This Under-the-Radar Semiconductor Stock Is Beating Nvidia and Broadcom in 2026

Navitas Semiconductor highlighted a new DC-DC power delivery board that converts 800 volts to 6 volts in one stage, targeting improved efficiency for Nvidia-linked AI data centers. Morgan Stanley raised its price target to $12.50 from $4.20 and Baird to $20 from $9, but the stock trades at 92x sales versus a five-year average of 11.8x, leaving valuation stretched. The article is constructive on the company’s technology but recommends waiting for a pullback before buying.

Analysis

The market is extrapolating a genuine product wedge into a platform-scale winner, but the real economic moat is still unproven. Power-delivery innovation can create a near-term design win cycle, yet in semis the monetization path is usually longer than the narrative: qualification, socket expansion, and volume ramps often lag investor excitement by 2-4 quarters. That makes the current setup more of a sentiment trade on AI capex intensity than a clean fundamentals compounding story.

The second-order implication is that the relevant beneficiaries are not just the company in focus, but the broader ecosystem of rack-scale power, thermal, and interconnect suppliers. If high-voltage conversion architectures gain adoption, it pressure-tests incumbent power-management vendors and creates a “pick-and-shovel” lane for adjacent names with deeper customer relationships and better gross margin durability. The risk is that hyperscalers and GPU-platform vendors push this functionality into reference designs or alternative supplier sets, compressing the premium before it can translate into durable share.

Consensus seems to be missing how reflexive the valuation is: when a stock trades at extreme sales multiples, even successful product execution can be insufficient if growth merely meets elevated expectations rather than accelerating. The setup is therefore fragile to any delay in AI infrastructure orders, margin slippage, or a rotation away from “AI infrastructure” themes over the next 1-2 quarters. On the other hand, if the company can show repeatable design wins with multiple hyperscale customers, the multiple can stay detached from near-term earnings for longer than skeptics expect.

For the listed tickers, the article itself is neutral, but the real signal for us is that the market is rewarding AI infrastructure adjacency more than realized cash flow. That suggests a tougher tape for high-quality financial and exchange names if capital rotates back toward semiconductor beta, but also creates opportunity to fade overowned, high-duration growth names versus proven infrastructure cash generators.

Key Decisions for Investors

• Avoid chasing the crowded AI-infrastructure momentum basket at current levels; wait for a 20-30% drawdown or a clear second-order customer validation event before initiating any long in the theme.
• Pair trade: long a diversified, cash-generative semicap/equipment name versus short the most expensive AI-adjacent power/story stock; target a 3-6 month horizon where execution risk can reassert itself.
• If seeking AI exposure, prefer call spreads over outright longs in high-multiple names to cap downside from multiple compression; use 1-2 quarter expiries tied to next product/earnings catalyst.
• For neutral exposure to the theme, rotate into more established infrastructure beneficiaries with repeat purchase behavior rather than single-product story stocks; prioritize balance-sheet strength and customer concentration under 20%.
• Set an event-driven alert for any announcement of additional hyperscaler design wins or reference-design adoption; only then consider a tactical momentum long with a tight 12-15% stop.