Better Artificial Intelligence Stock: Applied Digital vs. CoreWeave

Demand for AI compute is driving rapid expansion in neoclouds, with Applied Digital reporting fiscal Q2 revenue of $126.6M (up ~250% YoY), cash and equivalents of $1.9B, debt in excess of $2.6B and an operating loss of $31M, while CoreWeave posted record Q3 revenue of $1.4B (from $583.9M) with roughly $368.8M cost of revenue and nearly $1B gross profit but has amassed more than $14B of debt. CoreWeave benefits from a $6.3B partnership with and a $2B equity investment from Nvidia (including a purchase obligation for residual capacity through April 13, 2032), trades at a materially lower forward price-to-sales multiple than Applied Digital, and is presented as the preferable investment given higher sales, stronger gross margins and superior valuation. Applied Digital’s heavy buildout-related leverage and thin gross margins are the principal downside risks highlighted.

Analysis

Market structure: Neoclouds (CRWV) and GPU vendors (NVDA) are the primary winners as AI training demand outstrips bespoke hyperscaler capacity; landlords (APLD) benefit from leaseback economics but show weaker unit economics. CoreWeave’s $1.4B revenue and ~$1B gross profit in Q3 vs Applied’s $126.6M revenue and $26M gross profit imply the value accrues to operators with scale and preferred OEM relationships (Nvidia’s $6.3B pact and $2B equity). Supply/demand is tight for high-performance racks and GPUs through 2026–2032 (Fortune forecast $59B→$356B), signaling persistent capex and constrained GPU spot markets — upward price power for rented AI compute in the near term.

Risk assessment: Key tail risks include a sudden Nvidia supply disruption or export-control shock (months), a rapid rate-driven liquidity squeeze that raises debt servicing costs (90–180 days), and power/regulatory permitting bottlenecks for data-center builds (6–24 months). Hidden dependencies: CoreWeave’s apparent safety via Nvidia demand purchase commitment (through Apr 13, 2032) masks counterparty concentration and collateral obligations; Applied’s $2.6B debt vs $1.9B cash leaves modest liquidity cushion if capex or lease commencements slip.

Trade implications: Favor long CRWV and NVDA exposure with protective sizing and hedge credit/interest risk; implement a pair trade long CRWV / short APLD to capture operational quality and valuation dispersion (use 12–18 month horizon). Option structures: buy CRWV 12-month LEAP calls or call spreads to limit capital and sell OTM calls on NVDA to finance cost; short-duration shorts on APLD equity or buy CDS if available to express downside on leverage sensitivity.

Contrarian angles: Consensus underestimates counterparty and power risks — if GPU supply normalizes or Nvidia prioritizes internal use, scramble risk grows and margins compress. The market may be underpricing Applied’s landlord durability under long-term lease rolls if tenants default; a stressed-credit scenario could flip value to landlords with long-term contracted cash flows. Historical parallel: 2010s datacenter cycles where landlords lagged operators until lease roll-ups proved resilient — monitor lease commencement vs. capex cadence closely.

Key Decisions for Investors

• Establish a 2–3% portfolio long position in CRWV within 30 days (or buy 12–18 month LEAP calls equal to ~2% notional) citing superior gross margins and Nvidia support; set a tactical stop-loss at -25% and reassess at Q1/Q2 2026 earnings or upon any Nvidia capacity purchase notice.
• Initiate a dollar-neutral pair trade: long CRWV (2% notional) and short APLD (1–1.5% notional) within 60 days to exploit P/S multiple divergence and margin differential; cover the short if APLD falls >30% or if gross margins improve sequentially for two quarters.
• Buy 1–2% long NVDA exposure (shares or 9–12 month calls) as a hedge and asymmetric upside play given its supply/partnership role; finance via selling 3–6 month OTM calls (covered) to collect premium and lower basis.
• Monitor credit and macro triggers: trim levered AI infra exposure if 10-year Treasury yield rises >50bps within 30 days or if APLD/CRWV 5-year CDS widens by >200bps; require no-covenant-breach proof points from both companies within next 90 days before adding leverage.