Phase Space AI

Valuation

Xylem [XYL]

Xylem [XYL] — Valuation

As of 2026-07-30. Spot $116.63.

---

COMPANY STATE — declared first

STATE A — mature and structurally stable.

Test applied and evidence:

• Profitable throughout, 14 consecutive filed years, no loss year.
• Low margin variance across ≥5 years: filed operating margin 7.5% (2020, COVID) to 13.5% (2025), with a 9.5%–12.8% band in every other year of the 2012–2022 window. Gross margin has sat in a 36.9%–39.6% band for fourteen straight years — the tightest of the three names in this cluster and the clearest single piece of State A evidence.
• No exogenous cycle drives it. Demand is municipal and industrial water infrastructure — utility capex, regulated and multi-year. This is not fab utilisation or a commodity price.
• Regime change test: the Evoqua acquisition (2023-05-24) was a transformative acquisition and it disqualifies the pre-2024 window. FY2024, FY2025 and 1H 2026 are three clean fully-consolidated periods and are what State A is asserted on. The pre-2023 series is used only for the gross-margin stability observation, which is unaffected by the deal.

Consequence: terminal margin is built from Xylem's own normalised economics plus an explicit forward bridge; industry data is a sanity band, never an override. Solve for price-implied growth with the margin fixed.

Explicitly NOT State B. Xylem's revenue fell in exactly one of fourteen years (2020, −7.1%, COVID) and the margin recovered within two. There is no cycle to date and no mid-cycle to normalise to. Mean-reverting this name's margin downward would be the reflex the brief warns against.

---

TERMINAL MARGIN — 16.0%

Basis: the company's own demonstrated economics plus a forward bridge from FY2025 10-K lines.

The opex bridge, from 10-K lines (FY2025)

Bridge closes: 38.46 − 21.28 − 2.50 − 1.14 = 13.54% ✓

The forward bridge to 16.0%

Hard constraint

m_EBIT,T (16.0%) ≤ m_gross,T (39.5%) ✓ — 23.5pp of headroom against a 23.1%-of-revenue combined SG&A + R&D base. The full expense bridge, not just the literal ceiling, is satisfied.

steady_state_check.py — what it flagged, and the named mechanism

python3 steady_state_check.py --ticker XYL --terminal-margin 0.160 --exit-multiple 9.59

Flag 1 is real and I am accepting it with a named mechanism, which is what the tool demands. The tool's own words: "Defensible only with a named mechanism — operating leverage on a fixed cost base is a reason; 'it has been rising' is not."

The named mechanism is threefold and each part is quantified in a filing:

1. Q2 2026 already printed a 16.7% GAAP operating margin — above the 16.0% terminal, in a filed quarter. TTM is 14.48%. The terminal is not an extrapolation past anything achieved; it is 0.7pp below the most recent quarterly actual.
2. Restructuring ends on a stated date. The 10-K says the M&C actions run "through the end of 2027." Removing 74bp of the FY2025 1.14% is not a forecast, it is reading the disclosure.
3. 350bp of productivity savings and 240bp of price realisation are quantified in the Q2'26 10-Q as realised, against 270bp of inflation and 160bp of volume drag. The net operational contribution is +160bp and it recurs while the productivity programme runs.

Flag 2 is also real and I am NOT overriding it — see the exit-multiple section, where I use the identity-derived multiple rather than the market's.

Honest caveat, stated because the tool is right to force it: 180bp of Q2'26's 340bp of GAAP margin expansion came from lower charges, not operations. The terminal bridge already accounts for this by moving restructuring only to 0.4%, not to zero.

---

THE IMPLIED-PATH TEST — the Valuation Criteria

Verified inputs

EV basis statement, as required: headline EV of $28,881.6m is LEASE-EXCLUSIVE. Lease-inclusive is ~$29,457m, +2.0% — the largest lease adjustment in this cluster. Both are stated. EV/EBIT moves 21.86x → 22.30x; the conclusion does not change, but a lease-exclusive anchor previously invalidated a seven-name ladder in this project, so it is declared rather than assumed away.

Current multiples

The instrument is a discounted cash flow — confirmed

reverse_dcf.py was executed. It discounts a five-year interim FCF stream plus a discounted terminal value at an explicit WACC and solves for the revenue CAGR that equates present value to today's EV. It is not the undiscounted required-revenue-level test that CORZ and IREN substituted.

--fcf-margin — sign and magnitude

Xylem's free cash flow is POSITIVE and stable, so the correction direction is conservative and cannot manufacture a false PASS.

• TTM operating cash flow $1,301m; TTM capex $341m; TTM FCF $960m = +10.5% of revenue.
• Capex is 3.7% of revenue — this is a capital-light business.
• normalized_fcf.py --ticker XYL returns INDETERMINATE: "no growth capex identified above the maintenance rate — either the company is not building, or maintenance capex is overstated. Trailing FCF may be usable directly; check." That is the correct output and I checked it: Xylem is genuinely not building. Trailing FCF is therefore usable directly, which is the tool's own stated condition.
• Used: --fcf-margin 0.105.
• Forward/marginal check (the GOOGL lesson): no quarter in the last eight had negative free cash flow. Q1'26 was the weakest at $108m OCF against $90m capex (+0.8% margin) — seasonal working capital, recovered to $290m − $89m in Q2. The sign does not flip on any basis.

