Phase Space AI

Valuation

Ondas [ONDS]

Ondas Inc. [ONDS] — Valuation

as of 2026-07-29 · spot $6.80 (SIP close) · two outputs, both required

---

0. A twelfth acquisition closed 27 days ago and it is not in the screen record

On 2026-07-02 Ondas acquired 100% of High Point (from Highlander Partners Defense, LLC) under a Unit Purchase Agreement. 8-K filed 2026-07-06, Items 1.01 / 2.01 / 3.02 / 7.01 / 8.01. Consideration:

Lock-up: six months from closing, extended a further six months for only 50% of the Locked-Up Shares if the average of the thirty daily VWAPs over the thirty consecutive trading days immediately preceding 2027-01-02 exceeds $20.00 per share. Spot is $6.80. That condition will not be met on any plausible path, so the full 44,999,998 shares should be expected to become deliverable and resaleable from 2027-01-04. The sellers are additionally subject to a daily volume cap of 10% of the stock's daily trading volume, and Ondas agreed to file Rule 424(b)(7) resale prospectus supplements — which is what the weekly Item 3.02 8-Ks in the filing index are.

Every valuation input in the screen record predates this. The consequences:

High Point's revenue is not disclosed in the 8-K. Financial statements for a significant acquisition are due by amendment within 71 calendar days of the initial report's due date — on or about 2026-09-14 — and that filing is the single most informative dated event on this name (ONDS_Catalyst_Calendar.md).

Because High Point's revenue is unknown, every multiple below is computed on the $96.605m TTM revenue that excludes it. The multiples are therefore overstated to the extent High Point carries revenue, and that conservatism is stated rather than buried. It does not change the verdict: crediting High Point with even $100m of annualised revenue would leave EV/sales at 12.7x and the required CAGR at a 15% terminal margin and 18x exit at 50.6% — still far above anything the company has organically demonstrated.

---

1. Enterprise value — four defensible numbers, spanning 2.0x

(c) and (d) are the two internally consistent ways to capture the warrant claim and must not be combined — (c) values it at the market's own mark, (d) counts the shares it would become. Both say the same thing: the screen's 25.2x understates the true multiple by 42%–100%.

Fully diluted share build (construction (d)):

  535,762,648  outstanding after the High Point Immediate Shares (2026-07-02)
 + 44,999,998  High Point Locked-Up Shares, deliverable 2027-01-04
 +195,527,101  warrants to purchase common stock   (excluded from diluted EPS as anti-dilutive, 2026-03-31)
 +109,456,221  contingently issuable shares        (acquisition earn-outs, same source)
 +    654,350  options to purchase common stock
 ------------
  886,400,318  fully diluted

Net-cash derivation, from the 2026-03-31 balance sheet plus the one known subsequent cash movement:

  1,026,003  cash and cash equivalents
     11,081  restricted cash, current
    447,842  short-term investments  (a DISTINCT balance-sheet line — verified; contrast ASTS, where it is not)
  ---------
  1,484,926  gross cash and investments
       (243) notes payable, current
       (188) notes payable, non-current
       (528) convertible note payable, current
     (3,410) convertible notes payable, non-current
     (1,870) government grant liability, current
     (1,586) government grant liability, non-current
    (39,621) accrued purchase and contingent consideration, current
    (88,481) accrued purchase and contingent consideration, non-current
  ---------
  1,349,000  ≈ $1,352.5m on the reported roundings
   (200,000) High Point cash consideration, paid 2026-07-02
  ---------
  1,152,500  net cash used throughout this document

The $1,058,990k warrant liability is deliberately not netted here — it is share-settled, so it belongs in EV or in the share count, never in net debt. The screen deducted none of the items above and saw none of the warrant liability. Its net_cash_complete: true is false.

Other inputs:

---

2. Terminal margin is UNIDENTIFIED

Ondas has never earned an operating profit in any period in its filed history. FY2024 operating margin −481.2%; FY2025 −115.1%; Q1-2026 −85.1%. The absolute operating loss is widening: $(10.3)m → $(42.7)m year-on-year.

