Better AI Chip Stock: Nvidia vs. Navitas Semiconductor

Nvidia remains the dominant data-center GPU supplier (over 90% of discrete GPU market) and is projected to grow revenue and EPS at CAGRs of ~47% and ~45% from fiscal 2025–2028, trading at roughly 25x projected fiscal 2027 EPS and ~13x projected sales. Navitas, whose stock has rallied ~165% in the past year on a supply deal to provide GaN and SiC power chips for Nvidia data centers, faces near-term revenue declines (analysts expect -45% in 2025 and -21% in 2026 before +84% in 2027) and remains unprofitable with a valuation of ~63x trailing sales; mass production for the Nvidia partnership is not expected until 2027. The piece concludes Nvidia is the more attractive risk-reward today due to scale, moat (CUDA ecosystem), faster growth and cheaper fundamentals, while Navitas’s upside hinges on successful 2027 ramp and resolution of inventory/supply-chain headwinds.

Analysis

Market structure: NVDA is the clear short-to-medium-term winner — >90% discrete GPU share and CUDA lock-in give it pricing power and a durable moat; NVTS is a niche supplier whose Nvidia OEM win is binary (mass production in 2027) and has been bid to a 63x sales multiple. Expect GPU supply tightness and strong data-center pricing through 2026, supporting NVDA margins; GaN/SiC raw-material demand lifts specialty semiconductor suppliers and certain commodity inputs (SiC substrates, specialty gases). FX and rates: stronger USD and a risk-on AI rally pressure long-duration tech multiples, pushing 10y yields slightly higher if capex expectations accelerate.

Risk assessment: Tail risks include expanded US export controls on advanced nodes or GPUs to China, an NVTS ramp failure or single-customer dependency, and a demand pullback for AI capex (20–30% downside to consensus) if macro weakens. Timeframes: immediate (days) – NVDA guidance/earnings and option IV spikes; short-term (weeks–months) – inventory digestion at EV/solar customers hitting NVTS; long-term (2027+) – NVDA Rubin/Ferryman and NVTS production ramp. Hidden dependencies: foundry capacity (TSMC, GF) and power-architecture adoption by hyperscalers; catalysts are NVDA product launches, Navitas manufacturing milestones, and trade-policy updates.

Trade implications: Primary direct play is overweight NVDA (fundamental/valuation gap) and avoid/short NVTS until 2027 ramp confirmation; consider sector rotation into semiconductor capital equipment and power-semiconductor suppliers (e.g., NXPI, ON) that benefit from GaN/SiC growth. Options: use 9–18 month NVDA call spreads to gain upside while capping cost; for NVTS use put spreads or very small long-dated call punts tied to production proofs. Entry/exit: accumulate NVDA into any pullbacks of 10–20%; reduce/close NVTS exposure if the company misses a 2027 qualification milestone by Q4 2026.

Contrarian angles: Consensus understates two things — (1) the probability hyperscalers accelerate in‑house accelerators over 3–5 years, capping NVDA upside, and (2) the opposite risk that NVTS becomes strategic and accelerates production before 2027, which would re-rate its multiple quickly. The market may be overpaying for pre-revenue optionality in NVTS (valuation implies flawless 2027 execution); a simple threshold to watch: if NVTS revenue run-rate >$200m by end-2027, re‑rate risk reduces materially. Unintended consequence: Nvidia’s dependence on a single external GaN/SiC supplier could prompt vertical integration or acquisition, a binary corporate event that would swing NVTS dramatically.

Key Decisions for Investors

• Establish a 3–5% portfolio long in NVDA (ticker NVDA) with a 12-month horizon; hedge cost by buying a 12-month call spread (buy 20% OTM, sell 50% OTM) sized to the 3–5% position and plan to trim if NVDA forward P/E >30 or price rallies >30% from entry.
• Initiate a tactical 1–2% short or put-spread on NVTS (ticker NVTS): buy a 9–12 month 30% OTM put and sell a deeper 50% OTM put, or short outright if liquidity allows; cover/close if NVTS confirms >$200M annualized revenue run-rate by Dec 31, 2027, or if market cap/sales falls below 20x.
• Execute a pair trade: dollar-neutral long NVDA / short NVTS at 1–2% each to capture relative-value; rebalance monthly and widen stops to 15% on either leg for risk control.
• Overweight semiconductor-capital and power-semiconductor suppliers (examples: NXPI, ON) by 1–3% as a thematic play on GaN/SiC adoption; reduce cyclic EV/solar supplier exposure by 2–4% until Navitas end-market inventories show stabilization (watch inventory-to-sales ratios quarterly).
• Set event-driven rules: reduce NVDA exposure by 50% if next-quarter data-center revenue misses consensus by >5%; increase NVTS short size if Nvidia publicly delays mass-production beyond H1 2027 or if Navitas misses a validated foundry qualification by Q4 2026.