The AI Infrastructure Boom Is Just Getting Started. Here Are 2 Stocks to Buy.

Fluence Energy, a 2018 Siemens-AES JV, is positioning its industrial-scale battery modules as a reliability solution for AI data centers, with 36 GWh of data-center projects under development and 46 GWh total installed capacity as of September; analysts project annual sales could rise ~57% within two years despite flat-to-down 2025 revenues and no signed data-center contracts yet. Credo Technology’s Active Electrical Cables deliver up to 0.8 Tbps (and double that in fourth‑generation parts) over copper with lower power draw, fitting dense AI data centers; the company reported preliminary Q3 2026 revenues near $406 million versus $135 million a year earlier, though FY2025 revenue concentration was high (one client = 67%).

Analysis

Winners are modular-battery suppliers (FLNC), low-power high-speed interconnect specialists (CRDO), and GPU incumbents (NVDA) as hyperscalers pour ~ $1T into data centers; losers include long-reach fiber-transceiver incumbents and diesel/backup-only genset suppliers whose economics deteriorate as batteries displace peaker assets. Pricing power will shift to vendors that cut OPEX (power + cooling) per rack; expect incremental margin expansion for hyperscalers and concentrated supplier rents for patented AEC/serializer IP.

Key tail risks: CRDO’s 67% revenue concentration creates a single-client cliff risk; FLNC’s 36 GWh pipeline has execution and contract-recognition risk — none are in backlog today. Time horizons: expect stock moves on near-term (days–weeks) earnings/backlog updates, medium-term (3–12 months) backlog-to-revenue conversion, and structural adoption over 1–3 years driving the projected +57% sales for FLNC. Hidden dependencies include raw-material pressure (lithium, copper) and hyperscaler capex cadence; catalysts are signed multi-GWh contracts, hyperscaler capex guidance, and major OEM endorsements.

Trade implications: tactical long CRDO exposure captures reported 3x revenue growth momentum but size to 2–3% equity risk due to concentration; FLNC is a longer-duration optionality play—favor 12–24 month LEAP calls or a buy-and-collar with rollout milestones. Relative-value: long NVDA vs short INTC expresses GPU-led compute share gains; options: use 90-day call spreads on CRDO around earnings to cap premium and collars on FLNC to limit drawdown. Rotate overweight to data-center infra and AI semis, underweight legacy fiber/transceiver suppliers.

Contrarian angles: consensus underestimates timing friction — FLNC upside is binary on contract signing rather than smooth conversion, so implied premium may be overdone now. CRDO’s scale could invite low-cost competition or royalty disputes; historical parallel: Mellanox’s rapid adoption then margin compression post-acquisition. Unintended consequence — lower cooling OPEX can enable denser racks that stress PDUs and create a secondary market for power distribution — another investment angle to monitor.

Key Decisions for Investors

• Establish a 2–3% long position in CRDO ahead of Q3 2026 earnings using a 90-day call spread (buy ATM, sell 25% OTM) to express upside while limiting premium; set a hard stop-loss if share price falls 20% or if the largest customer contribution remains >60% after earnings.
• Allocate 1–2% as long-dated optionality in FLNC via 12–24 month LEAP calls or purchase-stock + 12-month collar (sell 20% OTM calls, buy 25% OTM puts) and scale into equity only after FLNC converts ≥10% (≥3.6 GWh) of the 36 GWh pipeline into signed contracts within 6 months.
• Implement a 1–2% pair trade: long NVDA (1–2%) and short INTC (1%) to capture AI compute share shift over the next 6–12 months; trim if NVDA outperforms by >30% or if hyperscaler capex guidance weakens materially.
• Underweight fiber-optic transceiver and legacy genset suppliers by reducing exposure 50% in the data-center hardware sleeve; re-evaluate if copper prices climb >15% YoY or if CRDO loses >15% share of its hyperscaler revenue mix within 12 months.
• Monitor specific triggers for position review: signed FLNC data-center contracts (target ≥3.6 GWh in 6 months), CRDO customer concentration dropping below 50% in 12 months, and LME copper or lithium carbonate moves ±15% which should prompt re-sizing within 30 days.