How to survive the AI hype… without missing the next Amazon

March 15, 2000.

Jeff Bezos sits in his Seattle office, reading the morning headlines.

The dream is over.

His company’s stock has collapsed from $113 to $6 in twelve months; a 95% wipeout of shareholder wealth. Analysts are writing obituaries for e-commerce itself. He doesn’t panic. He keeps building.

Twenty-five years later, we’re watching the same movie with different actors. Nvidia has become the most valuable semiconductor company on Earth. Palantir trades at 380× earnings. Every week, another AI startup raises billions before generating a single dollar of profit.

The pattern of every bubble

Every bubble follows a familiar rhythm:

Innovation → Imitation → Inflation → Implosion → Integration

• Innovation begins with genuine breakthroughs, the kind that make the impossible suddenly look inevitable.
• Imitation floods in as capital and copycats chase the early winners.
• Inflation drives valuations far beyond earnings, anchored on belief instead of math.
• Implosion is when the belief breaks.
• Integration is when the survivors, the real builders, lay the next foundation.

But here’s what most investors miss: Not all survivors create wealth. Some companies endure but never recover. They’re operationally sound, but financially ruinous. In short, they survive but don’t grow. On the other hand, a handful of companies don’t just survive the reckoning; they define the next era.

So the question isn’t whether a bubble will pop (it always does)…

The question is: What separates Amazon from Pets.com? Nvidia from Cisco?

That distinction between survival and wealth creation, is what the next two stories reveal.

Amazon: Building through the rubble

When Amazon lost 95% of its value during the dot-com crash, e-commerce was declared dead. Bezos was written off as reckless.

Imagine being that investor in 2001: Your portfolio down 95%. Every financial paper mocking your judgment. Your spouse wondering if you’ve lost your mind. Would you have held on? Most wouldn’t.

Over the next two decades, Amazon then went from $6 to $4,968… (I know, I wished I had bought some back then!) But then again, that’s just hindsight. But at the time? It sure feels like insanity. And that’s what makes Amazon’s story so instructive.

Because even in the ashes of a burst bubble, Bezos didn’t buy the media narrative, he kept building the business. He doubled down on logistics when analysts demanded layoffs. He hired engineers when others froze headcount. He reinvested every dollar into speed, reliability, and scale. Basically, he did things that didn’t move the stock, but built the business.

What Amazon did right

1. Long-term obsession. Bezos measured success in decades, not quarters. His 1997 letter already said it:

“We will continue to make investment decisions in light of long-term market leadership rather than short-term profitability.”

That philosophy turned out to be survival itself.

2. Irreplaceable value. Amazon wasn’t selling a dream. It was delivering something people came to depend on: better prices, faster shipping, and more selection than any physical store could match. This wasn’t “nice-to-have”, it was habit-forming. By 2003, while Pets.com and Webvan were punchlines, Amazon customers were checking the site daily.

3. Reinvention through necessity. In solving its own internal computing chaos, Amazon created AWS – a side project to rent out excess server capacity. That “side project” now generates more operating profit than all of retail. AWS didn’t exist because Bezos was a visionary. It existed because Amazon built infrastructure to solve real problems, then realized others had the same problems.

So yes, it’s easy to celebrate Amazon today as a $2 trillion titan. But back then, survival didn’t look visionary… it looked like stubborn denial. That’s what surviving a bubble feels like.

Cisco: When great businesses become wealth destroyers

Now, the other side of the story.

At the peak of the dot-com boom, Cisco was the backbone of the Internet. Dominant, profitable, indispensable. Its routers powered nearly every byte of data online. Revenue was growing 40% annually. Profit margins were world-class. In March 2000, investors valued it at 148× earnings… making it, briefly, the most valuable company on Earth.

Then the bubble burst.

Cisco’s revenue only dipped modestly. Its profits stayed solid.

But the stock?

It collapsed 86%, from $80 to $11. Twenty-five years later, it still hasn’t recovered its 2000 peak. (Yes, TWENTY-FIVE YEARS…) Let that sink in. You could have bought one of the best businesses in the world at that point, held it for two decades, and still be underwater.

Cisco wasn’t alone:

• Qualcomm dominated mobile chips, yet it took 16 years to break even.
• Oracle controlled enterprise databases, and it took 17 years to recover.
• Verisign ran the backbone of the internet, and still took 21 years to reclaim its peak.
• Intel was the unquestioned king of processors, but it never matched its 2000 valuation.

These weren’t bad companies. They were mispriced companies.

The lesson? You can be completely right about the company and catastrophically wrong about the price.

In a bubble, valuation isn’t a side issue. It becomes the whole story.

Which brings us to 2025…

The multi-trillion-dollar question

Everyone knows AI is the future.

And Nvidia has, without a doubt, become the beating heart of the AI revolution. Its GPUs power every major foundation model. Its CUDA ecosystem creates the kind of lock-in that made Amazon and AWS legendary. Revenue has grown 10× in three years. It has the ecosystem gravity and irreplaceable value of a young Amazon. But it trades at 50× earnings.

