Phase Space AI

Valuation

Harrow [HROW]

Harrow, Inc. [HROW] — Valuation

2026-07-29 · Spot $39.32 (close 2026-07-28) · Framework: Criteria (2026-07-29)

Two outputs are produced, over two horizons. Neither replaces the other.

These do not contradict each other and the difference is the point: the implied-path test asks what the current enterprise value requires over five years; the 12-month target asks where the tape lands as estimates revise. A name can rerate for four quarters on an H2 ramp while the five-year cash flows embedded in today's EV remain unsupportable.

---

1. Screen inputs verified against primary filings — three material discrepancies

Never silently adopt either number. All figures verified against the Q1-2026 10-Q (period 2026-03-31, filed 2026-05-11) and the FY2025 10-K.

Net debt, reconciled from the balance sheet, not from the screen. Harrow's debt is a single instrument: $300,000,000 of 8.625% Senior Notes due September 2030, upsized by $50m during Q1-2026. The maturity table shows nil in 2026–2029 and $300,000,000 in 2030. The balance sheet carries $292,087k = $300,000k less $7,913k of unamortised issuance costs.

Net debt = $300,000k face − $94,644k cash = $205,356k.

The screen used the carrying value net of issuance costs (−$197,923k). Unamortised issuance cost is not an economic reduction in what must be repaid; the face amount is. The screen understates net debt by $7.4m. Also excluded from both: $8,792k of operating-lease liabilities and a $7,000k accrued milestone.

Scale cross-check (net income ÷ shares ≈ filed EPS): FY2025 −$5,139k ÷ ~36.6m = $(0.14) = filed $(0.14). Q1-2026 −$27,602k ÷ 37.27m = $(0.74) = filed $(0.74). Share count confirmed. Not dual-class.

---

2. Terminal value share — the reverse DCF is mandatory here

At a 12.0x exit EBIT multiple, 16.2% terminal margin, 10% discount rate and the solved 38.8% CAGR, the discounted terminal value is the entire modelled EV — Harrow's interim free cash flow over the forecast window is negligible (TTM operating cash flow $15.2m; Q1-2026 −$9.0m; $25.9m of annual cash interest). Terminal value therefore exceeds 60% of EV by a wide margin and the reverse DCF is the primary long-horizon output, with a forward DCF demoted to supporting evidence. That threshold is not a judgement call here.

---

3. What "demonstrated" means for a roll-up — three numbers, all real

The strategy ranks on demonstrated − required. On a company that bought its revenue base, "demonstrated" requires disambiguation, so all three are reported:

Definition (b) is the honest choice: it is a real, achieved, twelve-month rate off the current base, it does not embed the one-time asset step-up in (a), and it does not over-weight the single distorted quarter in (c).

---

4. The exit multiple — growth-matched, and the screen's anchor does not survive it

The Criteria permit an exit multiple only from a comparator set whose growth brackets the subject's growth at the exit year. The required CAGR here is ~22–39%, so the set must bracket that. Peers pulled live from EDGAR companyfacts + Alpaca (2026-07-28), full data in data/peers.json:

(CORT and OCUL were pulled and are excluded as anchors: CORT's XBRL extraction returns an understated revenue and margin, and OCUL is effectively pre-revenue. Neither is relied on.)

A growth-matched set exists. ANIP (43.8%), COLL (23.6%), HRMY (21.5%) and ADMA (19.6%) bracket the required CAGR from both sides, are all US branded/specialty pharma, and all carry positive EBIT.

• Median EV/EBIT of that set: 9.4x (range 5.1–17.6x)
• Median EV/Sales of that set: 2.05x

Base exit multiple selected: 12.0x EBIT. This is above the growth-matched median of 9.4x and above three of the four anchors, credited for Harrow's 74% gross margin, its patented branded ophthalmic franchise and J-coded reimbursement. It satisfies the rule that the base may not sit below every stated anchor. It is not set to ANIP's 17.6x, because ANIP grew at 43.8% with a comparable margin and is the single most generous point in the set.

The screen's 22.5x is above every anchor in a properly-scoped set and is recorded as unsupportable. It was tagged GROWTH_MATCHED with peer_n = 160 — a 160-name comparator set cannot be growth-matched to an ophthalmic specialty-pharma roll-up in any meaningful sense; that is universe-wide, which is exactly the defect the Criteria document records against the "sector median that was universe-wide."

