MannKind Corporation [MNKD] — Research

Tier-2 memo · analysis only · no position verdict. Every Criteria is scored PASS / FAIL / INDETERMINATE per criteria.md; the book decides.

0. The one-paragraph answer

The screen's headline — 51.8% demonstrated revenue CAGR against a 7.0% required CAGR, a +44.8pp margin — does not survive contact with the filings. 76.8% of MannKind's three-year revenue increase came from a single counterparty, United Therapeutics, and a further 9.3% came from an acquisition completed in October 2025. Organic own-product revenue contributed 13.9%. The United Therapeutics block is decelerating hard (royalty growth 42% → 25% → 9% over three periods) and MannKind's own 10-K carries a risk factor stating that United Therapeutics called its next-generation replacement product a "category killer" aimed at dry-powder inhalers, with a filing planned in 2026 and launch in 2027. Meanwhile the screen's balance sheet is wrong by a quarter of a billion dollars: it records +$57.4m of net cash against a filed position of $325.0m of principal borrowings versus $133.9m of cash and investments — $191.1m of net debt. Correcting the balance sheet and the share count moves the required 5-year CAGR from 7.0% to 11.8%, and correcting the demonstrated rate from the 51.8% launch-ramp fossil to organic TTM growth of +8.2% turns the screen's +44.8pp margin into −3.6pp. There is a real, named, dated mechanism here — Furoscix and the ReadyFlow autoinjector approved 23 July 2026 — but it is 11% of revenue being asked to carry a levered business whose other 64% is controlled by a partner that has publicly named its successor.

1. What the company actually is

Three revenue lines, three completely different qualities of revenue. This is the trap the brief asked to resolve, and the filings resolve it cleanly because MannKind discloses the split on the face of the income statement.

1.1 The verified split — three years plus the current quarter

Source: 10-K FY2025 (filed 2026-02-26) and 10-K FY2024 (filed 2025-02-26), Consolidated Statements of Operations; 10-Q Q1 2026 (filed 2026-05-06).

Three-year CAGR by line (FY2022 → FY2025):

The mechanism behind 51.8% is the launch of somebody else's product. Tyvaso DPI was approved in May 2022 and United Therapeutics began commercialising it in June 2022. FY2022 therefore contains roughly half a year of a product going from zero. 76.8% of the three-year revenue increase is United Therapeutics (royalty + cost-plus manufacturing), 9.3% is the scPharmaceuticals acquisition, and 13.9% is MannKind's own pre-existing products.

1.2 The royalty terms and duration — established

Source: 10-K FY2025, Note 11 (Collaborations, Licensing and Other Arrangements) and Note 16 (Commitments and Contingencies); MD&A "Trends and Uncertainties".

• Rate: 10% of United Therapeutics' net sales of Tyvaso DPI. MannKind recognises the full 10% as revenue.
• It only collects 9%. In December 2023 MannKind sold 1% of future net sales to Sagard Healthcare Partners Funding Borrower SPE 2, LP for $150.0m upfront. Because MannKind retains continuing involvement (it manufactures the product and defends the IP), the proceeds sit on the balance sheet as a Liability for sale of future royalties rather than as a gain, and — critically — MannKind keeps recognising the full 10% as revenue while remitting the 1% to Sagard.
• "The Company will continue to recognize the full 10% of future royalty revenues in its consolidated statements of operations, with the Sagard Royalty being non-cash revenue for the Company." (10-K FY2025, Note 16)
• Quantified by the company: $12.8m (FY2025), $10.2m (FY2024), $2.1m (Q4 2023) of reported royalty revenue is remitted to Sagard.
• Duration of the Sagard arrangement: net sales generated 1 Oct 2023 through 31 Dec 2042, after which the 1% reverts to MannKind. The 10-K discloses no expiry date for the underlying 10% UT royalty itself — that is a disclosure gap, and it is stated as such here rather than filled in. What is disclosed is that the royalty is consideration for an IP licence interdependent with the development work that supported Tyvaso DPI's approval.
• Sagard contingent milestones (potential cash IN to MannKind): $50.0m if Tyvaso DPI net sales reach $1.9bn in any 12 consecutive months on or before 31 Dec 2026; failing that, $45.0m if net sales reach $2.3bn in any 12 consecutive months on or before 30 Sep 2027. See §4.3 for why neither is likely.
• Manufacturing (the Commercial Supply Agreement, Aug 2021): MannKind manufactures Tyvaso DPI and UT purchases on a cost-plus basis; UT supplies the treprostinil at its own expense. This is a low-quality, volume-follows-the-partner revenue line, not a franchise.

1.3 What is inside "Collaborations and services"

Source: 10-K FY2025, Note 11.

The ~$12.2m/yr "recognition of previously deferred revenue" is amortisation of a prior upfront — non-cash in the current period. Deferred revenue on the balance sheet at 31 Mar 2026 is $11.525m current + $38.905m long-term = $50.4m, so this line has roughly four more years to run and then stops.

1.4 The other UT item: a $5.0m upfront in 2025

In August 2025 UT exercised an option under the licence (the "First Amendment") for MannKind to formulate MNKD-1501. MannKind received a $5.0m upfront in 2025, of which $1.0m was recognised as revenue in FY2025 and $4.0m sits in deferred revenue. It is eligible for up to $35.0m in development milestones, of which the company judges ~$10.0m probable and $30.0m constrained. This is genuinely lumpy revenue, but it is small — $1.0m of FY2025's $349.0m.

