Phase Space AI

Valuation

SharkNinja [SN]

SharkNinja [SN] — Valuation

Spot $160.795 · 142.359m shares · market cap $22,891.6m · net debt $214.5m (lease-exclusive) · EV $23,106.1m · TTM revenue $6,589.4m (to 2026-03-31) · EV/Sales 3.51x

EV is stated LEASE-EXCLUSIVE. Lease-inclusive (adding $134.6m of noncurrent operating lease liabilities) EV is $23,240.7m, EV/Sales 3.53x. The required-CAGR difference between the two bases is 0.2pp (13.9% vs 14.1%) — immaterial, but stated rather than assumed, per the CRWV precedent where a lease-exclusive anchor invalidated a seven-name ladder.

---

Company state: A (mature and structurally stable) — with one dissent recorded

Test applied (valuation.md STATE A): profitable; operating margin low-variance across ≥5 years; no structural regime change.

Evidence for A: consistently profitable across the full six-year window including the 2022 freight/inflation trough; positive operating cash flow every year; no transformative acquisition; the JS Global separation was a change of ownership, not of business model, and the pre-separation statements are the same legal entity's (see SN_Research.md §1).

Evidence against A, recorded rather than suppressed: operating margin has moved 8.6% → 14.4% over three years, a +5.8pp ramp, and gross margin +11.1pp over four. steady_state_check.py computes the ramp from the 2020 endpoint and reads it as −2.5pp; measured from the 2022 trough it is +5.8pp, and the trough is the more informative anchor because 2020 was a COVID stay-at-home peak. A margin that has moved 5.8pp in three years is closer to STATE C's "scaling but economically observable" than to A's "low-variance across ≥5 years."

Declared A, with the consequence that the terminal margin is built from the forward bridge rather than simply set at the trailing actual — which is what STATE C's discipline would require anyway. The state assignment does not change the instrument here; the reverse DCF is the primary output either way because terminal value exceeds 60% of EV.

evidence_grade: B. The disclosure set is complete on revenue, categories, customers and advertising, and absent on volume/price, manufacturing geography and tariff magnitude.

---

Terminal margin: 15.0%, and what it assumes about A&P

Basis: the company's own demonstrated FY2025 operating margin of 14.4%, plus 0.6pp.

Opex bridge reconciled from 10-K lines, FY2025 (all as % of net sales of $6,399.2m):

Terminal bridge (year 5):

What this assumes about A&P, stated explicitly as the brief requires: nothing. It assumes A&P does not lever at all. Advertising is held at 9.9% of sales in perpetuity, which is above the FY2023 9.6% and below the FY2024 10.6% — i.e. inside the company's own three-year observed range and at its most recent print. The entire +0.6pp of terminal expansion comes from R&D, non- advertising S&M and G&A, and totals 0.7pp of leverage on $6.4bn of revenue growing to ~$12.7bn.

The reason for that construction is in SN_Research.md §2–3: 72.3% of FY2025 growth came from categories launched inside the window, and a new category is exactly what requires advertising. A terminal margin built on A&P leverage would be asserting that SharkNinja needs less marketing support while still needing a new category every year. I do not think both can be true, so I have not assumed either.

Constraints checked: - m_EBIT,T (15.0%) ≤ m_gross,T (49.0%) ✔ — with 34.0pp of headroom, so this is not the binding constraint on this name (unlike the low-gross-margin cases where it is). - Terminal margin is above the trailing actual (14.4%), so the "21 of 84 names below their own trailing actual" defect does not apply here. - Full expense bridge, not just the ceiling, reconciles above ✔.

---

Implied-path test (primary output — terminal value is 74% of EV)

reverse_dcf.py --spot 160.795 --shares 142.359 --net-cash -215 --revenue 6589.4
               --years 5 --wacc 0.09 --terminal-margin 0.15 --exit-multiple 17.0
               --fcf-margin 0.057 --hist-cagr 0.198

--fcf-margin passed as +0.057 (TTM FCF $378.3m / TTM revenue $6,589.4m = +5.74%). The sign is positive, so omitting it would have overstated required CAGR and suppressed a true PASS. The measured bias at a ~5% FCF margin is ~1.2pp (DATA_DEFECTS.md). Trailing and forward signs agree here — there is no capex cycle turning FCF negative, unlike GOOGL — so the trailing basis is the right input. Basis used: trailing TTM.

Required revenue CAGR: 13.9% Demonstrated: 19.8% (screen's 3-year figure, verified against the 10-K: $3,717.0m FY2022 → $6,399.2m FY2025 = 19.87% ✔) Margin: demonstrated − required = +5.9pp

Verdict: PASS. The price does not require acceleration.

Implied compression: SN trades at EV/EBIT of 24.4x on TTM operating income of $946.8m. Solving at 17.0x is an assumed −30.3% multiple compression over five years, which is the correct direction for a business decelerating from ~20% to ~14% revenue growth.