---

5. The implied-path test — the Valuation Criteria

Parameter solved for: revenue CAGR. Held fixed: terminal EBIT margin 16.2%; exit multiple 12.0x EBIT; discount rate 10.0%; horizon 5 years; revenue base $268.675m TTM; net debt $205.356m; shares 37.2751m; target EV $1,670.8m.

THE MARKET REQUIRES: a 38.8% revenue CAGR for five years.

Demonstrated: +26.2% (TTM YoY, definition (b)).

MARGIN = demonstrated − required = −12.6pp → FAIL.

The exit multiple used and the implied compression, stated as numbers. Terminal EV/Sales implied by 16.2% × 12.0x = 1.94x. Harrow trades at 6.22x EV/Sales today. The required path therefore embeds a compression of 4.28 turns of sales, −68.8%. On EBIT: today 85.0x TTM EBIT against a 12.0x exit = 73.0 turns, −85.9% of compression. Even at the screen's 22.5x the implied terminal EV/Sales is 3.65x, a 41% compression from spot.

Sensitivity over the exit multiple — the highest-variance parameter, per the Criteria. Not over scenario probabilities.

Read the bottom two columns together and the screen's headline dissolves. The screen's +23.1pp margin is produced by two choices stacked on each other: an exit multiple 2.4× the growth-matched median, and a demonstrated rate that is a bought step-change rather than a run-rate. Fix only the multiple and the margin falls to −0.4pp. Fix only the demonstrated rate and it falls to +3.8pp. Fix both and it is −12.6pp.

Secondary sensitivities (base 12.0x):

Using the built-up 12.3% WACC rather than the 10% screen convention makes the requirement worse, at 41.7%. The 16.2% terminal margin is roughly what management's own FY2026 guidance implies on a GAAP basis (~$90m adjusted EBITDA less ~$14m SBC less ~$21m D&A on ~$357m ≈ 15.4%), so it is not conservative — it is management's number, and it sits below the 23.5% peer median.

Result: FAIL. The price requires materially more than the business has demonstrated from its current base, and the argument for the excess — an unproven H2-2026 step-change — is not yet evidenced by a single quarter. It is not classed PASS WITH ARGUMENT because the argument's supporting evidence (a sales-force build, an in-office channel conversion, a promised 20–25% net-price improvement) has produced no realised revenue.

---

6. The 12-month target

Built per method: near-term consensus → named product-cycle events → the name's own multiple history with the percentile stated → expressed as a percentage to spot.

6.1 Near-term consensus (the base, not the answer)

Company guidance: FY2026 revenue $350–365m, adjusted EBITDA $80–100m, Q2 $71–81m. Street sits at or just below the low end on revenue while having cut FY2026 EPS by a third in ninety days.

Implied NTM (Jul-2026 → Jun-2027) consensus revenue ≈ $441m. The house does not adopt it: it embeds a Q4-2026 of roughly $132m, which is +48% above Harrow's all-time-record quarter of $89.1m — itself inflated by the IHEEZO channel load-in.

6.2 Named product-cycle events inside the window

All appear dated in HROW_Catalyst_Calendar.md: IOPIDINE 1% permanent J-code 2026-07-01; first IHEEZO retina data at ASRS, July 2026; Q2-2026 results ~Aug 2026 (guide $71–81m); IHEEZO multi-unit packaging and the stated 20–25% net-price improvement H2-2026; QUELL study data Q4-2026; FY2027 guidance ~Mar-2027. Working against: IHEEZO ASC pass-through already expired 2026-03-31, and the California Board of Pharmacy licence surrender (Jan-2026) is a permanent reduction to the compounding base.

6.3 The multiple — HROW's own trading history, percentile stated

EV/Sales computed daily from 2022-03-16 to 2026-07-28 (968 observations) using the TTM revenue that was actually published at each date plus the balance-sheet debt and cash of that quarter. Full series in data/evs_history.json.

The regime split at October 2023 is deliberate and declared: that is when the Novartis portfolio closed and Harrow became branded-led. By year: 2022 median 3.44x, 2023 5.96x, 2024 6.78x, 2025 6.87x, 2026 6.44x. The name is trading slightly below its own post-transition median — it is not, on its own history, expensive. This is why the 12-month target is above spot while the five-year implied path fails.

