Why Some Analysts Believe Artificial Intelligence (AI) Winners Will Look Very Different This Year

Analysts argue the next phase of the AI cycle will shift from AI chips and software to the infrastructure 'backbone' powering deployments, highlighting Texas Instruments' analog chips and data-center-related revenue (data center sales grew 70% in 2025). Energy/infrastructure plays cited include Bloom Energy, which supplies hydrogen power cells and electrolyzers and has signed deals with American Electric Power and Brookfield Asset Management, and Brookfield Renewable, which supplies Microsoft and Google and whose partnership class yields 5.2%. The note frames a multi-year investment opportunity in analog semiconductors, on-site/clean power and renewable providers as AI deployments scale.

Analysis

Market structure is shifting from pure AI compute/software winners (NVDA, SoundHound) toward the physical AI backbone: analog chips, on-site power and renewable supply. TXN’s data-center sales (+70% in 2025) signals multi-year demand for power-management analogs, boosting pricing power for suppliers and accelerating data-center capex, which should lift copper, transformer, and power-electronics demand by mid-2026. Faster onsite power (Bloom Energy) and contracted renewable supply (Brookfield Renewable) gain share versus incumbent utilities that face long interconnection lead times.

Key risks include an AI growth pause, grid/permitting bottlenecks, hydrogen/NG feedstock price spikes, and execution failure on BE project rollouts; regulatory changes to hydrogen subsidies or clean-power procurement would be binary. Time horizons: immediate (earnings/contract announcements in next 30–90 days), short-term (6–12 months for order flow and LEAP repricing), long-term (2–5 years for infrastructure rollouts and contracted revenue). Hidden dependencies: interconnection queues, EPC labor availability, and TI’s analog supply chain constraints could produce supply squeezes.

Trade implications: favor durable cash-flow/asset owners (BEP/BEPC) and industrial chipmakers (TXN) while selectively allocating to BE for asymmetric upside in fast-deploy power. Use income overlays (covered calls on BEP) and directional LEAPs on BE to capture optionality; avoid unhedged small-cap AI software exposure. Catalysts to watch: TXN quarterly data-center revenue cadence, BE large-scale utility/enterprise contracts, DOE hydrogen policy updates.

Contrarian view: the market underestimates the value of “connective” hardware and energy contracts — the highest alpha may come from infrastructure owners, not model makers. Historical parallel: post-dotcom telecom backbone winners outperformed many application-layer names; unintended consequence is commodity-driven margin pressure for installers, which can compress returns if not hedged.

Key Decisions for Investors

• Establish a 2–3% long position in Texas Instruments (TXN) within the next 2 weeks for a 6–12 month horizon to capture analog/data-center share gains; size a hedge by buying 6–9 month puts 5% OTM not exceeding 1% portfolio cost to limit downside.
• Allocate 1.5% of portfolio to Bloom Energy (BE) via 18–36 month LEAP calls (e.g., Jan 2028 expiries) or shares if calls illiquid; target >50% upside in 24 months and plan to sell 33% if BE fails to announce ≥$50M in new contracts within 6 months.
• Buy 3% in Brookfield Renewable (BEP/BEPC) to capture ~5.2% yield; implement covered-call overlay (sell 3–4 month calls 5–10% OTM) to boost income and hold 24+ months, trimming if NAV falls >15% or counterparty losses emerge.
• Enter a pair trade: long BEP 2% notional and short AEP 1.5% notional for 6–18 months to express faster renewable/contract advantage; unwind if relative spread narrows to <100bps or if BEP declines >15% absolute.
• Reduce speculative AI software small-cap exposure (e.g., trim SOUNW by ~50%) and reallocate proceeds to TXN/BEP/BE within 30 days to reflect rotation from model-layer to infrastructure-layer investments.