Hood River Trims Applied Digital Equity Holding

Hood River Capital trimmed its stake in Applied Digital (NASDAQ: APLD) by 812,377 shares in Q3, leaving a post-trade holding of 21,046,654 shares valued at $482.8 million (5.7% of 13F AUM) as of Sept. 30. Applied Digital, which provides data-center and GPU/HPC infrastructure for AI and crypto workloads, reported TTM revenue of $219.02 million and a TTM net loss of $225.71 million; Q3 revenue was $64.2 million (+84.3% YoY) with an adjusted net loss from continuing operations of $0.03. The stock has rallied sharply (price $23.06 as of Nov. 13, up ~220% year-over-year), so the modest trim looks like profit-taking by the fund rather than a shift in conviction.

Analysis

Market structure: Hood River’s 3.7% trim of APLD is classic profit-taking, not a conviction sell—it still holds 21.0M shares (5.7% of 13F AUM). Direct beneficiaries: GPU vendors, AI-focused colo and turnkey build contractors that can capture follow-on spend; losers: legacy data‑center REITs if capital shifts to specialized HPC players. The move signals continued demand for AI/HPC capacity but also growing concentration risk in a few liquid names, which amplifies price moves on large block trades.

Risk assessment: Tail risks include a regulatory clamp on crypto mining or local power curtailments that could reduce utilization (low-probability, high-impact) and a financing/dilution event if cash burn exceeds runway; payout sensitivity is high given TTM revenue ~$219M vs market cap $6.34B. Near-term (days–weeks) expect volatility around 13F/earnings windows; medium-term (3–12 months) hinge on GPU supply and new data center ramp; long-term (1–3 years) depends on sustained AI workload growth >40% YoY to justify current multiples.

Trade implications: Favor tactical long exposure to APLD on defined dips (see decisions), paired with shorts in slower-growth data‑center REITs to hedge macro. Use options to buy time (LEAPs) or sell premium against positions if implied vol spikes; position size should be modest (1–3% AUM) because of high convexity and potential dilution. Rotate into suppliers (NVIDIA, selected power/transformer contractors) and away from cyclical legacy hosting if conviction in AI secular demand >2 years.

Contrarian angles: Consensus treats Hood River’s trim as profit-taking; it understates concentration risk — the fund still owning 5.7% AUM means selling pressure could be front-loaded in liquidity events. Market may underprice regulatory/power constraints and overprice near-term revenue conversion; historical parallels include crypto‑cycle infrastructure names that doubled then collapsed when demand normalized. A disciplined entry with explicit stop-loss and tranche-based buys will exploit mispricings if AI demand disappoints.

Key Decisions for Investors

• Establish a tactical 1.5% long position in APLD on a pullback to $16–$18 (≈30%–35% below 11/13 close); set a hard stop-loss at 15% below entry and target 40%–60% upside within 12 months contingent on continued >50% YoY revenue growth.
• If bullish longer-term, buy APLD Jan 2027 LEAP calls (strike $20) sized to 1% AUM as a directional asymmetric bet; alternative: sell cash-secured $18 puts exp Mar 2026 for net yield if willing to acquire stock at that level (collect premium and only buy if assigned).
• Implement a pair trade: long APLD (1%) vs short Digital Realty (DLR) (0.8%) to express growth-over-real-estate exposure; rebalance if spread contracts >20% or after next APLD quarterly report.
• Reduce exposure to broad data-center REITs by 1% AUM and rotate proceeds into suppliers/contractors (e.g., NVIDIA, MasTec MTZ) over 30–90 days to capture capex flow; increase cash if GPU supply indicators (inventory reports, NVIDIA guidance) weaken over next 60 days.
• Monitor four high-leverage catalysts over the next 90 days before adding size: APLD quarterly EBITDA conversion, GPU spot pricing trends, local permitting/power procurement outcomes for APLD projects, and any SEC/legislative action on crypto energy use — add no more than 1% AUM per positive catalyst.