Vicor Corporation (VICR) makes Vertical Power Delivery (VPD) modules for AI accelerator and hyperscaler ASIC power delivery. The company historically served NVIDIA through the A100 generation but was displaced on H100 and Blackwell (MPWR, Renesas, and Infineon now serve NVIDIA). The current lead customer is undisclosed but speculated to be a hyperscaler ASIC program. Q1 2026 printed this morning (8-K filed 2026-04-21) with a headline $301M backlog (+70% sequential) that will gap the stock up at tomorrow's open. The same filing settled a $28.6M Federal-Circuit-affirmed willful patent infringement judgment that the CEO's letter did not mention. Both items are primary-source verifiable. Only one will be in the headline.

What the filing says

Q1 2026 results (8-K 2026-04-21, Exhibit 99.1):

• Total revenue $113.0M (+20.2% YoY). Product revenue $98.0M — narrow miss on $100M. Royalty revenue $15.0M (+39.1% YoY), annualized run rate ≈$60M.
• Gross margin 55.2% (+800bps YoY). Product revenue +$14.8M YoY on only +$1.0M incremental COGS — 93% incremental gross margin. Fab fixed-cost leverage inflecting.
• Backlog $301M, +70% sequential from $177M Q4 2025 ($124M single-quarter jump on $113M revenue). Backlog-to-revenue ratio moved 1.65x → 2.7x.
• Capex $12.4M (+170% YoY). CEO: "planning a second fab... alternate source of high current density 2 Gen VPD modules will give OEMs and Hyper-scalers redundant access."
• Cash flow from operations −$3.9M, which includes $28.6M "payment of an award for past litigation." Balance sheet confirms: accrued litigation $28.3M → $0. Ex-payment OCF was +$24.7M.

What the market thinks

VICR closed $162.67 on Apr 18 (already −19% from the Feb ATH of $210 on coordinated insider selling). We estimate the Apr 22 open near $185-195 on the backlog headline. At $190 reference, implied scenario weights approximate bull 40-45% / base 40% / bear 15-20% — pricing the ramp thesis near fair value with a modest bull tilt.

Our decomposition, same $190 anchor: bull 35% / base 40% / bear 25%. Forward 12-month EV ≈ −5.6%, Sharpe ≈ −0.14. Edge is ≈5-10pp more weight on bear scenarios than market — directionally consistent with "moat narrowing in parallel with demand accelerating." Small in magnitude.

Why the gap exists

Two primary-source findings the headline will skip:

SynQor. The $28.6M is not a settlement — it is a Federal-Circuit-affirmed willful-infringement judgment against VICR as defendant in isolated DC-DC converter IP. Timeline: E.D. Texas jury verdict 2022-10-26; Amended Corrected Final Judgment 2024-05-20 (≈$26M: $6.5M damages + $4.5M enhanced for willfulness + $9.5M attorney fees + $5.4M prejudgment interest + $87K costs); Federal Circuit AFFIRMED 2026-02-13. The CEO's Q1 letter, published 67 days after the affirmance, claims "precluding unlawful importation of computing systems infringing Vicor IP is having an effect" — zero acknowledgement that the company lost an affirmed willful-infringement appeal as defendant on the same quarterly timeline. The infringed patent is not VPD, so the VPD portfolio is not invalidated. The narrative cost: VICR is simultaneously an IP enforcer (ITC v. Delta, Foxconn, Luxshare, MetaPWR) and an adjudicated willful infringer.

Three-competitor VPD landscape. MPWR Q3 2025 and Q4 2025 calls (2025-10-30 and 2026-02-05): CEO Michael Hsing cited "modules vertical power delivery custom ASIC ramp pretty big tailwind" — first material MPWR VPD call-out, explicit hyperscaler territory. Joins ADI (shipping Q4 2025, described market as "knee of curve") and TDK (stackable µPOL launched Feb 2026). Three well-funded competitors now in production on what VICR still frames as monopoly VPD. The "alternate source... redundant access" language in the Q1 letter is demand-pull from hyperscalers — but the same demand shape is consistent with customers hoarding VICR supply while qualifying alternatives.

Risks (ranked by impact)

1. Q1 backlog contains pull-forward. Discriminated by Q2: if end-of-Q2 backlog falls below $275M, the $124M sequential jump was one-time commitment, not recurring order flow.
2. Lead customer identity undisclosed. "Plural Hyper-scalers" language + supply-redundancy framing suggests strategic relationships, but single-customer concentration is still possible.
3. Multiple at 75-90x forward EPS requires FY2027 revenue and margin expansion to materialize. Any ramp slippage compresses rapidly.
4. Insider 10b5-1 selling active. 6+ insiders sold ≈$14.7M Feb-Mar at $195-210. CEO's 1M share plan could liquidate $150-200M at recent prices. Persistent supply headwind.

Catalysts

• Apr 21, 8:00 AM ET (today, already held): Q1 2026 earnings call. Transcript review pending. Live risk: did any analyst ask about the $28.6M SynQor payment, and how did the CEO characterize it?
• Apr 22 open: price discovery on $301M headline.
• Early May: MPWR Q1 2026 earnings. VPD language progression is cross-read.
• Late May: ADI Q2 FY26 earnings. VPD customer disclosure is the single highest-edge adjacent data point. If ADI names a hyperscaler that overlaps with VICR's implied lead, "share split" becomes verifiable.
• Late July: VICR Q2 2026 earnings. Backlog persistence test. Q2 product rev >$100M and backlog stays >$275M are the two levels that distinguish ramp from pull-forward.
• Dec 31, 2026: Formal second fab capex commitment and VPD monopoly erosion are the year-end tests.

What would change our mind

• Bull harder: Q2 backlog stays ≥$300M and VICR discloses a named multi-year committed hyperscaler. Kills the hoarding read.
• Bear harder: ADI Q2 FY26 names the same hyperscaler as VICR's implied lead customer. Confirms share split, not TAM expansion.
• Stay ambiguous: Q2 backlog $250-280M. Demand real with some pull-forward. Current scenario weights hold.

Evidence