Is the S&P 500 in an AI Bubble?

The S&P 500’s gains have concentrated in a small group of AI-linked giants, reviving a big question: is this another bubble or a rational re‑rating? This article breaks down the evidence in plain terms.

You’ll see what concentration and valuations really imply, how AI earnings and capex shape the story, and which risks could puncture the trade. We also outline practical ways to reduce single-theme exposure without abandoning the index.

It matters now because valuation metrics are stretched, leadership is narrow, and earnings are soaring for a handful of names—conditions that can end with either a soft rotation or a sharp unwind.

The S&P 500 is not a uniform bubble, but parts of it—especially AI hardware and megacap platforms—trade with bubble-like characteristics. Index-level valuations and concentration are elevated, while earnings leadership is exceptionally narrow. That mix argues for caution: a powerful secular trend may be real, yet prices in some leaders could be running ahead of sustainable cash flows. Diversified investors should focus on risk controls, not binary calls.

• Concentration is extreme: the top 10 companies neared 40% of index weight by mid‑2025 (S&P Dow Jones Indices (research)).
• Valuations are rich: the Shiller CAPE hovered around 41–42 in May–June 2026, near historic highs ex‑1999 (multpl.com).
• Leadership is narrow: a 28‑session rally into May 8, 2026 saw ~69% of gains from just 10 stocks (TECHi reporting Nomura).
• Earnings power is real but concentrated: NVIDIA’s fiscal Q1 2027 revenue hit $81.6B, +85% YoY, with $75.2B in Data Center (NVIDIA press release).

What does today’s market concentration signal?

High concentration can be both a symptom of genuine innovation and a vulnerability. When a few firms capture most investor enthusiasm, the index’s risk profile quietly shifts from “broad U.S. economy” to “a handful of cash-flow engines.” That’s what we see today with AI-linked megacaps.

By mid‑2025, the 10 largest S&P 500 companies represented almost 40% of the index, according to S&P Dow Jones Indices (S&P Dow Jones Indices (research)). This kind of concentration is historically uncommon and makes index returns more sensitive to idiosyncratic news from a small cohort.

Concentration alone doesn’t prove “bubble.” If outlier earnings persist and broaden through the value chain (chips → systems → software → services), leadership can remain narrow for years. But when concentration coincides with weak breadth, it often signals a vulnerable advance where one theme carries the market while the rest hesitates.

Are valuations consistent with a bubble or a re‑rating?

Valuations, measured by the Shiller CAPE, sat around 41–42 in May–June 2026—levels rarely seen outside the late‑1990s peak (multpl.com). That places the market in a zone where long‑term forward returns have historically been lower and drawdowns more frequent. It doesn’t time tops, but it calibrates risk.

Is this a classic bubble? Bubbles usually feature high multiples disconnected from cash flows. Today, some leaders actually post exceptional earnings and free cash flow, bolstered by an AI buildout. The question is durability: can supply, pricing, and customer budgets support years of current run rates?

A plausible middle path: a secular AI re‑rating inside an expensive market. That means large payoffs for proven winners but heightened index‑level risk if growth expectations reset. For diversified holders, valuation adds caution, not necessarily a sell signal; position sizing and rebalancing matter more than grand market calls.

How durable is the AI earnings and capex cycle?

Recent earnings underscore the power of AI infrastructure spending. On May 20, 2026, NVIDIA reported record fiscal Q1 2027 revenue of $81.6B, up 85% year‑over‑year, with Data Center revenue at $75.2B (NVIDIA press release). Those numbers reflect a capex boom by hyperscalers and enterprises racing to deploy generative AI.

The durability test is whether demand broadens beyond a few hyperscalers, hardware constraints ease without crushing margins, and second‑order beneficiaries (software, services, networking, power) see sustained profit lift. If AI value accrues to more nodes in the stack, the market’s earnings base deepens and bubble risk recedes.

Conversely, if customers slow orders to digest capacity, pricing compresses amid competition, or power and supply bottlenecks persist, growth expectations could deflate. That would hit the index through its most heavily weighted names.

• Checklist: Gauging AI cycle resilience
• Backlog visibility: multi‑quarter purchase commitments versus just‑in‑time orders.
• Customer breadth: rising share of enterprise, not just a few hyperscalers.
• Unit economics: stable margins despite supply expansion and competition.
• Power and infrastructure: secured capacity for data centers and networking.
• Downstream adoption: measurable productivity or revenue lift in software and services.

Is breadth improving or getting narrower?

Breadth tells you how many stocks are pulling the market higher. Narrow breadth can persist, but it increases fragility because a stumble in a few names can drag the index down.

In the 28‑session rally from March 30 to May 8, 2026, the S&P 500 rose about 16.6%, and Nomura’s attribution showed roughly 69% of that rally came from just 10 stocks (TECHi reporting Nomura). That’s the definition of concentrated leadership.

Watch for signs of broadening: the equal‑weight S&P 500 outperforming the cap‑weight version over multi‑week windows, stronger advance‑decline lines, and sector rotation that lifts cyclicals and mid‑caps. If those improve alongside robust AI earnings, the “bubble” thesis weakens; if they don’t, fragility remains elevated.

