My Top 2 Artificial Intelligence Stocks for 2026

Applied Digital reported fiscal 2026 Q2 revenue of about $127 million, up ~250% year-over-year and well above the $88 million consensus, driven by data-center fit-out services and initial lease revenue recognition for 100 MW deployed for CoreWeave; the company has leases inked for 600 MW across two North Dakota campuses and a 15-year prospective lease revenue pipeline of $11 billion, with management expecting lease revenue acceleration over the next 18–24 months. SoundHound AI's revenue is estimated to have nearly doubled to $172.5 million for the most recent year (midpoint of guidance), and it carries a $1.2 billion bookings backlog (end-2024); analysts' median 12-month price target of $15.50 implies ~32% upside from current levels, supporting a constructive outlook for both companies amid rising AI infrastructure and voice-AI adoption.

Analysis

Market structure: Hyperscalers and AI-specialist infra builders (APLD, server/equipment suppliers like NVDA) are primary beneficiaries as build-to-suit demand shifts share away from legacy colocation REITs; pricing power accrues to firms able to deliver turnkey MW quickly. Supply-side constraints—power hookups, transformers, skilled civil/construction crews—create 6–18 month lead times that support elevated pricing for fit-out services and premium leases. Cross-asset: accelerated AI capex will raise corporate debt issuance (pressuring BBB spreads while boosting demand for copper/transformer-related commodities) and increase idiosyncratic equity vol for exposed names (APLD, SOUN).

Risk assessment: Key tail risks are hyperscaler demand reversals, tenant concentration (CoreWeave is a material counterparty), grid/regulatory constraints, and higher-for-longer rates that raise WACC and construction finance costs. Time horizons: immediate (days) for earnings/guide reactions, short-term (next 18–24 months) for lease revenue ramp and MW come‑online, long-term (3–15 years) for revenue recognition from 15-year pipelines. Hidden dependencies include PPAs, municipal permitting and chip supply; catalysts include new hyperscaler contracts, quarterly lease commencements and bookings conversion.

Trade implications: Direct plays are convex: APLD favored to capture 600+ MW pipeline and potential new hyperscaler, SOUN as asymmetric growth play given $1.2B backlog. Use option structures to limit downside (9–12m call spreads on APLD; 12m LEAP calls on SOUN) and pair trades long AI infra vs short legacy colocation REITs (DLR/EQIX) to express share shift. Entry: accumulate on pullbacks of 15–25% or after confirmed lease revenue starts; exit on missed lease commencements, backlog conversion <20% in 4 quarters, or +40–60% realized gains.

Contrarian angles: Consensus underestimates conversion risk—$1.2B backlog or $11B 15‑yr pipeline are not immediate cash and depend on lengthy buildouts and power availability; APLD’s +50% YTD move may already price in new hyperscaler wins. Historical parallel: 2018–19 cloud capex surges show upside can reverse when overbuild and local constraints manifest; unintended consequences include renegotiated tenant terms and PPA cost shocks that compress returns.

Key Decisions for Investors

• Establish a 2–3% portfolio long in APLD via a 9–12 month bull call spread (limits capital, captures MW ramp in next 18–24 months). Add to size on confirmed lease revenue starts; trim 50% if APLD misses lease revenue guidance by >10% or on a >30% single-session drop.
• Initiate a 1–2% position in SOUN via 12‑month ATM calls/LEAPs to play backlog conversion; take profits at +50% or cut if bookings backlog falls >20% QoQ or revenue guidance misses by >10% in any quarter.
• Implement a pair trade: net 1.5% long APLD vs 1.5% short DLR or EQIX to capture share shift toward build‑to‑suit hyperscaler capacity. Monitor quarterly tenant commencements; close trade within 6–12 months or earlier if hyperscaler contract wins are confirmed/denied.
• Hedge tail risk: buy 6–9 month protective puts equal to ~25–33% of APLD long notional. Reduce net exposure by 50% if 10‑yr Treasury >4.5% or APLD new debt pricing exceeds 600bp over Treasuries, indicating financing stress for buildouts.