Vicor Gaining on Search for AI Power

Vicor reported record quarterly revenue of $113 million, up 20.2% year over year, with GAAP diluted EPS of $0.44 and backlog of $300.6 million. The company guided 2026 revenue to nearly $570 million, while EPS is estimated to rise 101.7% this year. The article emphasizes strong institutional buying and shares are up 179% year to date, nearly 250% since the July 2023 outlier signal.

Analysis

VICR is one of the cleaner ways to express the AI infrastructure spend trade without owning the obvious hyperscaler beneficiaries. The second-order edge is that power-delivery bottlenecks tend to monetize later in the cycle than compute orders, so a supplier with visible backlog and rising mix can keep growing even if headline AI capex slows. That makes VICR more of a “picks-and-shovels on the power side” beneficiary than a pure semiconductor beta name.

The market may still be underestimating how sticky the demand can be if this is embedded in data-center rack redesign rather than a one-off procurement burst. Once OEMs and integrators qualify a power architecture, switching costs rise materially because thermal, reliability, and uptime certifications become part of the design win, which can extend revenue durability for multiple quarters. That said, the stock’s move has likely pulled forward a lot of good news, so the next leg is more likely to depend on margin mix and conversion of backlog than on another top-line surprise.

The main risk is that consensus treats backlog as near-certain revenue, when in power hardware the timing and mix can slip if customer deployments are deferred or reconfigured. A second-order negative is that a sustained run-up can invite supply-chain and channel normalization: if customers over-order to secure capacity, the stock can look extended before the actual demand curve cools. In the near term, the catalyst path is 1-2 quarters of beats/raises; over 6-12 months, the question is whether VICR can prove this is a durable platform shift rather than a cyclical AI ordering wave.

The contrarian read is that the market may already be paying growth-stock multiples for something that still has execution and concentration risk. If investor positioning is as crowded as the flow data implies, the asymmetric trade may now be in buying dips rather than chasing strength, especially if the next print is merely in line. Any disappointment in backlog conversion, gross margin, or guidance cadence would likely hit harder than a normal miss because expectations have shifted from re-rating to proof-of-sustainability.

Key Decisions for Investors

• Long VICR on a pullback rather than momentum chase; prefer entry after a 5-8% retracement or post-earnings consolidation. Thesis: AI power-content scarcity should support multi-quarter demand, but valuation risk is elevated after a large run.
• Use a call spread instead of outright equity for 3-6 months (e.g., at-the-money / +20% strike) to express upside while capping multiple-compression risk if growth merely meets expectations.
• Pair trade: long VICR / short a basket of over-owned AI compute beneficiaries with less visible backlog conversion. Goal is to isolate the power-infrastructure spend rather than broad AI sentiment beta.
• If you already own VICR, take 25-33% profit on any post-earnings gap-up and trail the rest with a 10-12% stop; the risk/reward shifts quickly once positioning becomes consensus.
• Monitor the next two quarters for backlog conversion and gross margin expansion; if revenue beats but backlog growth decelerates, reduce exposure because the market will likely re-rate from growth story to execution story.