The question isn't whether this is a good company. It's whether the margin inflection ahead of it is priced.

The Setup

CPSH had a bad 2024. A quality issue consumed production capacity, revenue dropped from $27.6M to $21.1M, gross margins went negative. They added a third manufacturing shift in August 2024. Revenue recovered 54% to $32.6M in FY2025. But gross margins only recovered to 16.2% — well below the 25-27% they ran in 2022-2023.

The third shift still isn't at full proficiency. Fifteen months in, Q4 2025 posted an operating loss of -$0.1M despite 38% year-over-year revenue growth. The market saw "still not profitable" and sold the stock from $6.85 to $3.81.

Here's what the market missed: CEO Mackey warned on the Q3 call that Q4 would be weaker due to holidays. He literally said "unlikely to achieve record revenue." The sequential decline wasn't a surprise — it was a calendar effect the CEO flagged 90 days in advance. The -25% selloff punished expected seasonality.

On the Q4 call, Mackey called the company at an "inflection point" and said they have "further to go on both revenue and gross margins." This isn't hedging language. He's telling you the margin improvement is forward.

The Math

Operating leverage at CPSH is steep because the cost base is small:

Fixed costs: ≈$5.4M (SG&A $4.8M + D&A $0.6M)
Current:     $32.6M revenue × 16.2% GM = $5.3M gross profit ≈ breakeven
At 20% GM:   $35M × 20% = $7.0M GP - $5.4M = $1.6M operating income
At 22% GM:   $38M × 22% = $8.4M GP - $5.4M = $3.0M operating income
At 25% GM:   $40M × 25% = $10.0M GP - $5.4M = $4.6M operating income

Every percentage point of gross margin improvement drops ≈$350K to operating income. The move from 16% to 22% — which is where they were running two years ago — creates $2.1M in operating income on current revenue alone. At $3M operating income on a $55M enterprise value, that's 18x forward earnings. Reasonable for a niche manufacturer with sole-source claims and secular demand tailwinds.

The $26M backlog heading into 2026 covers 80% of prior-year revenue. Revenue visibility is strong.

Why SiC Matters

CPSH's core product — AlSiC baseplates — goes into power semiconductor modules. The industry is transitioning from silicon IGBTs to silicon carbide (SiC) devices. SiC runs hotter than silicon. Hotter semiconductors need better thermal management. Better thermal management means AlSiC baseplates become MORE critical, not less.

The SiC power semiconductor market is growing at 25-29% CAGR ($2.7B in 2025 to $8.4B by 2030). This isn't speculative. Infineon's Q1 FY2026 earnings call (February 4, 2026) explicitly confirmed grid infrastructure as the only growing sub-segment within their Green Industrial Power division. They're investing EUR 500M in AI power supply manufacturing expansion.

CPSH's largest customer — almost certainly Infineon, based on the 10-K showing Customer A at 39% of revenue and Germany at 39% of revenue (identical), plus the Q3 call describing a "$15.5M contract with a long-standing multinational semiconductor manufacturer" for "high-speed rail and grid infrastructure" — is at the center of this transition.

The critical finding from factor regression: CPSH trades with ZERO correlation to Infineon (β = 0.04). Despite 39% revenue dependence. The market has no idea these two companies are connected. The SiC secular tailwind is not being priced through CPSH because nobody connects a $68M Norton, MA manufacturer to the global power electronics supply chain.

The Insider Signal

October 8, 2025. CPSH priced a 3.45M share secondary offering at $3.00/share, raising $9.5M net proceeds through Roth Capital.

On the same day, five insiders bought open market at $3.00:

Total: 67,166 shares, ≈$201,500 of personal capital. Cavoli added another 1,500 shares on November 13. Hughes and Griffith exercised options in December and February and held — no paired sales.

This is the strongest insider signal type. Not a scheduled 10b5-1 plan. Not option exercises. Open market purchases with personal money, coordinated across the entire leadership team, at the same price as the institutional offering. They're betting on the margin inflection.

The Hidden Option: SBIR Phase III

CPS has three active Phase II SBIR programs: an Army tungsten warhead (40mm controlled fragmentation), a Navy anti-ship missile thermal management system, and a DoE modular radiation shielding project.

