Phase Space AI

Valuation

Halozyme Therapeutics [HALO]

Halozyme Therapeutics [HALO] — Valuation

2026-07-29 · framework: Criteria, 2026-07-29 Two outputs are mandatory and both are produced.

---

Part A — Inputs, verified against primary filings

Convertible-note treatment, stated. HALO carries $2,179.6m of convertible principal across four series: 2027 ($209.6m, conversion $77.17), 2028 ($470.0m, conversion $56.02), 2031 ($750.0m, conversion $87.20) and 2032 ($750.0m, conversion $87.20). At $83.62 the 2027 and 2028 series are in the money. The treatment used here — GAAP diluted shares plus full debt at carrying value — is the conservative one and slightly overstates EV if the in-the-money series convert to equity. It is stated rather than optimised.

Recency asserted. Latest 10-Q filed 2026-05-11 for the period ended 2026-03-31 — 79 days old. FY2025 10-K filed 2026-02-17. HALO's XBRL companyfacts is current (max filed 2026-05-11), unlike ALNY's.

TTM used, never last-FY. Last-FY revenue ($1,396.6m) would understate the base by 7.4%.

---

Part B — The implied-path test (the Valuation Criteria)

B1 — The exit multiple, its basis, and the argument for discounting it

Step 1: establish the growth-matched anchor. HALO's projected FY2031 revenue growth is in the 10–18% range (decelerating from the guided 22–30% in FY2026, with the 2027/2029 royalty step-down offsetting the Wave-2 and early Wave-3 ramps). The comparator set below spans 1.0% to 21.4% and brackets it.

Computed 2026-07-29 from Alpaca prices and EDGAR FY2025 filings, one method, one date:

Step 2: argue the discount, separately and explicitly.

valuation.md: "The base exit multiple may not sit below every stated anchor without a separately argued reason." The base used here — 15.0x EBIT — sits 45% below every anchor. Here is the argument, and it is not a feeling about biotech:

All three anchors own franchise exclusivity that extends past the 2031 exit year. HALO does not. The 10-K states that HALO's US rHuPH20 patent expires in 2027 and its European patent in 2029; that the royalty rate is reduced in a country upon that expiry where no collaboration-specific patent covers the product; and that "Janssen's patents covering DARZALEX SC do not impact the timing for this royalty reduction." DARZALEX is 34.6% of total revenue. A multiple applied at a 2031 exit year is therefore being applied after the step-down on the largest revenue line, in both the US and Europe. A comparator whose exclusivity is intact at the exit year is not pricing the same instrument.

This is not the NTRA defect. NTRA named anchors of 7.0x and 7.5x and set base at 5.0x as a silent distant-year haircut on a multiple that had already fully mean-reverted. Here the anchors are current market multiples for intact-exclusivity assets, the discount is to a specific, filed, dated contractual event, and it is argued in the open with the sensitivity carrying the alternative.

Basis label: ARGUED_DOWN_FROM_GROWTH_MATCHED (anchor 27.4x GROWTH_MATCHED; base 15.0x). The 27.4x case is run in the sensitivity and reported.

B2 — The terminal margin, and its basis — the screen's number is corrected here

Base terminal EBIT margin: 50%. Below the clean FY2025 figure (57.5%), below the FY2026 implied (~55%), and above the clean Q1-2026 (49.0%) — deliberately, because the royalty rate step-down removes near-100%-margin revenue and the Hypercon/Surf Bio investment (~$60m in 2026 per guidance) is an ongoing operating cost against pre-2030 revenue.

The screen's 33.6% is used as the bottom of the sensitivity, so the effect of the correction is visible rather than assumed.

B3 — The result

python3 reverse_dcf.py --spot 83.62 --shares 122.875 --net-cash -1823.757 \
    --revenue 1388.2 --years 5 --wacc 0.10 \
    --terminal-margin 0.50 --exit-multiple 15.0 --hist-cagr 0.2836

On the total TTM base ($1,508.5m, including $120.3m of trailing milestones) the requirement is 11.5% and the margin is +16.9pp. The recurring run governs.

B4 — Implied compression, stated as a number

The price requires the sales multiple to compress 14% over five years while revenue compounds at 13.4%. That is a coherent, non-heroic path.

B5 — Sensitivity, run over the exit multiple (never over scenario probabilities)

Recurring base $1,388.2m, terminal margin 50%, WACC 10%, 5 years.

The flip point is 8.0x EBIT — equivalently a 4.0x exit sales multiple against 8.7x today, a 54% de-rating. Below that, today's price requires more than HALO has demonstrated and the Criteria fails. That is where the judgement actually lives, and it is stated as a number rather than buried. A book that believes the post-2027 royalty step-down halves the terminal sales multiple should regard this name as fully priced.