Parameters

The exit multiple is determined by the identity

EV_T / EBIT_T = (1−t)(1 − g/ROIC) / (WACC − g)

t = 19.56% (FY2025: 231 / 1,181), g = 2.5%, WACC = 8.5%, ROIC = 8.80% (NOPAT $1,062.1m = TTM EBIT $1,321m × (1−0.1956); invested capital $12,071m = net debt $1,650m + equity $10,421m):

(0.804)(1 − 0.025/0.088) / 0.060 = (0.804)(0.7159) / 0.060 = 9.59x

steady_state_check.py returns 6.1x at its own 6.5% ROIC measure. I use the higher 9.59x.

The ROIC ambiguity is the honest complication on this name and it must be stated. Reported ROIC of 8.80% is depressed by $8,256m of goodwill and $2,150m of intangibles — almost all of it Evoqua purchase price. Ex-goodwill-and-intangibles invested capital is $1,665m and ROIC is 63.8%. The underlying water-technology business earns extraordinary returns on tangible capital; the company earns 8.8% because it paid $7.5bn for Evoqua.

Which is right for a terminal value? The reported figure, and I am using it deliberately. A terminal value is what an owner earns on the capital actually deployed, and the Evoqua capital was actually deployed. Using the 63.8% tangible figure would give an exit multiple of ~12.9x and a required CAGR of ~13.7% — still a large FAIL. The verdict is robust to this choice, which is why it is reported rather than argued.

RESULT

  EV implied by today's price         28,882m   (3.2x current revenue)
  Held fixed              terminal margin=16.0%, exit multiple=9.59x
  Discount rate / horizon 8.5% over 5y

  >>> THE MARKET REQUIRES: revenue CAGR of 19.3%

Sensitivity — over the exit multiple

Xylem fails at every exit multiple in the plausible range, and it fails even at the screen's own anchor. That is a materially stronger result than KGS's, where the verdict hinged entirely on the multiple. Here the problem is the numerator: 0.5–1.3% organic growth cannot support a 3.2x EV/Sales multiple under any terminal assumption that respects the ROIC identity.

The screen's +8.2pp PASS came from combining a 23.5x exit multiple (from a 286-name "peer" set) with an 18.1% demonstrated CAGR that is the Evoqua acquisition. Correct either one and it fails.

---

THE 12-MONTH TARGET

Multiple anchor — the name's OWN history, percentile stated

Daily EV/Sales from Alpaca split-adjusted closes against as-known annual revenue (FY revenue stepping in at its 10-K filing date), current shares and net debt held constant.

Window: 2020-07-27 → 2026-07-29, n = 1,506 trading days (six years).

Xylem trades at the 6th percentile of its own six-year multiple history. That is the single most striking fact on the name, and it is the mirror image of the BSX finding recorded in this project (PASS at the 0th percentile of its own 8-year EV/Sales history).

Trailing one-year window: median 3.94x, today's value at the 33rd percentile. Trailing three-year: median 4.14x, 13th percentile.

Regime-change check, per valuation.md: the six-year window spans the Evoqua acquisition, which raised revenue ~30% and lowered the multiple mechanically. That biases the pre-2024 observations upward relative to today's business. I therefore anchor at the 25th percentile (4.03x), not the median, and round to 3.95x — which is also the median of the trailing one-year window. Two independent constructions land within 2%.

NTM revenue base and revision direction

Revisions are positive but small, and monotone — every 7/30/90-day step is upward on FY2026. No whipsaw. This is the profile of a name being revised on execution rather than on a surprise.

Target

Cross-check on earnings: $153.06 ÷ NTM EPS of $5.884 = 26.0x, against today's 19.8x. Xylem has traded in a 22–35x forward P/E band for most of the last six years, so 26x is inside its own range and is not the aggressive end of it.

What the target actually asserts, stated plainly: essentially all of the +31.2% is multiple re-rating from the 6th percentile to the 25th, not growth. NTM revenue of $9,465m is +3.7% on TTM. The named events that could drive that re-rating are (a) the Q2'26 orders spike being confirmed in the FY2026 backlog figure, (b) Applied Water data-centre revenue being quantified for the first time, and (c) the operating margin holding above 16% for a second and third quarter. All three are in the catalyst calendar with what is and is not a disclosed date.

No external professional target is on file for XYL in this project, so the sanity band is UNAVAILABLE rather than passed.

---

COMPARATOR SET — XYL only

Water and flow-control infrastructure equipment, US-listed, growth bracketing Xylem's low-single-digit organic rate.

Growth bracketing: organic growth across this set runs roughly 0–6%, which brackets Xylem's 0.5–1.3%. The set is admissible under the growth-matched rule.

Load-bearing status: NOT load-bearing on any number in this memo. The exit multiple comes from the ROIC identity; the 12-month multiple comes from Xylem's own history. The set is recorded because the brief requires a separate comparator set per name and because it is what a reader should check the output against. It shares no ticker with the KGS or FSS sets.

The screen's exit_multiple_peer_n: 286 is the defect this replaces.

---

SUMMARY