The Artificial Intelligence (AI) Stocks I'm Watching Closest in March

CoreWeave posted $5.1B revenue in 2025 and is guiding $12.0–$13.0B for 2026, secured a multiyear deal with Perplexity and a minimum $6.3B arrangement with Nvidia, but recorded a ~$1.2B net loss in 2025 and carries >$20B of debt. TTM Technologies’ FY2025 revenue rose 19% to $2.9B, net income more than doubled to $177.4M, and management expects 15–20% sales growth in FY2026 with Q1 data-center sales forecast +66% YoY. Fastly reached record revenue of $624M in 2025 (from $543.7M), projects $700–720M for 2026 and reduced its net loss to $121.7M, though ~1/3 of revenue remains concentrated in its top 10 customers.

Analysis

CoreWeave’s model leverages third‑party GPU flow and scale but embeds a classic leverage mismatch: operating cash burn and fixed real‑estate/power commitments front‑load risk while revenue remains usage‑sensitive. The practical effect is that any micro‑cycle weakening in AI training demand or a step‑up in power costs will compress coverage quickly; credit markets will reprice this within 6–18 months because growth alone won’t substitute for negative free cash flow.

TTM sits in the opposite quadrant — higher cash conversion and time‑to‑market specialization give it optionality to capture margin from OEMs shortening supply chains and bumping safety inventory. Expect the secular shift to edge/data‑center hardware to favor PCB suppliers that can promise sub‑quarter turnarounds and integrated RF/assembly services; suppliers of laminates, testing, and high‑end copper-clad materials are second‑order beneficiaries.

Fastly’s edge/CDN positioning is structurally relevant to AI traffic patterns, but customer concentration and valuation expansion create a fragile reply to any churn or contract renegotiations; a modest increase in customer attrition or a 10–20% multiple compression would erase recent gains. Across the group, the clearest systemic risk is power and chip supply volatility — think regional outages, spot GPU shortages, or an OEM prioritizing captive cloud customers — which act as binary catalysts on 3–24 month horizons.

Key Decisions for Investors

• Long TTMI (6–12 months): initiate a 3% NAV position, add to 5% NAV on a 10% pullback. Rationale: cash conversion + order book visibility; target 35–50% upside in 12 months vs a -15% stop. Pair with sector hedges (SMH put) if semicap weakness emerges.
• Pair trade — Long TTMI / Short CRWV (12 months, equal‑dollar): size 2% NAV each. Objective: capture dispersion between margin discipline (TTMI) and balance‑sheet leverage (CRWV). Risk: CRWV outperforms if enterprise spot demand reaccelerates; capped loss = 2% NAV on divergent moves.
• FSLY asymmetric hedge (3–9 months): buy a defined‑risk put spread (e.g., buy 1x FSLY 6–9 month OTM put, sell a lower strike put) sized to 0.5–1% NAV to monetize customer‑concentration downside while limiting premium. Reward: protects against 20–40% downside with limited premium paid.
• Credit/volatility hedge for CoreWeave (12–24 months): purchase IG/HY protection tied to data‑center REITs or power utilities exposed to AI capex (or buy NVDA put spreads only as short‑dated volatility hedge) if GPU spot rents decline — allocate 0.5–1% NAV to options/credit to limit tail exposure.