Under federal law, successful Phase II SBIR awardees receive non-competitive Phase III production contracts. No bidding. Sole-source. No dollar limit. Twenty-year IP protection.

The Army warhead program is producing samples that "exceed Army performance benchmarks." The CEO described the addressable market — Army artillery — as "big numbers."

At $68M market cap with one analyst, nobody is modeling SBIR Phase III option value. A single $5-10M Phase III production contract would represent 15-30% of current revenue. The timeline is 2027-2028 — not imminent, but real. IEHC, a comparable nano-cap defense connector manufacturer demonstrates the pattern: sole-source defense contracts at small companies translate to sustained revenue with pricing power. IEHC's CEO explicitly stated "much of this business is sole-source and thus highly profitable."

SBIR reauthorization lapsed September 30, 2025, but CPSH's four ongoing programs continue funded. Congress was showing "indications of compromise" as of the Q4 call, with potential reauthorization through September 30, 2031.

The Bear Case

Customer concentration is the kill risk. Customer A (Infineon) is 39% of revenue, up from 32% last year. Concentration is worsening, not improving. There are no long-term contracts — purchase orders only. Infineon can reduce orders with minimal notice. And because CPSH shows zero IFNNY correlation, an Infineon demand cut would arrive as a pure idiosyncratic shock with no early warning from Infineon's stock price.

Mitigants exist — AlSiC is a duopoly (CPSH and Denka of Japan), qualification cycles are 3-4 years, Infineon's grid/SiC segment is growing — but 39% single-customer concentration on purchase orders is structurally fragile.

CFO retiring at the wrong time. Chuck Griffith announced retirement after Q4 earnings. Successor search is ongoing. This is happening simultaneously with the facility relocation decision, third-shift ramp, and defense program pursuit. Three execution challenges at once for a 117-person company.

The growth is partially catch-up. Revenue was $26-27M in 2022-2023 before the 2024 quality issue cratered it to $21M. The 54% growth in 2025 is partly recovery to trend, not purely organic expansion. Organic growth net of gold pass-through ($1.4M inflated revenue) is closer to $31.2M — still above prior peak, but less dramatic.

Facility relocation = near-term disruption. The Q4 call confirmed DAO Corporation has been selected as general contractor and the move begins "several months from now." Customer requalification is required after the move. Any production disruption in H2 2026 could delay the margin improvement thesis.

Factor Decomposition

iev regress CPSH --factors "SPY,IFNNY,AMSC,ITA"

Idio:   89.4%    (well above 75% threshold)
R²:     10.6%    (almost nothing explains this stock)

AMSC (grid peer):     5.1%
ITA (defense):        4.4%
IFNNY (customer):     0.0%   ← market doesn't price the relationship

Structural decomposition of where edge exists:

≈80% of the thesis sits in factors where we have edge. The 20% where we don't (concentration + execution) is the irreducible risk.

The Verdict

This is a doorway state. 60% bull, 40% bear. The pattern collapses on April 29 when Q1 2026 earnings drop.

Pre-resolution forward α: ≈6%. Not enough to size. The probability-weighted expected value — 60% chance of $5.50 target, 40% chance of $2.50 — produces thin edge before the pattern resolves.

Post-resolution (if Q1 confirms): ≈19% forward α. If gross margins tick above 18% on continued revenue growth, the bull case confirms and the stock re-rates toward normalized earnings.

Post-resolution (if Q1 disappoints): Kill. Another operating loss with margins stuck at 16% means the third-shift proficiency story was wrong. No nursing.

The interesting edge here isn't the stock itself — it's monitoring Infineon's Q2 FY2026 call (~May 2026) as a leading indicator on CPSH demand — a signal nobody else is watching because the market doesn't know these companies are connected.

Scenario:

EV = $4.53 (+19% from current $3.81).

LR = 1.4. Mild bullish. The market is slightly underpricing the margin inflection and completely missing the SiC supply-chain connection. But the evidence is mostly management commentary (credibility 0.85-0.90) and structural inference, not binding contracts or hard catalysts. Customer concentration is a genuine kill risk that could overwhelm the bull case. Pre-resolution, this is watch-and-wait. Post-April-29, if margins confirm, the forward α becomes meaningful.

Evidence