There are an increasing number of financial articles and viral social media posts comparing the current rally in artificial intelligence and semiconductor chips to the dotcom era bubble. They highlight the same pattern: a sharp correction, a vertical recovery, and then full-blown euphoria. The warning is clear — the last time this happened, the Nasdaq crashed nearly 80%. Should we panic?

It’s a compelling visual. But does the chart pattern tell the whole story? Or are we seeing history rhyme without repeating?

Here’s a clear, data-driven breakdown of where the AI boom in 2026 looks similar to the dotcom bubble… and where it looks fundamentally different.

Both periods feature explosive rallies in a narrow group of “picks and shovels” stocks tied to a transformative technology. In 1999–2000, it was internet infrastructure (Cisco, Lucent, JDS Uniphase). Today, it’s AI infrastructure (Nvidia, Broadcom, TSMC, and memory names). Both saw:

In both eras, a handful of stocks drove the majority of market gains. The “Magnificent 7” (or more accurately, the AI infrastructure leaders) have shouldered an outsized portion of S&P 500 returns, just as a small group of internet stocks did in 1999–2000.

Tech valuations are stretched. Nvidia trades at roughly 30–35x trailing earnings and ~22–24x forward earnings as of mid-2026. While lower than their recent peaks, these multiples remain elevated relative to the broader market.

This is where the comparison breaks down.

This is the single biggest distinction. During the dotcom era, the vast majority of internet companies were unprofitable. Many had little to no revenue. Only about 7% of companies in the broader tech universe were generating positive free cash flow at the peak. Cisco — often compared to Nvidia — traded at 130x to 200x+ forward earnings at its March 2000 peak. Today’s AI leaders are highly profitable:

You cannot say the same about most dotcom companies in 1999–2000.

Even at current levels, Nvidia’s forward P/E of ~22–24x is dramatically lower than Cisco’s peak multiples. The broader S&P 500 forward P/E sits around 23x — elevated, but nowhere near the Nasdaq-100’s ~60x at the dotcom peak. More importantly, these multiples are being applied to companies with actual, accelerating earnings growth backed by real customer spending.

Hyperscalers (Microsoft, Amazon, Google, Meta, etc.) are on track to spend over $600 billion in capex in 2026 alone, with a large portion going directly to AI data centers and GPUs. Long-term forecasts point to trillions in AI-related infrastructure spending through 2030. This is not speculative hope — it’s actual money being deployed by the world’s richest companies because they believe AI will drive productivity and competitive advantage. In contrast, much of the dotcom-era spending was fueled by cheap capital chasing unproven business models.

The internet was real. AI is also real. The difference is the speed and scale of monetization. Enterprises are already using AI tools at scale. Hyperscalers have massive backlogs of orders for AI capacity. Semiconductor demand is being driven by rapidly growing training and inference workloads.

The core reason AI stocks can sustain higher valuations than dotcom stocks is earnings power and cash flow.

Nvidia’s revenue has grown at extraordinary rates on the back of real GPU demand. The big tech companies spending on AI infrastructure have strong balance sheets and are funding much of this themselves rather than relying on speculative IPOs and venture capital alone.

History shows that transformative technologies can justify premium valuations if the underlying economics eventually deliver. The internet did — just not on the timeline or for the companies investors bet on in 1999.

Artificial intelligence and the semiconductor sector have stronger near-term economics supporting the rich valuations.

That said, this doesn’t mean AI stocks can’t fall sharply. Elevated valuations always carry risk. Potential triggers for a correction include:

A 30–50% correction in AI leaders is always possible. A full 78% Nasdaq-style crash like 2000–2002 is much less likely because the companies have real earnings and cash flows to fall back on.

The chart patterns in viral social posts are eye-catching and worth respecting. Market psychology and technical setups can rhyme across decades.

But the fundamentals are meaningfully different this time:

The AI boom is not immune to volatility or even a painful correction. However, calling it “the exact same mistake as the dotcom era” ignores the vastly stronger earnings foundation underneath today’s leading companies.

The internet changed the world. AI very well might too. The question isn’t whether the technology is real — it’s whether current prices have gotten too far ahead of the earnings trajectory.

Only time will tell. But based on profitability, cash flow, and real-world spending, 2026 looks more like the early innings of a multi-year buildout than the final euphoric stage of a pure speculation bubble.

At Nicoya Research, we are buying any major dips of 20% or more in quality AI infrastructure companies. We focus on the ‘picks and shovels’ infrastructure plays, including rapidly-growing, profitable companies that make the critical components needed in AI data centers or semiconductor manufacturing. We believe this is the sweet spot that will continue to generate outsized returns for investors.

If you are not yet a premium member and would like to access our portfolios, newsletters, and chat room, upgrade by clicking here.

Cheers,

 Jason Hamlin, Founder, Nicoya Research