Navitas Semiconductor at Citi’s 2026 Global TMT Conference: pivot to ai power

Navitas said its pivot away from mobile chips is largely complete, with mobile declining from roughly 85% of sales to a low-single-digit share by year-end and AI infrastructure expected to exceed one-third of revenue. Management cited backlog visibility into 2027 and expects the planned October 2026 acquisition of Claros Power to roughly double its serviceable market opportunity to $8B-$9B, although Claros-related revenue is expected to become meaningful only in 2028-2029. The company is targeting multiyear demand from the transition to 800-volt AI data-center power systems and grid modernization, but execution risk remains significant given trailing-12-month revenue fell 46% to $36.53M and shares have risen 110% over the past year to $12.44.

Analysis

NVTS is attempting to earn a data-center power multiple before its financial model has demonstrated data-center scale. The near-term monetization sits primarily in SiC power-conversion content, where established suppliers including IFX and onsemi retain qualification, reliability-data, and channel advantages; GaN and IVR are later-cycle opportunities whose meaningful ramps depend on architecture choices not yet locked across hyperscaler fleets. The key non-obvious constraint is that merchant power-system vendors—not the hyperscalers alone—control component qualification and carry the warranty risk, making design-win announcements materially more valuable than broad AI-exposure rhetoric.

The Claros transaction increases strategic optionality but introduces execution risk: integrating a new process-node supply chain and proving an array-based IVR at extreme current levels are separate hurdles from closing the deal. A fabless model should support cash efficiency, but it also leaves NVTS exposed to foundry allocation, yield, and qualification slippage precisely as customers demand multi-year supply assurance. Over the next 1-3 months, the stock needs revenue guidance and gross-margin evidence rather than additional TAM framing; absent that, a premium valuation on a still-small revenue base leaves it vulnerable to a sharp de-rating on even modest schedule delays.

Contrarian view: the staged transition supports a broader power-infrastructure basket more reliably than a concentrated NVTS bet. VRT, ETN, and FLEX can monetize rising rack power density and electrical-system complexity regardless of which semiconductor material wins each conversion stage, while ADI has a more credible route into point-of-load integration following its power portfolio expansion. NVTS upside is real if it converts backlog into repeatable revenue, but the market appears to be capitalizing 2028-29 IVR economics ahead of independently verifiable customer deployment.

Key Decisions for Investors

• Do not chase NVTS following the rerating; place on an earnings-alert watchlist for two consecutive quarters of AI-infrastructure revenue growth, stable-or-improving gross margin, and explicit 2027 revenue/backlog conversion disclosure. Failure to provide quantified conversion metrics or a reduction in forward outlook is the thesis falsifier.
• Prefer a 6-12 month long VRT / short NVTS relative-value position, sized modestly: VRT has direct system-level exposure to higher-density deployment, while NVTS bears technology qualification and valuation risk. Target 15-20% relative outperformance; stop if NVTS discloses a named hyperscaler production ramp with material 2027 revenue contribution or VRT guides to a data-center order slowdown.
• Accumulate ETN or FLEX on broad AI-infrastructure pullbacks for a 12-18 month horizon. Both benefit from rising electrical content and power-system complexity without requiring native 800-volt adoption on a specific timetable; reassess if hyperscaler capex guidance rolls over materially.
• Use IFX as the more defensive SiC exposure over the next 6-18 months: early high-power conversion adoption can favor incumbent reliability and supply-chain credentials before NVTS's GaN/IVR upside is commercially proven. Exit the relative preference if NVTS demonstrates sustained design-win conversion and IFX signals margin pressure from data-center pricing competition.
• Treat the Claros close as an event-risk checkpoint rather than a catalyst to buy: monitor transaction terms, retention of key technical personnel, Samsung qualification progress, and any disclosed customer sampling. A delayed close, revised economics, or lack of 2028 customer specificity would increase downside risk in NVTS.