What risks could derail the AI trade and the index?

Several pressure points loom. First, supply and power constraints: the AI stack relies on advanced chips, specialized networking, and ravenous data center power. Delays in capacity additions or energy availability can stretch deployment timelines.

Second, competition and pricing: if leading margins invite aggressive rivals, unit pricing could soften faster than volume grows. Third, customer digestion: after a spend surge, large buyers may pause to optimize existing deployments, delaying follow‑on orders.

Higher-for-longer rates tighten financial conditions and compress multiples for long‑duration assets. A growth scare could reprice cyclicals and megacaps together, reducing diversification benefits when you need them.

Pro tip: Separate your view on “AI is real” from “this price is justified.” You can believe in the theme yet manage timing risk with staggered entries, periodic rebalancing, and guardrails on single‑name or sector weights. None of this is financial advice—just risk hygiene.

How can investors express a view without overconcentrating?

If you worry about concentration but still want core U.S. equity exposure, there are ways to dial the theme up or down without making an all‑or‑nothing bet. Consider how index construction, factor tilts, and explicit risk controls alter your exposure to megacaps.

Approach What it does Where it helps Key trade‑offs Cap‑weight S&P 500 Tracks market value; highest megacap/AI sensitivity When leaders outperform persistently Concentration risk; theme reversals hurt more Equal‑weight S&P 500 Spreads weight evenly across constituents When breadth improves and mid‑caps participate Can lag when megacaps dominate Quality/Profitability tilt Emphasizes balance‑sheet strength and margins During late‑cycle or volatility spikes May reduce exposure to high‑beta AI plays

Additionally, set explicit guardrails: cap any single stock at a defined percentage, rebalance on a schedule, and consider holding some cash or short-duration bonds as a volatility buffer. The goal isn’t timing perfection—it’s avoiding concentration mistakes that become obvious only after the fact.

What evidence would invalidate the bubble case?

Strong, sustained diffusion of AI profits beyond the top cohort would argue against a bubble. If software and services show multi‑year revenue acceleration tied to AI deployments, while margins remain healthy despite rising supply, the earnings base for the index broadens.

Market‑structure markers would also help: improving breadth (equal‑weight leadership), lower single‑name skew in index attribution, and more moderate valuation measures even as earnings grow. In short, if leadership persists but broad participation lifts, the “bubble” label loses power.

Finally, if the Shiller CAPE cools from extreme levels without a deep drawdown—because earnings growth outpaces price gains—that would look more like an orderly re‑rating than a speculative blow‑off (multpl.com).

Common Mistakes

1. Chasing parabolas: Buying vertical moves without a plan invites regret. Stagger entries and pre‑commit to rebalancing thresholds.
2. Confusing themes with prices: AI can be transformative and still be overpriced. Separate fundamental conviction from risk management.
3. Ignoring breadth: Narrow leadership raises fragility. Track equal‑weight vs cap‑weight performance and advance‑decline data.
4. Using CAPE as a timing tool: High CAPE warns about risk but doesn’t call tops. Use it to shape allocation ranges, not binary decisions.
5. Concentration blindness in index funds: Cap‑weighted exposure can quietly become a megacap bet. Set position caps and review holdings periodically.

For cross‑asset context on liquidity, correlations, and how equity leadership spills into digital assets, see ongoing coverage at Crypto Daily.

Frequently Asked Questions

Does a high Shiller CAPE mean a crash is imminent?

No. A high CAPE has historically correlated with lower long‑run returns and higher drawdown risk, but it does not time crashes. It’s a risk gauge, not an alarm clock.

Could AI productivity justify higher multiples for years?

Possibly. If AI materially boosts revenues and margins across many sectors, the market can support higher steady‑state multiples. The burden of proof is on durable, broad earnings growth rather than one or two leaders.

Are equal‑weight funds automatically “safer” now?

They reduce single‑theme concentration, which can help if megacaps wobble. But they can lag during periods when leaders keep winning. “Safer” depends on your risk tolerance and time horizon.

How might an AI drawdown affect crypto markets?

When megacaps wobble, risk appetite and liquidity often tighten across assets. Crypto can see higher volatility and correlations rise, especially during de‑risking episodes. Effects vary by cycle.

What would signal that breadth is genuinely improving?

Persistent outperformance of equal‑weight over cap‑weight, broader 52‑week high lists, and sector rotation beyond AI and platforms, sustained over weeks—not just a few sessions.

Is it better to wait for a pullback before adding?

There’s no universal answer. A common approach is to phase entries over time and rebalance on schedule to avoid anchoring on a single price. This manages regret risk without prediction.

Which single indicator should I watch most?

No single metric suffices. Combine valuation context (e.g., CAPE), earnings follow‑through (especially beyond top names), and breadth indicators. Together they frame risk better than any one signal.

Disclaimer: This article is provided for informational purposes only. It is not offered or intended to be used as legal, tax, investment, financial, or other advice.