So here’s the multi-trillion-dollar question facing every investor today: Is Nvidia the next Amazon or the next Cisco?

Let’s do the math…

At about 50× earnings, the market is assuming Nvidia can grow EPS by roughly 50% a year. The challenge is whether Nvidia can sustain that level of expansion, given its size, market dominance, and the inevitable slowdown that comes with scale. That would mean that after just three years, Nvidia would have to grow its earnings to nearly 3× their current level, a high hurdle for a company already this large.

Is it possible?

Maybe…

The AI total addressable market is genuinely massive, and Nvidia’s software moat through CUDA is very real. But the company will still need to continually fend off competition from AMD and custom chips developed in-house by Google, Amazon, and Microsoft — all while managing ongoing supply-chain bottlenecks.

Then there’s Palantir, which spent 17 years in the red before finally turning profitable, and is now growing more than 60% year over year as it embeds itself deeper into government and enterprise infrastructure. Yet it trades at a staggering 380× earnings!

Investors are essentially betting that Palantir will become the operating system for enterprise AI, a massive claim that already appears fully priced in. At this valuation, the margin for error is razor-thin. Even small disappointments in growth, adoption, or execution could lead to significant downside long before the long-term story plays out.

Amazon has proved that, occasionally, companies actually do justify the insane faith early investors placed in them. But for every Amazon, there are fifty Pets.coms that had the same narrative and none of the execution.

My Take?

Nvidia has Amazon-level business strength and will likely remain dominant, highly profitable, and central to the entire AI infrastructure stack. But at 50× earnings, Nvidia must deliver steady, compounding growth (which becomes increasingly difficult at its scale) to justify its current valuation.

Palantir shows the same obsessive culture and deep customer stickiness that once defined early Amazon. But at 380× earnings, it would be a risky investment right now. Palantir may be an exceptional company, but its valuation is even more extreme.

The point isn’t to avoid great companies. It’s to understand that valuation is the difference between a great company and a great (or not so great) investment.

The survival blueprint

If history is any guide, the next few years will separate the builders from the bystanders.

Here’s the blueprint, ranked by what truly matters:

Tier 1: Non-negotiables

Valuation. This comes first because nothing else matters if you overpay. Cisco survived. Intel survived. Oracle survived. Their shareholders didn’t… not for decades. The best business at a bad price is still a bad investment.

Reality check: If the company doubles its profits in five years and you still don’t make much, the problem wasn’t the business; it was the price.

Irreplaceable value. Amazon didn’t deliver “nice-to-have”, it delivered something customers came to depend on. Pets.com also delivered tangible value (cheaper dog food at your door!), but it was replaceable. Discretionary. A luxury, not a necessity.

Reality check: If this company vanished tomorrow, would customers scramble… or shrug?

Tier 2: Force multipliers

Ecosystem gravity. Cisco had this. Microsoft has this. Nvidia has this. AWS has this. It’s where others build on your infrastructure, which makes your infrastructure more valuable, which attracts more builders, which… you get the point. Amazon didn’t just sell books. It became the platform where millions of merchants built their businesses, locking them into AWS, Fulfillment by Amazon, and the entire Amazon flywheel.

Reality check: Are others building on this company’s infrastructure, or merely buying its products?

Founder-led obsession. Jeff Bezos, Elon Musk (can’t believe I’m saying this) and Jensen Huang are visionaries who think in decades, not quarters. This is what lets them build while others retreat.
Reality check: Is leadership measuring success in decades, or trying to impress analysts next quarter?

Tier 3: Resilience

Capital discipline. When the bubble bursts, capital disappears. Companies that depended on endless cheap funding die whereas companies that can self-fund or control burn survive. Amazon did this by cutting experiments that didn’t scale and doubling down where it had conviction.

Reality check: Can this company survive three years without external capital?

Adaptability. Amazon started selling books. Then everything. Then cloud computing. Then streaming. Then logistics. Bezos didn’t predict AWS in 1997… but he built a culture that allowed it to emerge from solving real problems.

Reality check: Is this company defending its past or creating its future?

The fifth perspective

In 2000, Cisco traded at 148× earnings, a price that assumed perfection in a world that never delivers it. Its business endured, but the stock’s journey showed how long excellence can take to justify euphoria. Around the same time, Amazon traded at $6, a price that assumed failure in a business still quietly compounding underneath the panic. Two decades later, those assumptions reversed.

Fast forward to today: Nvidia trades at 50× earnings and Palantir at 380×. Both embody the same paradox that has defined every major market cycle — extraordinary companies burdened with extraordinary expectations.

Maybe this AI wave will prove larger than any before it. Maybe some of these names will justify or even surpass what’s priced in.

But the math hasn’t changed.

Even when you’re right about the company, the outcome still depends on the price you paid to believe in it.

Our goal is not to predict the pop.

It’s to own what survives at a price that lets you compound through the chaos.

Because when the dust settles, the real builders, the next Amazon, the next Nvidia, or the next big thing… will define the next decade.

Disclaimer: The author holds positions in Amazon and Palantir. The views expressed are his own and do not constitute financial advice.