Nvidia Sold All of Its Applied Digital Stock. Should You Follow Suit?

Nvidia sold its entire ~7.7 million-share Applied Digital stake (about $177M) during Q4, disclosed in its Dec. 31, 2025 13F. Applied Digital reported fiscal 2026 Q2 revenue up 250% YoY (period ended Nov. 30) while net loss widened as it invests heavily in data-center buildout; the article views Nvidia's sale as portfolio noise and remains bullish on long-term cash-flow potential, advising patience as capacity comes online.

Analysis

The NVDA sell-down is a headline but likely noise for Applied Digital’s underlying economics: the second-order beneficiary of hyperscale AI demand is the physical stack—power, cooling, transformers and long-term leases—where margin accrual is sticky once facilities hit utilization. Applied Digital’s near-term P&L will remain capex-heavy; true free cash flow inflection depends on reaching sustained 70-80% rack-level utilization across new pods, not quarterly revenue prints, which creates a multi-quarter runway (6–24 months) for re-rating if utilization and multi-year contract tenure are announced.

Counterparties and suppliers to Applied Digital will see asymmetric wins: power equipment and substation installers (long lead times, constrained supply) get squeezed as capacity ramps, which can bottleneck expansion and give pricing power to those vendors. Conversely, incumbent colo landlords (e.g., large REITs) face revenue pressure in the highest-density AI corridors as customers seek turnkey, high-PUE capacity at scale—I expect selective displacement in greenfield markets within 12–18 months.

Key risks are financing and concentration: a stretched balance sheet or a single large customer negotiating below-commitment pricing could force equity raises and meaningful dilution; that reversal can happen quickly on a missed utilization cadence or if power interconnects slip by 2–4 quarters. The practical catalyst set to watch is: monthly/quarterly utilization ramps, signed multi-year leases with collateralized payment schedules, and PPAs/utility interconnection confirmations—these are the binary items that will move the valuation multiple over the next 3–12 months.

Key Decisions for Investors

• Tactical long APLD (size 1–2% NAV) with downside protection: buy shares and purchase a 6–9 month protective put ~20% OTM to cap dilution-risk during the next financing window; target 2.5–3x upside if utilization moves to 70%+ across new capacity within 12 months, downside limited to ~20% premium paid for protection.
• Long-dated APLD LEAP call-spread (buy Jan-2027 ATM/30% OTM call, sell Jan-2027 80%+ OTM call) to express asymmetric upside through the multi-year buildout while limiting theta bleed; size 0.5–1% NAV, payoff skewed to capture FCF conversion after 12–24 months.
• Relative value pair: long APLD / short a large colo REIT (e.g., DLR/EQIX) sized neutral to portfolio—this isolates exposure to hyperscaler direct-build wins versus legacy colocation pricing pressure; horizon 12–18 months, close if Applied Digital misses two consecutive utilization milestones.
• Buy INTC (6–12 month horizon) as a diversified play on AI infrastructure demand that is less binary than pure-play colo builders; risk: execution vs AMD/NVDA competition—limit position to 1–2% NAV and reassess after Intel publishes updated datacenter design wins or guidance uplift.