The screen assigned 9.1%, described as "industry median of mature profitable peers (pre-profit subject)". Rejected. It is the construction that produced the +51pp error on ADMA and the +32pp error on INOD, applied here to something that is not even a single business but a portfolio being assembled — and whose segment reporting collapsed to one line in the same quarter it closed six acquisitions.

Declared: terminal EBIT margin UNIDENTIFIED. A grid follows, not a number.

No exit anchor is shared with OUST or ASTS in this cluster. The three sit in different end markets (industrial sensing; Israeli counter-UAS and airborne ISR; LEO direct-to-device satellite), have different revenue models, and in ASTS's case there is no operating-margin history to anchor at all. Each name is gridded on its own.

The observable constraint on ONDS specifically: gross margin is 49.2% and rising (from 35%), which is genuinely good and does not cap the terminal margin at a low level. The binding constraint is operating expense — G&A alone is 86% of revenue. For a 15% terminal EBIT margin, total operating expenses must fall from 134% of revenue to 34%, a 100-percentage-point reduction. Not impossible for a company multiplying revenue; simply unevidenced — and $5.8m per quarter of that cost is acquisition due-diligence and legal fees that only stop when the roll-up stops.

---

3. The implied-path test

Method as assets/reverse_dcf.py. Solved for the revenue CAGR, with every held parameter named. 5 years, WACC 13.0%, TTM revenue $96.605m.

3.1 Construction (b) — corrected net cash, 535.76m shares. EV $2,491m (25.8x sales)

3.2 Construction (c) — + the $1,059m warrant liability in EV. EV $3,550m (36.7x sales)

3.3 Construction (d) — fully diluted, 886.40m shares. EV $4,875m (50.5x sales)

3.4 On estimated organic revenue — the honest base

57% of the reported top line was purchased rather than built, so the path that matters starts from what the company actually generates. 15% terminal margin, 18x exit:

3.5 The required parameter, with everything held fixed named

At a terminal EBIT margin of 15.0% and an exit multiple of 18.0x EBIT, discounting at 13.0% over 5 years, today's price of $6.80 requires:

Revenue base	Equity-claim construction	EV	Required 5-year revenue CAGR
Reported TTM $96.605m	(b) 535.76m shares, corrected net cash	$2,491m	77.4%
Reported TTM $96.605m	(c) + $1,059m warrant liability	$3,550m	90.5%
Reported TTM $96.605m	(d) fully diluted 886.40m shares	$4,875m	103.0%
Estimated organic TTM ~$42m	(b)	$2,491m	109.6%
Estimated organic TTM ~$42m	(d)	$4,875m	139.7%

Demonstrated: UNDETERMINABLE. Ondas does not disaggregate acquired from organic revenue by period, so the margin demonstrated − required cannot be computed. Per item D1 a missing input is INDETERMINATE, never FAIL — but the Valuation Criteria does not rest on that comparison alone. A required rate of 77% to 140% compound for five years exceeds every rate the company has demonstrated on any basis it does disclose, and exceeds the 60.4% this framework treated as extreme on OUST. FAIL.

3.6 Implied compression, stated as a number

Today: 25.8x TTM revenue on (b); 50.5x fully diluted. The (b) 15%/18x cell implies year-5 revenue of 96.605 × 1.774^5 = $1,691m, year-5 EBIT of $254m, exit EV $4,566m, discounted at 13% to $2,491m. Year-5 EV/sales = 2.7x.

Compression: 25.8x → 2.7x, a factor of 9.6x, or −90%. Fully diluted: 50.5x → 2.7x, a factor of 18.7x. For scale, the required path has Ondas reaching $1.69bn of revenue within five years — the size of a mid-cap defence prime — and still being worth only 2.7x sales.

3.7 Verdict

FAIL. The price requires 77%–140% compound revenue growth for five years. The evidenced organic driver is $11.4m of year-on-year growth at one subsidiary from two named products — Airobotics' Optimus System and Iron Drone Raider — which is real, and two orders of magnitude short of what the path needs. Everything else in the growth record was bought, and buying it consumed roughly $1.83bn of equity in fourteen months, created a $1.06bn warrant claim, and has now added 85.0m more shares plus $200m of cash for High Point.

