Initiated 2026-09-22 · Price $111.45 (2026-09-21, stockanalysis.com) · Mkt cap $12.65bn · Health care / biotech royalties & drug delivery · Model: verified

Rating: Sell — Conviction: Medium

Halozyme is a superb business and, at a 52-week high after a 43% twelve-month run, a poor security. It collects royalties on the subcutaneous conversion of blockbuster biologics — 62% EBITDA margins, ~1% of revenue in capex, 32% ROIC, revenue guided up 34% this year — and trades at 11.4x forward earnings because the market is pricing the end of the patent estate those royalties rest on. That discount is not an error: the core rHuPH20 patent lapses 2027-09-23, and the Mdase portfolio meant to carry protection to 2034 is under a seven-patent inter partes review campaign from Merck, who has already won one. The Base case values the stock ~30% below today's price; only the Bull case, in which the estate substantially holds, justifies it, and the sell-side's own $99.56 average target also sits below the market. Medium conviction, not High, because the bull outcome is genuinely live.

Model value range vs price
Bear $24.40Base $78.10Bull $148.93Price $111.45

Business overview

Halozyme licenses ENHANZE, a recombinant human hyaluronidase (rHuPH20) that lets large-molecule biologics be injected under the skin in minutes instead of infused over hours, and collects royalties on partner sales. Royalty revenue was $307.7mm in Q2 2026 (+50%) and is guided to $1,220–1,245mm for the year against total revenue of $1,835–1,910mm; the rest is product sales ($129.6mm in Q2) and collaboration payments. DARZALEX SC, Phesgo and VYVGART Hytrulo drove 52% royalty growth in 2025; newer launches (Opdivo SC, Ocrevus SC, Rybrevant SC) grew 80% quarter on quarter. Earnings turn on exactly two things: how fast partners convert infused biologics to subcutaneous, and how long the patents last.

Bull case

  1. Subcutaneous conversion is a secular wave still early in its run — the newer launches grew combined royalties 80% quarter on quarter, and five new ENHANZE and Hypercon collaborations were signed through July 2026 (Vertex, Oruka, GSK, Incyte among them), against an original target of three for the year. Plays out if every large-molecule franchise eventually needs an SC presentation. Model: rev_growth
  2. A patent loss is not a cliff to zero — the Mdase estate is roughly 100 patents running to 2032 in Europe and 2034 in the US, and collaboration royalty terms run to the later of a contractual period or the last valid claim. Plays out if the estate substantially survives and contractual floors protect the rest. Model: rev_growth, terminal_growth
  3. The unit economics are close to ideal — royalties carry no cost of goods, capex is about 0.75% of revenue on trailing numbers, and Q2 2026 ran a 59.8% GAAP operating margin. Incremental royalty dollars fall almost entirely to cash. Model: ebitda_margin, capex_pct_rev
  4. The market is paying nothing for the option — 11.37x forward earnings for 34% revenue growth and 32% ROIC, while the company repurchased 4.8mm shares in Q2 at an average of $69.30 and funded capped calls to blunt convertible dilution. Plays out on any favourable litigation outcome. Model: not in the value range — share_change, peer_pe move EPS, net cash or a cross-check only (see model note)

Bear case

  1. It is a wasting asset with a dated fuse — US Patent 7,767,429, the core rHuPH20 composition claim, runs to 2027-09-23 even after 1,297 days of patent-term adjustment. Everything beyond that depends on the Mdase estate holding. Model: rev_growth, terminal_growth, exit_ev_ebitda
  2. Merck is winning the fight, not just picking it — Merck filed inter partes reviews against seven Mdase patents, and in May 2026 the PTAB issued a final written decision finding the challenged claims of US 11,952,600 unpatentable on written-description and enablement grounds. Halozyme's counter-suit over subcutaneous Keytruda has produced a German preliminary injunction, which is a jurisdictional win, not a validity one. Plays out if the remaining six go the same way. Model: rev_growth, exit_ev_ebitda
  3. FY2025's reported margin collapsed and this session could not fully explain it — EBITDA margin on a consistent definition went 62.3% (FY2024) → 39.9% (FY2025) → about 64% in Q2 2026. Two technology acquisitions and active litigation plausibly sit in that year, but the bridge was not sourced. Plays out if any of it turns out to be recurring. Model: ebitda_margin
  4. $2.15bn of convertibles against $231mm of cash — four series outstanding at 0–1.00% coupons, with the 2027 notes nearest. September 2026's $1.3bn issue at 1.50% shows the window is open now and shows the direction of travel: each refinancing resets the coupon upward, and this is a company that must keep tapping a market that closes precisely when sentiment on contested patents sours. Model: wacc

