Semiconductors
PublicSKYT is no longer a foundry — it is an IonQ deal-stub trading $38.40 vs a $35.00 closed-vote merger value, i.e. a +10% premium that is pure leveraged FTC-cleared bet on IonQ rallying past $60.13; the standalone foundry (43% Infineon, negative op cash flow, $22M cash) is the break-case floor, not the thesis.
Research
The verdict
SKYT is no longer a foundry — it is an IonQ deal-stub trading $38.40 vs a $35.00 closed-vote merger value, i.e. a +10% premium that is pure leveraged FTC-cleared bet on IonQ rallying past $60.13; the standalone foundry (43% Infineon, negative op cash flow, $22M cash) is the break-case floor, not the thesis.
Primary sources
SkyWater is a US-headquartered (Bloomington, Minnesota), exclusively domestic, pure-play semiconductor foundry — a Delaware corp, Nasdaq: SKYT, CIK 0001819974, FY ends late December (FY2025 = 28 Dec 2025). It is one of the very few merchant foundries with a DMEA-accredited Category 1A Trusted Foundry status — it can fabricate classified / mission-critical defense and intelligence chips on US soil. That accreditation, not its process technology, is its scarcest asset.
What it actually sells — two service categories across two reportable segments:
Business model. The thesis SkyWater sells is that differentiation in semis has migrated from circuit design into process technology, materials, device physics, integration and packaging — so customers co-develop the full stack with the foundry. It targets foundational/specialty nodes (not leading-edge logic) for end-markets with multi-decade product lives: aerospace & defense, automotive, industrial automation, medical devices, energy infrastructure, and emerging computing (quantum/photonics).
Contract structure — a key weakness. SkyWater has no significant backlog; purchase orders are cancellable until shortly before production. Customers provide only 12-month rolling forecasts. Management flags explicitly that this makes revenue and margins hard to forecast and that expense levels (largely fixed fab cost) can't flex with revenue shortfalls. The Infineon Fab 25 layer is the exception — a multi-year supply agreement valued >$1B anchors the Texas output.
Map: specialty materials / process chemicals / gases + semiconductor capital equipment (the "Tools" customers fund) → SkyWater's three US fabs (MN 200mm + FL packaging + TX 200mm) → fabless / system customers in defense, compute, automotive, medical → end systems (defense primes, hyperscalers' bespoke silicon, quantum platforms).
Named stakeholders along the chain:
Chokepoints / single-source dependencies: (1) Infineon is the single biggest chain risk — 43% of revenue and one customer ≈ 59% of accounts receivable as of 28 Dec 2025. (2) The fabs themselves — concentrated US footprint with no offshore redundancy (a feature for Trusted-Foundry buyers, a fragility for operations/yield). (3) Capacity build depends on customer-funded Tools — when that funding completes, the Tools line collapses (it did, −62% in FY2025).
Net: the moat is regulatory (Trusted Foundry) + relationship (co-development), not economic (scale/cost). It is exactly the kind of asset worth more inside a strategic acquirer than standalone — which is the whole logic of the IonQ bid.
segments.csv is empty (headers only) — so all segment numbers are pulled directly from the 10-K, labeled ``.
By service type (FY2025 vs FY2024):
| Service type | FY2025 | FY2024 | Δ |
|---|---|---|---|
| ATS development | $212.5M | $238.6M | −11% |
| Tools | $28.9M | $76.8M | −62% |
| Wafer Services — Legacy | $25.5M | $26.9M | −5% |
| Wafer Services — SkyWater Texas | $175.3M | $0 | NM (new) |
| Total | $442.1M | $342.3M | +29% |
The single most important fact in the financials: every legacy line declined. The headline +29% growth is entirely the bolted-on Texas fab (Infineon off-take). Strip Texas and SkyWater's organic business shrank ~22% (Legacy segment revenue fell from $342.3M to $266.8M).
By segment (FY2025):
Cause of the legacy decline: ATS development −$26.1M was driven by a $41.8M collapse in aerospace-&-defense revenue ("recent U.S. government policy shifts and changes in defense spending priorities… delayed contract awards"), partially offset by +$12.9M in advanced-compute (quantum). Tools −$47.9M because customers finished their capability-build funding cycles. So the differentiated, sticky part of the business (defense ATS) is the part that rolled over.
FY2025 (10-K, FYE 28 Dec 2025):
Latest quarter — Q1 FY2026 (10-Q, ended 29 Mar 2026, filed 8 May 2026):
Market reaction: decoupled from fundamentals — SKYT trades on the IonQ deal, not the print. Despite the operating loss and FTC Second Request, the stock hit an all-time high $39.93 on 2 Jun 2026 and trades ~$38.40. What's "priced in" is deal certainty + IonQ upside, not earnings.