6.4 The target

Applied to TTM revenue as it will stand at the target date (July 2027), with net debt held at $205m and shares at 37.6m (modest SBC dilution):

12-month target: $50.00 — +27.2% above spot ($39.32).

• Bear assumes FY2026 lands near $285m (H2 at ~$165m — still above the FY2025 exit run-rate but far below guidance), IHEEZO's in-office channel does not fully replace ASC volume, and compounding keeps eroding. The multiple compresses to the full-history p25, which is above the all-time low of 2.71x.
• Base assumes FY2026 ~$320m, an ~8% shortfall to the low end of guidance, with 2027 growth resuming; the multiple simply holds at the name's own post-transition median.
• Bull is roughly consensus NTM ($441m) modestly beaten, at the own-history p75.

Sanity band vs the external reference. The house base case ($380m TTM at Jul-2027) sits ~14% below consensus NTM of $441m, and the house FY2027 view is ~27% below the Street's $517.5m. The divergence is explained, not asserted: the Street's number requires a Q4-2026 of ~$132m against an all-time record of $89.1m that itself contained a channel load-in. That is the variant view, and it is what the Consensus Criteria records.

Note that the target is above spot while the Valuation Criteria fails. That is the two-instrument design working as intended, not a contradiction: over twelve months a 66%-volatility name mean-reverts within its own multiple range far more reliably than five years of cash flow gets underwritten.

---

7. Downside Criteria — the permanent-loss case (MEASURED; logged, blocks nothing)

Named cause: the IHEEZO Medicare reimbursement reset compounding with an inflexible fixed charge.

Not a volatility figure. The mechanism: IHEEZO was 30% of FY2025 revenue. Its ASC pass-through expired 2026-03-31; Q1-2026 IHEEZO revenue was $1.85m against $35.88m in Q4-2025. Management's stated remedy is that the in-office retina channel "fully offsets" the lost ASC volume this year — an assertion with no realised quarter behind it. Simultaneously the compounding segment is −32.7% YoY with California's 503B and 503A out-of-state licences surrendered in January 2026.

If both continue, FY2026 revenue lands near $280m rather than $350m. At that level adjusted EBITDA falls well short of the $80–100m guide, GAAP EBIT is near zero, and $25.9m of annual cash interest is not covered from operations. Stockholders' equity is $28.7m against $181.4m of intangibles — tangible equity is −$152.7m — so an impairment of acquired product rights would take book equity through zero. The permanent-impairment path is a dilutive equity raise at a depressed price, or an intangible write-down that forces one.

• Bear price: ~$20 (−49%). ($280m × ~3.4x EV/Sales, near the 2022-era trough multiple, less $205m net debt.)
• Probability: 30%.
• Going concern: explicitly NOT argued. The $300m notes are a single 2030 bullet with nil due in 2026–2029, and Harrow held $94.6m of cash at 2026-03-31. This is an equity-impairment case, not a solvency case, and the distinction is stated so it can be scored honestly.

---

8. Peer Spread Criteria

Named same-end-market peer: Bausch + Lomb [BLCO] — ophthalmic pharmaceuticals and surgical, $5.1bn revenue, EV/Sales 2.10x. HROW at 6.22x trades at a 2.96x premium; HROW grows faster (26.2% vs 6.5%) and carries a much higher gross margin, so a premium is warranted — the question is 3x.

Named structural peer: ANI Pharmaceuticals [ANIP] — the closest analogue in kind: a leveraged specialty pharma that grows by acquiring branded product rights. FY2025 revenue $883m, +43.8% growth, 12.6% EBIT margin, EV/Sales 2.21x, EV/EBIT 17.6x. ANIP is growing faster than Harrow and trades at one third of Harrow's EV/Sales.

Own-history percentile: 44.8th (post-Oct-2023 regime), 55.9th (full history). Against peers HROW is expensive; against itself it is mid-range. Both are true and both are recorded.

---

Reverse DCF run with ~/.claude/skills/investment-memo/assets/reverse_dcf.py. Peer multiples: EDGAR companyfacts + Alpaca SIP closes, 2026-07-28. Own-multiple history: Alpaca SIP daily bars 2019-01-02 → 2026-07-28 joined to filed quarterly revenue, debt and cash. Consensus: Alpha Vantage EARNINGS_ESTIMATES, retrieved 2026-07-29.