Valuation & balance sheet

Metric (definition) Current Own history (range or 5y avg) Peers Source, as-of
Forward P/E 11.37x trailing 32.95x RPRX 10.57x stats / peer_rprx, 2026-09-22
EV / EBITDA (TTM, third-party definition) 13.89x n/a (unverified) RPRX 26.79x — not comparable, different royalty accounting stats / peer_rprx, 2026-09-22
EBITDA margin (GAAP op. income + D&A + intangible amort. / revenue) 39.9% (FY2025); ~64% in Q2 2026 62.9% → 48.0% → 50.9% → 62.3% → 39.9% (FY2021–FY2025) n/a (unverified) edgar_opinc / edgar_da / edgar_amort / q2pr
FCF yield (TTM FCF / market cap) 6.37% ($806.2mm) n/a (unverified) n/a (unverified) stats / quote, 2026-09-22

Model-implied value range (from model-summary.json; generic module, DCF-Gordon and DCF-exit-multiple, midpoints): Bear $24.40 · Base $78.10 · Bull $148.93 per share, i.e. implied returns of −78.1% / −29.9% / +33.6% vs $111.45. These ranges show how the bull and bear drivers translate into value; they are not price targets. The market trades above the Base case and roughly a quarter of the way to the Bull — it is pricing the Mdase estate largely holding. Two assumptions carry this call and a reviewer should attack them first: terminal growth of 0.0% in the Base, which assumes new collaborations exactly offset decay in perpetuity, and an exit EV/EBITDA of 7.0x applied in 2030, when even on the company's own 2034 reading the US estate has four years left. Move either toward the Bull settings (2.0% and 10.0x) and the call changes — which is precisely the bet the market is making.

Balance sheet: the model prints net leverage of 3.44x and gross 3.85x, but that is computed on FY2025's depressed EBITDA and overstates current leverage — against 2026 guided adjusted EBITDA of $1,225–1,280mm the same $1.92bn of net debt is about 1.5x. Coverage 45.9x, flattered by near-zero convertible coupons. Nearest maturity is the 0.25% 2027 series ($209.0mm at 2026-06-30, ~$151.7mm since repurchased). Liquidity is $231.0mm of cash and Treasuries plus ~$806mm of trailing free cash flow. No agency ratings retrieved.

Model note: Built and LibreOffice-verified across all 3,188 formula cells; 0 failing error checks, 0 assumptions without a basis, no consistency CHECKs. One unverified input, material for one thing only: the $1,937.7mm long-term convertible balance could not be split by series, so it is all placed at 2031 and the maturity ladder should not be relied on here — the 1.00% 2028 series sits inside that lump. Two post-balance-sheet events are deliberately not modelled: the $1.3bn 1.50% convertible due 2033 priced 2026-09-17 and the related 2027/2028 repurchases, which extend maturities without changing net debt much. The FY2025 margin anomaly in bear case 3 is disclosed in meta.note and is the largest judgment in the file. Generator note (2026-09-24): in this module share_change, capital_return_pct and cost_of_debt move EPS and the net-cash path only, and peer_pe feeds a cross-check only; none changes the value range, so bull/bear tags no longer cite them. Scenario deltas were rebuilt after multiple_pct shocks were extended to the Gordon method (previously they moved only the exit-multiple half), and EPS now uses average diluted shares.

