History Has Changed My Mind on These Millionaire-Maker AI Stocks (At Least for Now). Why I Have Suspended My Earlier Call.

The article flags CoreWeave (CRWV) and Iren (IREN) as higher-risk AI infrastructure plays due to heavy debt reliance. CoreWeave’s total debt has surged to $35B (from $21B in six months) versus just $5B in stockholders’ equity, raising solvency risk if AI demand stalls; the author also notes Nvidia-related “circular financing” concerns. Iren is less levered ($3.7B notes payable; $2.7B equity) but revenue is falling even as AI cloud revenue rises, implying both companies may need debt to fund expansion. With the Shiller P/E at 42 near highs, the author recommends keeping both stocks on a watch list until they can withstand a meaningful AI downturn.

Analysis

CRWV and IREN should be treated less like pure AI “growth” names and more like levered infrastructure credits. In a market where equity investors are already paying up for duration, the hidden vulnerability is refinancing: if AI capex growth merely normalizes, the equity can reprice much faster than the operating business because debt service absorbs the upside asymmetrically. The key second-order effect is that these names are effectively dependent on a permanently receptive bond market; once credit spreads widen, dilution becomes the bridge finance of last resort.

The more important winner is not the obvious chip supplier, but the better-capitalized toll collectors in the AI stack. NVDA still benefits if demand is real, but the market should distinguish between end-demand and financing-enabled demand: if customers are funding purchases with increasingly expensive leverage, order growth can decelerate even before AI inference volumes do. That would compress sentiment across the entire AI complex, while balance-sheet-light beneficiaries and hyperscalers with operating cash flow should outperform on relative multiple stability.

Catalyst path matters: over days, this is mostly a narrative drag; over 1-3 months, watch earnings calls, debt issuance windows, and any sign of slower contracted utilization or a pause in AI-related purchase commitments. Over 6-18 months, the thesis breaks only if these companies prove they can fund expansion from internally generated cash or long-duration, low-cost capital. The contrarian point is that debt is not automatically reckless here if supply remains scarce and contracted utilization stays high; in that case, leverage may simply be the fastest way to convert scarce compute into equity value, not a red flag.

Key Decisions for Investors

• Short CRWV into strength or via 1-3 month put spreads ahead of the next financing/refinancing window; thesis only works if credit spreads stay benign and utilization remains exceptional.
• Pair trade: long NVDA / short CRWV or IREN to isolate AI demand exposure from balance-sheet risk; this is a cleaner expression if the next leg of the AI trade is more selective than broad beta.
• For more tactical downside, buy IREN put spreads around the next earnings date if management needs to fund incremental build-out with debt; downside convexity improves if revenue mix keeps shifting away from legacy mining slower than expected.
• Set a watch item on investment-grade/HY AI-infrastructure issuance and lender appetite; if new funding comes at meaningfully wider spreads, that is the signal to increase shorts or reduce any longs in levered neoclouds.
• If AI capex guidance from hyperscalers reaccelerates and both names secure non-dilutive funding at tight spreads, cover shorts quickly; that would falsify the near-term credit stress thesis.