Will a market-cap-weighted composite of major AI-infrastructure stocks (NVIDIA, Broadcom, AMD, Oracle, Micron, TSMC, Vertiv, CoreWeave, Palantir, Super Micro Computer) fall at least 50% from its post-January-2025 peak closing value at any point before December 31, 2027?

probability: 26%

rationale:

Composite Structure and Current Drawdown Because this composite is market-cap weighted, it functions as a mega-cap semiconductor index heavily dominated by NVIDIA (~44% weight) and the rest of the top five—TSMC, Broadcom, Micron, and AMD—which together account for over 90% of the basket. The smaller, high-beta names like CoreWeave, Super Micro, and Vertiv collectively represent less than 1.5% of the weight and cannot independently trigger a resolution investors.coreweave.com. The running peak is also smoothed by staggered constituent highs, meaning the composite's all-time high is lower than the sum of its individual maximums. Reconstructing the aggregate market cap shows the composite is currently only about 10–15% below its post-January-2025 peak. Therefore, reaching the 50% drawdown threshold requires these highly profitable mega-caps to fall roughly another 40–45% from current levels.

Robust Fundamentals and Base Case The near-term fundamental evidence remains exceptionally strong, which firmly anchors the base case against a collapse. NVIDIA recently reported record quarterly revenue up 85% year-over-year nvidianews.nvidia.com, while TSMC and Broadcom are seeing similar demand surges 2 sources. Hyperscalers have reconfirmed 2026 capex guidance of roughly $700 billion combined, with signals for further increases in 2027 2 sources. Unlike the dot-com era, the dominant weights in this basket are massively cash-generative companies with strong balance sheets. Furthermore, broad high-yield credit spreads remain historically tight (with option-adjusted spreads around 2.7% recently) macrotrends.net, suggesting that an imminent, systemic credit crisis is not yet priced in.

Vulnerabilities and Debt-Financing Risks Despite these strong fundamentals, there are credible structural and cyclical risks that could drive a severe correction. The AI infrastructure buildout is increasingly dependent on debt and circular financing commitments, with hyperscaler debt issuance spiking and tech investment-grade spreads widening compared to broad indices 2 sources. Goldman Sachs estimates that roughly one-third of AI capex in 2026 is debt-financed, peaking around 35% in 2027 goldmansachs.com. If AI monetization fails to meet expectations, this leverage and counterparty risk could amplify a sharp selloff. Additionally, this sector exhibits extreme baseline volatility. The SOX index dropped 29% from its June 2026 peak to its July trough before rebounding 2 sources, and individual dominant names like NVIDIA and AMD suffered drawdowns of over 50% as recently as 2022.

Quantitative Dynamics and Outlook From a quantitative perspective, the roughly 16-month remaining window leaves ample time for a sudden AI capex pause, a memory-cycle rollover, or a macro shock to trigger a synchronized de-rating. Furthermore, the trailing-peak mechanic means any near-term rallies will ratchet the 50% threshold higher, increasing the surface area for a future drawdown. However, calibrating against recent shocks shows notable resilience: the severe July 2026 semiconductor rout produced only about a 20% drawdown in this specific cap-weighted composite because NVIDIA held up better than average chips. Evaluating the likelihood of an AI infrastructure bust alongside projected timelines for a broader AI market bubble pop and frontier lab IPO valuations also indicates a lower risk of a near-term crash. Ultimately, while this parabolic, highly concentrated basket faces elevated risks of a debt-amplified bust, forcing a 50% decline requires a genuine and sustained break in mega-cap capex and credit that current evidence does not yet support, anchoring my estimate at a 26% probability.