‘Yet another way in which 2026 is looking like 1999’: Top analyst fears bubble popping with investors and Wall Street out over their skis

AI-linked megacap tech stocks drove the worst market sell-off since October, with hundreds of billions of dollars in value erased in one session after a stronger jobs report revived Fed-hike fears and Broadcom issued weak guidance. The article highlights rising bubble concerns around stretched valuation multiples, 20.2% expected long-term S&P 500 earnings growth, and aggressive 2026 forecasts from Wall Street, alongside warnings from Dimon and Dalio. It frames Friday’s drop as a risk-off signal for AI and high-multiple equities rather than evidence that the bubble has already burst.

Analysis

The selloff is less about a one-day macro scare than about an increasingly fragile earnings bridge supporting the AI complex. When the market is pricing a decade of near-perfect monetization, even a modest reset in rate expectations or capex optics can force multiple compression before fundamentals have to move. The vulnerable part of the chain is not just semis; it is every company whose equity story depends on the market continuing to fund a capital-intensive AI arms race at ever-richer terms.

The second-order effect is that tighter scrutiny of ROI should hit the ecosystem unevenly. Hyperscalers and chip suppliers can still grow revenue, but their customers, software vendors, and IPO aspirants face a harder financing backdrop if the market stops rewarding “growth at any price.” That is especially true for businesses whose usage is subsidized or whose valuation depends on perpetual share gains rather than durable unit economics; those names are the most exposed to a rotation from narrative scarcity to cash-flow discipline.

The more interesting near-term risk is that this does not need to become a recession to matter. In the next 2-8 weeks, a sequence of hot data, higher-for-longer rhetoric, or another weak AI guidance print could keep pressure on the highest-duration names while broader indices remain resilient. Over 3-12 months, the bigger catalyst is not a “bubble burst” but a capex air pocket: if buyers start demanding proof of payback, the market can re-rate before revenues roll over.

Consensus is still treating the AI spend cycle as self-financing because top-line growth has been visible. The missed point is that visible revenue does not equal durable economics when pricing is promotional and customer concentration is extreme. If token subsidies or cloud discounts normalize, reported demand can fade quickly, and the market may discover that a lot of the current growth is more circular than organic.

Key Decisions for Investors

• Short NVDA vs long QQQ on a 4-8 week horizon: use a 1:1 pair to isolate AI-duration compression; target a 10-15% relative drawdown if rates stay sticky and capex skepticism deepens.
• Initiate a tactical short in AVGO into any post-guidance bounce; the stock is the cleanest way to express vulnerability to slowing AI spend momentum and multiple compression. Risk/reward improves above prior support after weak forward commentary.
• Buy put spreads on MSFT or AMZN 2-3 months out to hedge hyperscaler capex disappointment; these names can still report solid revenue while margins and free cash flow face pressure from AI infrastructure intensity.
• Short an equal-weight basket of AI-adjacent IPO/late-stage software names against JPM or other balance-sheet beneficiaries if primary issuance reopens; the trade plays the shift from narrative funding to capital discipline.
• Keep a standing trigger to cover shorts if a new catalyst re-accelerates estimated earnings revisions by more than ~5% in a quarter; that would indicate the market is getting fresh evidence the spend is converting to monetizable demand.