When Euphoria Peaks: AI Stock Mania & the Case for Real Assets

AI will change the world—but parts of the market may already price perfection. We compare today’s AI valuations with the dot-com era and make the case for allocating to durable, cash-flowing real estate.

The air is getting thin at the top

In every cycle there’s a story investors fall in love with. In the late 1990s it was the commercial internet; in 2006 it was housing-linked credit; in 2021 it was SPACs. Today, it’s artificial intelligence. The transformative potential is real—but even world-changing technologies can be priced for more than perfection. When that happens, portfolios built only on momentum become fragile. The smarter move is balance: own durable, cash-flowing assets—like well-underwritten real estate—alongside public equities.

Recent market action shows how quickly AI enthusiasm can push prices to extremes, with single-day moves that rival the dot-com era. Some flagship names now trade at valuation levels that leave little room for error.

Great company ≠ great stock at any price. The math of multiples is unforgiving when expectations reset.

A simple picture of where valuations stand

This is not a forecast; it’s context. Some leaders trade above the dot-com reference multiple; others remain well above long-term market norms. The timing of a top is unknowable—but stretched concentration and lofty expectations raise the cost of surprise.

What the dot-com playbook teaches

The internet didn’t fizzle after 2000—our economy was rebuilt on it. Yet investors who bought at peak valuations waited years to break even. Cisco, briefly the most valuable company on earth in 2000, traded near ~39× sales and fell ~88% in the subsequent bear market. Two lessons endure:

1. Great company ≠ great stock at any price. The higher the multiple, the less room there is for upside surprises—and the harsher the penalty for small disappointments.

2. Capex waves are lumpy. Infrastructure buildouts (then routers; now GPUs, networking, and data centers) often overshoot near-term demand. As investment digests, growth slows and margins compress before the next leg higher.

Today’s AI cycle rhymes with both points. Analysts model multi-trillion-dollar infrastructure spend over the decade. If that spend slips, shifts vendor mix, or yields lower returns than modeled, the first place you’ll see it is in revenue growth forecasts and valuation multiples.

“But AI revenues are real this time”—agreed… and that’s the point

Acknowledging current strength does not weaken the argument for balance. It strengthens it.

- Revenue concentration risk. Several AI leaders rely on a small set of hyperscale customers or a single product cycle. A modest pivot in those budgets can cascade through the ecosystem.

- Capital intensity & returns. Massive capex can depress free cash flow if pricing power erodes or utilization lags expectations.

- Reflexivity. Elevated share prices enable cheap equity financing and aggressive spend—until prices wobble, then the feedback loop runs in reverse.

The rational conclusion isn’t to avoid AI exposure. It’s to avoid *over-reliance* on a single narrative priced for perfection.

Real estate’s role: ballast, cash flow, and sensible upside

Thoughtfully underwritten real estate can counterbalance equity-market cyclicality:

1) Cash flows you can underwrite. Stabilized assets produce contractual rents, renewal options, and quantifiable expenses—metrics within an investor’s control.

2) Inflation-hedging potential. Over long horizons, rent escalators and replacement-cost dynamics can help real estate keep pace with price levels.

3) Lower correlation (by design). Private real-estate returns—especially in niche operating segments—tend to exhibit lower short-term correlation to public equities.

4) Value creation beyond multiple expansion. In development, asset management, and repositioning, much of the return is generated through execution—entitlements resolved, NOI improved, capital deployed wisely.

For our firm, “meaningful real-estate projects” means assets with use-case durability—hospitality in truly supply-constrained locations, well-located multifamily with durable demand drivers, and specialized assets tied to real economic activity.

How much is “enough” ballast?

Every investor’s situation is unique, but three principles travel well:

- Right-size AI exposure to your horizon. If a 30–50% drawdown in your equity sleeve would imperil near-term goals, exposure is likely too high.

- Carve out a dedicated real-assets sleeve. A 10–30% allocation to income-producing or development-to-core projects can materially change portfolio risk while preserving upside.

- Favor projects with downside mitigants. In today’s rate environment, focus on conservative leverage, proven operating teams, and business plans that don’t require heroic exit multiples.

Risks to our view (and how we address them)

If AI’s S-curve is faster than modeled: We still benefit through targeted equity exposure while ensuring we don’t need multiple expansion to hit our targets in real estate.

If rates fall and long-duration equities rally: Lower discount rates can also compress cap rates and reduce development carry costs—a tailwind to quality projects.

If a recession hits: Quality leases and resilient demand drivers can cushion income, while development pipelines can be timed to deliver into the next up-cycle.

The opportunity right now

Today’s AI leaders are remarkable companies. History suggests that when valuation, concentration, and narrative crest together, prudence—not bravado—wins. That’s where real estate fits: owning the cash flows of indispensable places that benefit from long-term demographic and supply dynamics, regardless of which ticker is hottest this quarter.

Interested in a resilient allocation? Contact us to discuss current offerings and how they may complement your equity portfolio.

Figure Sources: Multiple valuation references compiled from reputable market data providers; dot-com reference based on Cisco’s ~39× P/S near the March 2000 peak and its subsequent drawdown.