Scenario stress test

Reasoned from the bull/bear drivers; the model column comes from the scenario overlays (Base + shock). Halozyme is the most macro-insensitive name covered so far — royalties are a percentage of net sales of approved drugs, so most rows move the multiple and the discount rate, not the cash flows. The risk that governs this security is idiosyncratic and sits in the case construction, not in this table.

Scenario Effect Mechanism Magnitude Model Δ value vs Base ($/sh)
S1 Fast equity crash − Near a 52-week high with 13.11% short interest; de-risking hits crowded contested names hardest even when cash flows don't move Med −$10.37
S2 Slow bear / recession − Almost no fundamental channel — DARZALEX and VYVGART demand doesn't track GDP. Only the multiple moves Low −$3.64
S3 Rapid rate shock − The worst row: $2.15bn of converts issued at 0–1.00% must eventually reprice, and a rate shock removes exactly the convertible bid that produced 1.50% money in September High −$14.58
S4 Slow rate grind − Same refinancing and discount-rate channel, slower, across the horizon Med −$9.02
S5 Soft-landing cuts + Long-duration cash stream with convertible funding: helps the discount rate, the multiple and the next refinancing at once High +$11.05
S6 Recession-driven cuts + One of the few names here where this row is positive: drug royalties don't fall with employment, so rate relief arrives with no offsetting demand hit Med +$3.82
S7a Credit liquidity shock − A shut convertible window is the one financing risk that matters, with the 2027 series nearest; September's issue reduces but doesn't remove it Med −$4.85
S7b Slow default cycle − Weak channel — partners are large-cap pharma, not levered credits; the effect is sentiment and convertible pricing Low −$2.42
S8 Stagflation − Structurally inflation-protected on revenue (a percentage of net drug sales, no cost of goods), so the damage is entirely discount rate and multiple High −$11.74
S9a Dollar spike − Royalties on ex-US net sales translate lower; modest because partner reporting is largely USD Low −$1.38
S9b Dollar slide + The same effect in reverse Low +$1.39
S10 Melt-up + High growth, heavy short interest, near its high — well placed for a momentum melt-up, and cheap convertible issuance gets cheaper High +$11.44
S11 Energy supply shock 0 No material effect, not modeled — a royalty stream has no energy input, no freight and no commodity-linked demand — $0.00
S12 Mega-cap/AI derating − Minor: no AI exposure and a value multiple, but it carries a high-growth label and some crowding Low −$1.21

Magnitude labels re-ranked 2026-09-25 to match the model's dollar deltas (skill 3a-v-b item 7; the 2026-09-24 generator fix enlarged multiple-shock rows): S10 Med→High.

Currently active/on watch per the playbook: S3 partially active (price-level leg no longer met at a 4.96% 10Y as of 2026-09-21); S8, S10 and S11 on watch. S3 is this name's worst row and the playbook currently flags it; S11, its only blank row, is also on watch — a useful reminder that watch status and exposure are different things.

Model value change vs Base, by scenario
S3 Rapid rate shock−$14.58S8 Stagflation−$11.74S1 Fast equity crash−$10.37S4 Slow rate grind−$9.02S7a Credit liquidity shock−$4.85S2 Slow bear / recession−$3.64S7b Slow default cycle−$2.42S9a Dollar spike−$1.38S12 Mega-cap/AI derating−$1.21S11 Energy supply shock$0.00S9b Dollar slide+$1.39S6 Recession-driven cuts+$3.82S5 Soft-landing cuts+$11.05S10 Melt-up+$11.44

What would change the call

Upgrades if: the PTAB denies institution or upholds claims on a meaningful share of the six remaining Merck reviews; or a Merck settlement with a running royalty on subcutaneous Keytruda converts the largest litigation risk into the largest new revenue line; or partner disclosures confirm contractual royalty floors surviving patent expiry on the major franchises. Downgrades if: already Sell — conviction deepens if further PTAB decisions go against the Mdase estate, if a major partner publicly guides to a royalty step-down after 2027, or if the 2026 margin step-up proves to include items that reverse.

Watch items

Sources