Second sensitivity, over the terminal margin, at the 15.0x base:

The terminal-margin correction is worth 9.3pp of required CAGR. This is the single largest input correction in either of tonight's two names and it runs in the direction of making the name look better, not worse — which is exactly the kind of screen error that goes unchallenged.

B6 — What could break the PASS

Two things, and both are dated:

1. A quantified royalty step-down. If HALO ever discloses the magnitude of the post-2027 rate reduction and it is severe, the correct exit multiple moves toward the 8.0x flip point and the Criteria fails. The disclosure would most plausibly come in the FY2026 10-K (February 2027) or in 2027 guidance (January 2027), as the 2027 US expiry enters the guidance window.
2. An adverse USPTO post-grant review on the MDASE patents. That removes both the Merck optionality and the 2032–2034 estate, and it is the event that most directly justifies an 8.0x terminal multiple.

---

Part C — The 12-month target

Built per valuation.md: near-term estimates, named product-cycle events, and the multiple anchored on HALO's own trading range with the percentile stated.

C1 — The near-term revenue base

FY2026 is management's own guidance (2026-02-17, reiterated 2026-05-11):

Management also states the expectation that ENHANZE royalty revenue exceeds $1 billion for the full year, and that guidance includes approximately $60m of new Hypercon and Surf Bio investment.

FY2027 is a house estimate — not guidance, not consensus:

TTM revenue reportable twelve months from now (through Q2-2027) ≈ (1,760 + 2,000) / 2 = $1,880m.

C2 — Named product-cycle events inside the 12-month window

Each appears in HALO_Catalyst_Calendar.md with its source.

1. Q2, Q3-2026 results — the test of the +22–30% revenue guidance and of whether bulk rHuPH20 sustains its +172% trajectory.
2. The $1bn share-repurchase programme, authorised to 2028-12-31, with at least $400m projected in 2026 — roughly 4% of the share count at today's price. This is a dated, funded, disclosed capital action and it is the single most reliable EPS driver in the window.
3. Target of 13 ENHANZE products in development by year-end 2026 (nine at Q1-2026).
4. Stated goal of three new SC-delivery-platform deals in 2026 — already exceeded with Vertex (April), Oruka (May) and GSK (May).
5. TAK-881 (Takeda) — positive Phase 2/3 in primary immunodeficiency reported May-2026; a filing and approval would make it the 11th ENHANZE product launch.
6. Merck litigation and the USPTO post-grant reviews — no date is disclosed, so this is an undated catalyst and is treated as such.

C3 — The multiple, anchored on HALO's own history with the percentile stated

EV/TTM-sales computed daily from 2022-01-01, using time-varying share counts and net debt by fiscal year, against the TTM revenue reportable at each date with a 45-day filing lag:

Base multiple chosen: 8.0x — today's multiple, held flat. No re-rating is assumed in either direction. Given that the name is simultaneously at a 52-week high and at the 19th percentile of its own multiple range, holding the multiple flat is the neutral assumption and the one least likely to be doing the work.

C4 — The target

12-month target: $107.5, +28.6% above spot.

Cross-check on the name's own earnings multiple. Today's price is 10.5x the FY2026 guided non-GAAP diluted EPS midpoint of $8.00. The base target of $107.5 against a FY2027 non-GAAP EPS estimate of roughly $9.75 (operating leverage on ~$2.0bn of revenue plus a ~4% reduction in the share count from the funded buyback) is 11.0x — essentially no multiple expansion. The target is revenue growth plus buyback arithmetic, which is the most defensible kind.

Note the asymmetry the bear case reveals. At a multiple below anything HALO has traded at in 4.5 years, the stock is still up 3%. The 12-month downside is protected by the growth in the denominator. The 5-year downside is not — that is the royalty step-down, and it lives in the implied-path test's 8.0x flip point, not here. These two instruments genuinely disagree about this name, and that disagreement is the finding: the next twelve months are underwritten by disclosed guidance and a funded buyback; the five-year terminal value is underwritten by an undisclosed royalty rate.

C5 — Sanity band against an external professional target

Not computable. No paid feed is connected; Alpha Vantage's quota was exhausted by parallel agents at run time (verified). The gap to a professional's target is stated as unavailable rather than estimated.

---

Part D — Reconciliation to the screen

Net effect of the corrections: the terminal-margin fix makes the name look better (+9.3pp of margin); the exit-multiple argument makes it look worse (−14.4pp). They very nearly cancel, and the memo's +15.0pp lands close to the screen's +18.9pp by a different and fully stated route. That coincidence is not reassurance — two compensating errors produce the same answer as none — which is why both adjustments are itemised rather than netted.