Unusual vs own history: (1) first GAAP profit ever — but it's a non-cash bargain gain; (2) first time operating income went negative since the IPO era; (3) operating cash flow went deeply negative; (4) customer concentration leapt from "two customers 40%/20%" to "Infineon alone 43%."
transcripts/ is empty — sentiment is reconstructed from filed 8-K results and press, labeled accordingly.
Tone shift: from "defense delays are temporary, quantum + Texas will more than offset" (2025, defensive-optimistic) → "we are being acquired by IonQ to build the only vertically-integrated US quantum platform" (2026, sold). What they stopped saying: the confident "record ATS year" framing — because A&D revenue fell $41.8M and the organic business shrank. The recurring 2026 phrase is "trusted domestic / US-owned and US-operated."
| Company | Ticker | Mkt cap | EV/Sales | EV/EBIT | P/E | Div yld | 5y avg ROE |
|---|---|---|---|---|---|---|---|
| SkyWater | SKYT | ~$1.8–1.9B | n/a | neg (op loss) | n/a (gain-distorted) | 0% | negative |
| Tower Semiconductor | TSEM | n/a | n/a | n/a | n/a | — | n/a |
| GlobalFoundries | GFS | n/a | n/a | n/a | n/a | — | n/a |
| UMC | UMC | n/a | n/a | n/a | n/a | — | n/a |
The comp that actually matters here is not a foundry peer — it is the deal. Because SKYT is a closed-vote acquisition target, the relevant "valuation" is the merger consideration, not a trading multiple:
SkyWater IPO'd in 2021 at $14. The names of the moves:
Pattern: pre-2026, SKYT reacted to defense-program funding and single-customer/Tools lumpiness (classic small-cap binary risk). Post-Jan-2026, it reacts almost exclusively to IonQ's stock price and FTC headlines — its beta is now to IonQ + deal risk, not to foundry fundamentals.
insider-transactions.csv not present — insider ownership not quantified from the research layer; n/a for precise %.Ground: 10-K + 10-Q financials, every figure labeled.
Regulatory findings (required sub-section).
This is a deal-stub, so the meaningful "projection" is the consideration payoff, not a standalone EPS path. I give both, clearly labeled.
(a) Deal-value projection (the live trade) — output: Per-share consideration = $15.00 cash + IonQ stock leg, where the stock leg = $20.00 fixed when IonQ's 20-day VWAP (3 biz days pre-close) is between $37.99 and $60.13; = 0.3326 × VWAP above $60.13; = 0.5265 × VWAP below $37.99.
| IonQ price at close | Stock leg | + $15 cash | Deal value/share | vs SKYT $38.40 |
|---|---|---|---|---|
| $25.89 (IonQ 52w low) | $13.63 | $28.63 | −25% | |
| $37.99 (floor) | $20.00 | $35.00 | −9% | |
| $56.55 (today) | $20.00 | $35.00 | −9% | |
| $60.13 (cap) | $20.00 | $35.00 | −9% | |
| $70.00 | $23.28 | $38.28 | ~flat | |
| $84.64 (IonQ 52w high) | $28.15 | $43.15 | +12% |
Base case: deal closes Q3 2026 with IonQ in the collar → holders receive ~$35.00; buying at $38.40 loses ~9% unless IonQ rallies above ~$70 before close. Bull case: IonQ runs past the $60.13 cap (quantum momentum continues; IonQ already +755% YoY Q1 revenue, raised FY26 guide to $260–270M) → stub worth $38–43. Bear case: FTC blocks/forces a re-cut, or IonQ stock falls below the floor → SKYT collapses toward standalone fair value (well below $35; a money-losing sub-scale foundry on $22M cash would not hold a $1.8B cap), a >35% drawdown from $38.40.
(b) Standalone EPS (the break-case floor) — output: If the deal broke, the run-rate is roughly breakeven-to-small-loss EPS: FY2025 underlying (ex-gain) net income ~$7.2M / ~48.7M diluted shares ≈ $0.15 "core" EPS, but Q1 FY2026 ran a −$0.25 loss as interest expense ($6.2M/qtr) and the defense air-pocket bit. A standalone FY2026 would plausibly print a small GAAP loss (negative operating margin + ~$24.6M annual interest vs ~$52M Adj EBITDA), with Texas/Infineon volume the swing factor.
Forecast log: Skipped per --watchlist rule (log a Brier forecast only on a genuine committed base case; do not create in the loop). The scoreable proposition if logged would be binary on deal completion ("SKYT/IonQ merger closes by 2026-12-31"), not an EPS line — appropriate for /thesis, not here.
Bull case (for owning SKYT at $38.40). The deal is closed-vote and strategically logical: IonQ needs a domestic, Trusted-Foundry, quantum-capable fab to vertically integrate (pull forward its 200,000-qubit QPU testing to 2028) and to qualify for US-government/defense procurement. A vertical deal (no horizontal foundry overlap) is hard for the FTC to block — Second Requests on vertical deals usually end in clearance with timing extension, not prohibition. With the stock leg uncapped above $60.13 and IonQ on a quantum tear (+755% YoY revenue, Strong Buy, $67.64 avg target ), SKYT is a leveraged long-IonQ call with a $15 cash floor — if IonQ runs to $70–85, the stub is worth $38–43 and you collected optionality cheaper than buying IONQ outright.
