn’t guess at business models.
Cisco didn’t try to reinvent the internet.

Cisco focused on one mission:

Build the hardware that connects the world.

And Cisco understood something crucial far earlier than most:

“The internet will grow faster than anyone expects.”

To capitalize on that insight, Cisco did four things exceptionally well:

1. It built the foundational infrastructure everyone needed.

Every website, every e-commerce store, every corporate network — nearly all ran on Cisco hardware.

2. It acquired aggressively and strategically.

Cisco bought dozens of companies in the 1990s, absorbing talent and tech before competitors could respond.

3. It partnered deeply with enterprises.

Cisco wasn’t a media darling.
But CIOs trusted it.
Enterprises standardized on it.
Distribution became Cisco’s superpower.

4. It expanded in lockstep with internet traffic growth.

Every time the internet doubled in users, Cisco doubled its footprint.

Cisco didn’t grow because of hype.
Cisco grew because the internet grew.

The deeper pattern:

When a new technology becomes infrastructure, the companies that power it become dominant.


What the Market Overestimated

During the bubble, investors didn’t just believe in Cisco —
they believed in an infinite internet explosion.

Assumptions included:

  • every company would need endless networking gear
  • usage would grow exponentially forever
  • valuations could climb indefinitely
  • Cisco would own all infrastructure
  • competitors would never catch up
  • margins would remain sky-high

These assumptions were unrealistic — not because Cisco was weak, but because expectations were too extreme.

When the bubble burst, narratives swung in the opposite direction.

  • People called Cisco a bubble loser.
  • Commentators said Cisco’s best days were gone.
  • The market punished it heavily.

But the truth?

Cisco didn’t fail.
Expectations failed.

Cisco survived, evolved, and remains one of the most important infrastructure companies in the world.

The lesson:

“A bubble does not mean the underlying technology is fake. It means people temporarily misprice the future.”


The Parallel to NVIDIA

The Cisco analogy fits NVIDIA almost perfectly.

Like Cisco, NVIDIA powers the infrastructure

AI is built on compute.
NVIDIA owns that compute.

  • GPUs
  • CUDA
  • Networking
  • Software stacks
  • AI frameworks
  • Enterprise distribution

Just as the internet couldn’t scale without routers, AI cannot scale without massive compute.

Like Cisco, NVIDIA grows with the ecosystem

As AI usage grows, NVIDIA’s total addressable market expands.

  • More inference → more GPU demand
  • More enterprise adoption → more data centers
  • Larger models → more compute

Like Cisco, NVIDIA faces overheated expectations

  • Analysts project impossible growth
  • Valuations soar
  • Commentators proclaim NVIDIA will be “bigger than all of tech”
  • People forget competition always arrives eventually

NVIDIA is not a bubble…
…but it lives inside a bubble environment.

This is the nuance most people miss.


Why Bubbles ≠ Scams (Bubbles Accelerate Adoption)

The word “bubble” triggers fear, suspicion, and memories of 2000 or 2008.

But historically, bubbles are not scams —
they are infrastructure accelerators.

They overbuild the systems the world later depends on.

Examples:

Railroad bubble (1800s)

Overbuilt track → powered manufacturing growth.

Electricity bubble

Overbuilt grids → fueled industrial productivity.

Dot-com bubble

Overbuilt fiber networks → enabled YouTube, Google, Amazon, Netflix.

Housing bubble

Scaled financial infrastructure → increased mortgage availability.

Bubbles waste capital —
but leave behind infrastructure.

Long term:

  • hype fades
  • weak companies die
  • infrastructure remains
  • survivors dominate

That’s why bubbles ≠ scams.
They’re chaotic, but they accelerate progress.

The AI bubble is doing the same:

  • hyperscalers overbuilding data centers
  • NVIDIA overbuilding compute
  • governments investing in GPU clusters
  • startups overexperimenting
  • enterprises overpiloting AI tools

Expensive now…
but it becomes the foundation for the next decade.


How Today’s AI Cycle Mirrors the Internet 1996–2001 Arc

In 1996:

  • internet felt magical
  • investors excited
  • startups everywhere
  • technology immature
  • infrastructure insufficient
  • business models unclear

In 1999:

  • hype exceeded reality
  • valuations unrealistic
  • money poured in too fast

In 2001:

  • bubble burst
  • weak companies died
  • strong companies survived
  • infrastructure remained

In 2005+:

  • real internet economy began
  • Amazon scaled
  • Google dominated
  • Facebook emerged
  • cloud arrived
  • mobile exploded
  • SaaS took off

AI in 2025 is exactly the same.

We are somewhere between 1998 and early 1999:

  • not early enough to feel experimental
  • not late enough for consolidation

The pattern is repeating:

  • early magic
  • hype overshoot
  • market frenzy
  • infrastructure boom
  • startup overcrowding
  • inevitable shakeout
  • platform winners
  • long-term transformation

Understanding this pattern helps you make better decisions as a builder, leader, worker, and investor.

Because:

The internet didn’t end in 2001. It began.
AI won’t end with the next correction. It will begin.