Navitas Semiconductor Is Flashy. This Boring AI Stock Might Make You More Money.

Navitas' shares surged 376% at their 2025 peak (> $17) but were about $9 on Mar 19 (up 23% YTD, +250% over 12 months); analysts project near‑term revenue declines as the company pivots to data centers/EVs/industrial, with a median price target of $8 implying ~9% downside and a rebound expected in 2027 when a Nvidia GaN/SiC data‑center contract begins. IBM grew revenue 8% and adjusted earnings 12% in 2025, expanded gross margin by 170bps to 59.5%, guides ~5% revenue growth and ~$1bn more free cash flow this fiscal year, signed a Nvidia deal for watsonX, acquired Confluent, trades at a ~20 forward P/E with a $340 median target (~36% upside) and yields 2.67% after 27 consecutive years of dividend increases.

Analysis

The shift toward GaN/SiC in high-density power conversion is creating concentrated bottlenecks upstream (substrates, specialized fabs and qualification/test flows) that will reprice economics for small power-IC specialists. Companies that control wafer supply and long lead-time test capacity will capture disproportionate margin and time-to-market advantages; conversely, PMIC incumbents with broad legacy portfolios face margin compression and forced product rationalization as customers consolidate around discrete GaN/SiC stacks.

Execution risk dominates the smaller specialist story: multi-year revenue cliffs are possible if a major OEM qualification slips or if foundry/substrate allocation is tightened. For larger incumbents pivoting into AI/streaming/quantum, the near-term readthrough is adoption cadence — enterprise wins and streaming throughput improvements will meaningfully compress payback periods for hybrid-cloud customers, while quantum remains a structural, multi-year optionality that won’t materially move cash flow in the next 12–36 months.

Practically, this bifurcates opportunities: take measured exposure to durable, cash-generative platforms that accelerate enterprise data movement and AI (low volatility, multi-quarter alpha) while using concentrated hedges to express skepticism on small-cap execution risk in power semis. Tactical option structures allow asymmetric upside to AI exposure while capping downside to idiosyncratic pivot failures in the supply chain over the next 6–18 months.

Key Decisions for Investors

• Pair trade (capital-efficient, 6–18 months): Long IBM via 12-month call spread (buy ATM, sell ~25% OTM) sized 1.5% NAV to capture platform/streaming adoption; fund with a size-0.75% NAV protection leg in NVTS by buying a 9–12 month put spread (tight width). Rationale: retain AI/cloud upside while hedging idiosyncratic qualification/funding risk in a small-cap power vendor. Target asymmetric payoff ~2–3x; max loss = premium paid on both spreads (set hard stop-loss if IBM spread loses 60% of premium).
• Directional AI infrastructure (6–12 months): Buy NVDA call spread (buy 6–12 month near-ATM, sell higher strike) sized 2% NAV to capture broad GPU-led adoption without single-supplier balance-sheet risk. Risk/reward: limit premium outlay, target 3:1 upside if sector re-accelerates; cut if sector breadth narrows or Nvidia misses data-center guide for two consecutive quarters.
• Event-driven small-cap short (3–12 months): For traders unwilling to carry short gamma, buy NVTS OTM puts 6–9 months out sized small (0.5% NAV) to express conviction around pivot/qualification timing risk. Reward: high asymmetry if a large contract delay or funding miss occurs; defined loss = premium, roll/trim after 50% premium erosion.
• Risk-rotation (12+ months): Rotate partial proceeds from realized gains in short-term AI beta into larger-cap enterprise data-streaming and hybrid-cloud names (e.g., NDAQ-sized positions for trade/market-structure exposure or direct IBM equity for stable FCF). Use position sizing to keep total AI/infrastructure exposure below 6% NAV to avoid concentration into execution-risk names.