Vicor Corporation

▲ Bull Case

• ◆PoPP becomes the industry standard for AI GPU power delivery across NVIDIA Rubin and successor GPUs, AMD MI400+, Google TPU v6/v7, and major custom AI ASICs. Vicor holds ~70% wallet share at $250–350 content per chip; by 2030 AI represents $1.4B of $1.7B total revenue with gross margins sustained at 58%+.
• ◆Second-fab construction comes online on schedule (2028) delivering the additional capacity needed to monetize design wins beyond Fab One's $1.5B ceiling, with the operating leverage of high-IP-content product lines producing 35%+ operating margins at scale and FCF margins of 22–25%—a structural shift from Vicor's historical 18% FCF margin profile.
• ◆Defense compounds at 7–10% as the underappreciated multi-year second engine—DoD modernization, ITAR-protected program stickiness, and the integration of advanced power systems into EW, autonomous, space, and directed-energy platforms adds $50–90M of high-margin revenue independent of AI capex cycle.

▼ Bear Case

• ◆MPWR or Infineon delivers a 'close enough' competing on-package solution at materially lower cost; NVIDIA dual-sources to manage BOM cost; Vicor's $/GPU content compresses 30–40% as the market commoditizes. Revenue still grows but gross margins compress to 50% and operating margins drop to 22–25%; the multiple compresses on slower growth—equity value drops to $80–110/share, a 70%+ drawdown.
• ◆Second-fab CapEx overruns or sits underutilized if design-win momentum stalls in the second-generation refresh, creating a stranded-asset write-down candidate (cumulative $350M+ committed) and depleting the net-cash buffer that has been the balance-sheet anchor of the thesis.
• ◆Vinciarelli succession event with no public plan creates strategic uncertainty at the exact moment when manufacturing scale, second-fab execution, and multi-customer relationship management require continuous decisive leadership; a health event or step-down would likely produce a 20–35% multiple compression even with operations intact.

“Consensus has shifted from 'PoPP is optionality' (May 2025) to 'PoPP is happening' (June 2026, confirmed by Q1 2026 results). The new debate has three layers: (1) How big does PoPP get? Bulls argue NVIDIA + AMD + custom-ASIC = $4–6B addressable Vicor content TAM by 2030; Bears argue captive or commoditized power solutions cap Vicor TAM at $2–3B. (2) What sustained margin survives volume? Bulls model gross margin at 56–58% as IP-content dominates BOM; Bears expect erosion to 50% as MPWR, Infineon, and hyperscaler-captive silicon enter with cheaper GaN solutions. (3) What's the right multiple? Bulls anchor on 50–69x EV/EBITDA (MPWR comparable); Bears anchor on 23–30x (TXN-like specialty semi at scale). The gap implies $200–400/share of outcome difference. Consensus 12-month price target is ~$262.50—meaningfully below current $338.”

• ◆Q2–Q4 2026 revenue beats (sequential acceleration above $120–130M/q) with high magnitude upside
• ◆AMD MI400 series PoPP adoption confirmation (6–18 months, very high magnitude)
• ◆Google/Meta custom AI silicon PoPP adoption (12–24 months, very high magnitude)
• ◆Second-fab groundbreaking + capacity guidance (6–12 months, high magnitude)
• ◆2028+ revenue guide upgrade above $1B (12–24 months, high magnitude)
• ◆Vinciarelli succession announcement (orderly, with named successor) — removes strategic overhang
• ◆Defense program contract awards (Hi-Rel growth, ongoing, medium magnitude)
• ◆MPWR or Infineon announces on-package competing product (12–36 months, very high magnitude negative)
• ◆Hyperscaler captive AI chip power-management announcement (12–36 months, high magnitude negative)

• ◆PoPP pricing erosion via competitive response—CRITICAL severity, 25% probability, 2027–2029
• ◆Vinciarelli succession event without plan—HIGH severity, 15% probability, open-ended
• ◆Customer concentration/hyperscaler captive silicon—HIGH severity, 15% probability, 3–7 years
• ◆Second-fab CapEx overrun or delay—MEDIUM severity, 30% probability, 2026–2028
• ◆Valuation compression on missed quarter—HIGH severity, 35% probability, ongoing
• ◆AI capex cycle correction—MEDIUM severity, 25% probability, 2027–2028
• ◆Patent expiration on core VTM IP—MEDIUM severity, certain, 5–10 years
• ◆Manufacturing gross-margin pressure on revenue stall—MEDIUM severity, 20% probability, 1–3 years