PASS WITH ARGUMENT is unavailable. The argument required is "the roll-up will keep working and will integrate", which is a narrative. The one hard data point on integration runs the other way: a $46.2m loss expensed on the Indo Earth Moving acquisition in the quarter it closed, because consideration exceeded the fair value of the net assets acquired.

---

4. Sensitivity — on the exit multiple, and on the equity-claim definition

Per criteria.md, sensitivity runs on the highest-variance parameter. Terminal value is 100% of modelled EV here — Ondas generates no interim free cash flow — so the reverse DCF is mandatory as the primary long-horizon output and the sensitivity must be on the multiple, never on scenario probabilities.

On the exit multiple (15% terminal margin, construction (b)): required CAGR moves from 92.4% at 12x to 60.2% at 30x — a 32.2pp range.

On the equity-claim definition: required CAGR moves from 77.4% (construction (b)) to 103.0% (construction (d)) — a 25.6pp range arising from a definitional choice about the same balance sheet, at the same multiple.

That second number is the unusual one. On most names the definitional uncertainty is small and the multiple assumption dominates. On ONDS the two are comparable: how you count the warrants and the earn-out shares matters almost as much as what multiple you exit at. No amount of care about the exit multiple compensates for getting the share count wrong — and the screen's construction gets it wrong by the largest margin available, missing 350.6m shares of claim (886.4m fully diluted against its 495.8m).

Flip point

There is no flip point, because there is no demonstrated rate to flip against. Stated the useful way instead: the required path falls below 50% — a rate the company might plausibly claim — only at a 20% terminal EBIT margin, an exit multiple of 30x, on reported rather than organic revenue, with the warrant claim and the Locked-Up Shares both ignored. All five conditions must hold at once.

---

5. What the price is not paying for

Worth stating plainly, because the grids above can obscure it: corrected net cash of $1,152.5m is 31.6% of the $3,643m market capitalisation. At $6.80 an owner pays $2,491m for a business generating ~$42m of estimated organic revenue and losing $42.7m a quarter at the operating line, and receives $2.15 a share of cash — $1.30 fully diluted. The cash is the floor, and it is a long way down.

---

6. The 12-month target — the multiple anchor is UNIDENTIFIED

6.1 Why, with the evidence

valuation.md: "Anchor the multiple on the name's OWN trading range, with the current percentile stated… If the history is too short or spans a regime change, declare it UNIDENTIFIED rather than substituting a peer median."

Ondas's own EV/TTM-revenue history, from the filed balance sheets and the SIP close at each date:

Median 36.2x · min 31.2x · max 125.3x · today 25.8x — below every observation, the 0th percentile.

This history is uninterpretable and must not be used. Across five observations revenue went from $7.2m to $96.6m (13.4x), the share count from 105.7m to 535.8m (5.1x) and net cash from $10m to $1,152m (115x); twelve businesses were acquired and the founding subsidiary was deconsolidated. A "multiple range" computed across that is a range of different companies. The 2025-03-31 quarter is absent outright because its cash balance is not separately tagged in the aggregated XBRL.

Declared: 12-month target multiple UNIDENTIFIED. No peer median is substituted — that substitution is precisely what valuation.md forbids and what produced this framework's recorded errors.

6.2 What is offered instead, with its basis stated

A reference band, explicitly not a multiple-anchored target and not to be used as one. On 535.76m shares except where noted.

Reference band $2.15 – $7.78, midpoint ~$4.97 (−27%). Every figure rests on an assumed multiple, which is why it is a band and not a target.

6.3 Warning: the reported EPS will look excellent and will mean nothing

The stock has fallen 24.8% since the 2026-03-31 balance-sheet date ($9.04 → $6.80). The warrant liability moves inversely to the share price, so Q2-2026 will very likely report another large non-operating gain and a positive EPS. Q1-2026's headline was EPS basic $0.58 against an operating loss of $42.7m, with 100% of the profit non-operating and 108% of it the warrant remeasurement.

Any screen, target or model that reacts to that EPS is reacting to the share price, reflected back at itself through the income statement. Flagged in advance so the Q2 print cannot be mistaken for progress.