Sensitivity over the exit multiple (never over scenario probabilities)

Flip point is between 14x and 15x. The verdict on this name is a judgement about the exit multiple, and I am saying so rather than presenting 17.0x as settled.

What steady_state_check.py flagged

steady_state_check.py --ticker SN --terminal-margin 0.15 --exit-multiple 17.0

• MARGIN RAMP: 2020 margin 16.9%, latest 14.4%, moved −2.5pp, "latest is window peak: False", terminal 15.0% = +0.6pp vs latest, −1.9pp vs peak. No flag raised on the margin. (As noted above, the 2020 anchor makes this read as a decline; measured from the 2022 trough it is a +5.8pp ramp and the tool would flag it. The tool's window start is doing work here — a finding about the tool, not the company.)
• ROIC measured 20.3%; warranted multiple at steady-state ROIC 9.6x; terminal reinvestment rate implied by g/ROIC 14.7%; SCALING CONTAMINATION 1.00x.
• FLAG RAISED — EXIT_MULTIPLE_UNWARRANTED_AT_STEADY_STATE: "17.0x is 1.8x the 9.6x warranted at a 20% steady-state ROIC. It is justifiable only if today's returns persist forever, which is the assumption under test."

My response, rather than dismissal: the flag is right that 17.0x is not derivable from EV_T/EBIT_T = (1−t)(1−g/ROIC)/(WACC−g) at g=3%, ROIC=20.3%, WACC=9%. At the warranted 9.6x the name FAILS by 6.8pp. The 17.0x base is defended on the traded record — SN's own EV/EBIT has not been below ~15x since listing — and on the observation that a warranted-multiple identity computed at a 3% terminal growth rate on a business that has compounded at 19.8% is measuring a different company. But that is exactly the assumption under test, and the honest statement is that this name's PASS is contingent on the exit multiple, not robust to it. Contrast CRDO, which this corpus found robust across 10x–27x.

---

12-month target: UNIDENTIFIED

Multiple history (own, EV/Sales, as-known TTM revenue lagged 45 days, split-and-dividend adjusted daily bars):

• 752 trading days, 2023-07-31 → 2026-07-29 — the entire post-listing record, 3.0 years.
• min 1.31x · p25 2.24x · median 2.63x · p75 2.98x · max 3.61x
• current 3.63x = 100th percentile.

The history does NOT span the separation — it begins on the listing date, so the WDC-style contamination is absent. It is nonetheless declared UNIDENTIFIED, for two reasons:

1. The sample has no support above spot. At the 100th percentile there is no observation above the current multiple. Any mean-reverting anchor drawn from this series is mechanically one-sided, so a "target" built on it is a statement that the stock falls, generated by the sample boundary rather than by analysis. That is the mirror image of item B16 (16 of 16 targets below spot) and it would be manufacturing the same defect from the other direction.
2. Three years, entirely inside one regime. The window contains no consumer-discretionary down-cycle and no multiple contraction event for this name. valuation.md requires declaring UNIDENTIFIED where the history "is too short or spans a regime change"; a window with no dispersion in the dimension being matched is the reference-class failure described in rule 6 — valid class, zero information.

What I will report instead of a fabricated point estimate, using consensus NTM revenue of $7,713m (5/12 × FY2026 $7,241.0m + 7/12 × FY2027 $8,055.7m, AV EARNINGS_ESTIMATES, 13 analysts):

Estimate-revision direction (AV built-in, 30-day trailing): FY2026 EPS 11 up / 0 down; FY2027 11 up / 1 down — strongly positive. But Q2 2026 is 1 up / 7 down — the near-term quarter is being cut while the years are being raised. That divergence is itself the signal: the Street is pushing the year's earnings into the back half.

---

Downside case with a named cause (Type: MEASURED — does not reject the name)

Named cause: a failed category launch coinciding with the receivable normalising.

The mechanism is specific. 72.3% of FY2025 growth came from Food Prep and Beauty & Home Environment. If the FY2027 adjacent-category launch does not take — and there is no contract, backlog or RPO that says it will — consolidated growth reverts to the legacy categories' 6.3%. Simultaneously, a retailer that has extended terms (AR +31.6% vs revenue +15.7%) tightens them, and the working-capital release that has been funding growth reverses.

Quantified: at 6% revenue growth and a 12.0% operating margin (FY2024's level, reflecting the advertising that a failed launch still costs), FY2028 EBIT is ~$940m. At a 12x exit multiple — the low end of the compression range, appropriate for a 6% grower — EV is $11.3bn, equity $11.1bn, $78/share, −51.5%.

That is the permanent-impairment case, not a volatility figure. It requires the category engine to stop, which is observable a year in advance from the category disaggregation table.