Bear case (2–3 things that permanently impair / lose money). (1) You are paying a +10% premium to a $35 deal — the base case is a negative-9% return unless IonQ rallies ~24%+. (2) FTC Second Request is a real overhang — even a clearance pushes close into Q3+ (time decay on the cash leg; risk of a IonQ-stock-driven re-cut). A break sends SKYT toward standalone value, and standalone is a sub-scale foundry that lost money operationally, burned cash, runs on a Revolver with $22M of cash, and just lost $41.8M of its stickiest (defense) revenue with 43% single-customer dependence — a >35% air-pocket. (3) The stub's downside below the $37.99 floor is IonQ's downside (0.5265 ratio): IonQ is a richly-valued, pre-profit quantum name with an $25.89–$84.64 52-week range — a 30%+ IonQ drawdown before close drags the stub under $30 even if the deal completes.
Pre-mortem (18 months out, thesis broke): FTC's Second Request dragged into Q4 2026; IonQ's quantum momentum cooled and the stock fell to the high-$30s/low-$40s; either the deal closed near the floor (holders got $30–33, a loss from $38.40) or, worse, the parties re-cut/terminated on the antitrust timeline and SKYT re-rated to standalone ($10–15 zone given the cash burn and lost defense revenue) — a ~60% loss from the premium entry.
Are multiples too high? As a standalone, unquestionably — ~4x sales / ~33x Adj EBITDA for a money-losing specialty foundry is a strategic-takeout multiple, not a fundamental one. As a stub, the question isn't the multiple but (IonQ price × ratio) + $15 vs the price you pay — and at $38.40 you're paying for IonQ upside you don't yet have.
Contrarian view (what the market is refusing to see): the crowd is treating SKYT at an all-time high as "deal certainty + IonQ rocket." What it's under-pricing is that at $38.40 the arb is negative against the $35 base case — you are not clipping a discount, you are paying up for a leveraged IonQ bet while wearing FTC-break risk whose downside is a cash-burning, customer-concentrated, defense-exposed foundry. The asymmetry has quietly flipped against new longs.
Dismantling the long. (1) Revenue concentration is grotesque and worsening — Infineon 43%, top-3 = 74%, one customer = 59% of A/R. The "growth" is one acquired off-take contract; lose or reprice Infineon and the P&L caves. (2) The moat bulls cite is the defense/Trusted-Foundry base — and that base just shrank $41.8M on government-spending shifts; the sticky part is the weak part. (3) The most dangerous competitors bulls underestimate: Tower, GlobalFoundries (which just bought AMF to lead silicon photonics ), UMC, X-FAB — all larger, better cost curves, also chasing specialty/photonics/quantum-adjacent work; SkyWater's only defensible edge is the accreditation, not the technology. (4) Worst capital-allocation optics: debt-funded acquisition that turned operating cash flow negative + VIE/failed-sale-leaseback financing structures + a related-party-flavored Infineon triple-role. (5) Assumptions that must hold for $38.40: the FTC clears and IonQ trades above ~$70 at close and nothing in the supply agreement / Texas ramp disappoints in the interim. (6) If growth disappoints 20–30% standalone: with negative operating margins and fixed fab cost + ~$24.6M interest, a 20–30% revenue miss pushes the company into meaningful losses and tighter Revolver covenant headroom. (7) Single scenario that permanently impairs: FTC forces termination → SKYT standalone with $22M cash, negative op cash flow, lost defense revenue, and 43% customer concentration must either dilute or lever further — a plausible 50–60% equity impairment from the current premium. Plausibility: low-but-not-trivial (vertical deal, but extended FTC reviews occasionally end in abandonment).
Research Trail
Covered in the Knowledge Base
Hardware — Memory & Compute Architecture
A barely-profitable IDM whose equity 10x'd on a recapitalization-and-validation narrative (US gov, Nvidia, SoftBank, hyperscaler 18A interest) while Foundry still bled $10.3B in FY25 — the story is priced as if the turnaround already happened. WATCHING, with a bearish lean on valuation.
The balance sheet is fixed but the business is not — at ~4x EV/sales with a -27% GAAP gross margin, a commoditizing substrate core, and a ~35% burned-strategic overhang (Renesas), WOLF is priced for an AI-datacenter / 200mm-utilization inflection the P&L will not confirm before FY27; WATCHING until gross margin crosses zero.
Source documents — open to read in full
The purest listed play on the Big-3's DDR4 exit — a real supply-shock oligopoly windfall (GM 30%→53% in six quarters, sold out through 2027), but +890% in a year prices a transient legacy-DRAM squeeze as if it were a structural moat; WATCHING, not chasing, at 52× trailing peak-cycle earnings.