2. Accounting quality — is the reported growth real?

2.1 Headline: strip the acquisition and revenue is falling

Furoscix arrived with the scPharmaceuticals acquisition (closed 7 October 2025, ~$296.5m cash to shareholders plus $82.6m to extinguish scPharma's Perceptive debt). It contributed:

• $23.2m net revenue in FY2025 (the 7 Oct – 31 Dec stub; gross $32.4m, 28% gross-to-net)
• $15.5m net revenue in Q1 2026 (gross $20.9m, 26% gross-to-net)

On a TTM basis (Q2 2025 – Q1 2026 versus Q2 2024 – Q1 2025):

The screen's 51.8% "demonstrated CAGR" describes a period that has ended. The sequence of actual total-revenue growth rates is: FY2023 +99.4%, FY2024 +43.5%, FY2025 +22.2%, Q1 2026 +15.1%, Q1 2026 organic −4.7%.

2.2 Line-by-line quality of the current book

TTM (Q2 2025 – Q1 2026), derived as FY2025 − Q1 2025 + Q1 2026:

Recurring vs lumpy, quantified:

• Lumpy / milestone revenue is small. In FY2025 the only identifiable milestone- or upfront-type revenue is the $1.0m MNKD-1501 upfront recognition (0.3% of revenue). FY2024 contained a $1.1m Cipla regulatory milestone (0.4%). The CAGR is not milestone-driven — the brief's stated hypothesis is not what is wrong with this name.
• What is wrong is that the recurring revenue is other people's. 64.2% of TTM revenue is a royalty and a cost-plus manufacturing contract on a product MannKind neither owns nor sells.
• Non-cash / pass-through revenue: ~$25.3m of TTM revenue (7.0%) — $13.1m Sagard pass-through plus $12.2m deferred-revenue amortisation. Neither is cash earned in the period.

2.3 Afrezza: twelve years, no inflection

Afrezza was FDA-approved in June 2014. In the quarter just reported (Q1 2026) the 10-Q states:

"Gross revenue from sales of Afrezza decreased by $1.1 million, or 5% … The decrease was driven primarily by lower demand. The gross-to-net adjustment was 31% of gross revenue … compared to 36% … As a result, net revenue from sales of Afrezza increased by $0.4 million, or 3%."

Net revenue grew 3% only because gross-to-net improved by 5 percentage points, which is a non-repeatable lever (rebate reduction). Underlying demand fell. V-Go net revenue fell 23%.

2.4 Furoscix: the one thing that is genuinely growing — and one thing to watch

Derived from the 10-K FY2025 pro-forma disclosure (Note 3, Supplemental Pro Forma Information):

• Pro-forma combined FY2025 revenue $396,150k vs reported $348,966k → scPharma's 1 Jan – 6 Oct 2025 revenue was $47,184k ($169k/day).
• Pro-forma combined FY2024 revenue $321,836k vs reported $285,504k → scPharma FY2024 revenue was $36,332k ($99k/day).
• Underlying Furoscix growth ≈ +70% year-on-year into Q1 2026 ($172k/day vs $99k/day). Real, and the strongest fact in the long case.

But: the Q4 2025 stub ran at $273k/day ($23.2m over 85 days) and Q1 2026 ran at $172k/day ($15.5m over 90 days) — a 37% per-day sequential decline in the first full quarter MannKind owned it. Q1 payer resets are normal in US pharma and the Q4 stub may contain channel stocking around the acquisition, but this is the single number to check on the Q2 2026 print. Flagged as an open question, not a conclusion.

2.5 Accruals and earnings quality — clean

Accruals are negative (operating cash flow above net income) in every period — earnings are not being inflated relative to cash. That is a genuine pass. But SBC of ~$25.3m TTM exceeds TTM operating income of $14.8m, and operating cash flow has fallen from $42.5m to $19.3m while the company took on $325m of debt costing ~$29.5m/yr in cash interest.

2.6 Transcript / disclosure mention-frequency — required core metric

Counted directly from the Exhibit 99.1 earnings press release attached to every Item-2.02 8-K, 2019 → 2026 (30 releases, all primary source). Counts are normalised by nothing — raw mentions, with release word count shown for context.

Three findings, all evidenced:

1. Tyvaso DPI mentions peaked at 11–12 in H2 2024 and collapsed to 3–4, where they have now sat for six consecutive releases. Over the same six releases Tyvaso DPI went from ~60% to 64% of revenue. The largest and fastest-decelerating revenue driver has been de-emphasised in management's own communication exactly as its growth rate fell from 42% to 9%. This is the Twist pattern the brief names as precedent.
2. A new narrative was substituted in on schedule. Furoscix went 0 → 8 → 15 → 15 across the three releases following the acquisition; V-Go fell from 8–10 to 2–3.
3. "Tresmi" appears zero times in any earnings release — including the two published after United Therapeutics called it a "category killer" on 25 February 2026. It appears only in the 10-K risk factors. The single largest identified threat to 64% of revenue is disclosed where it is legally required and nowhere management chooses to speak.

3. Validating the screen's inputs against primary filings

Required by the brief. Every figure below is from the source named.

3.1 The net-cash error, in full

Verified balance sheet, 31 March 2026 (10-Q, Condensed Consolidated Balance Sheets — every line cross-footed to the stated totals):

MannKind has had negative book equity in every quarter since at least Q1 2021 (XBRL instant facts, 21 consecutive quarters). There is no reading of this balance sheet that produces +$57.4m of net cash.

Debt terms (10-Q Note 9, Borrowings): Blackstone senior secured term loan, $325.0m principal, matures 6 August 2030, bullet — no amortisation. SOFR (2.00% floor) + 4.75%, stepping to +5.00% if debt/adjusted-EBITDA ≥ 5.00x. Effective rate 9.09% at 31 Mar 2026 → ~$29.5m/yr cash interest against FY2025 operating cash flow of $18.3m and Q1 2026 operating cash flow of −$5.4m. Secured by a first-priority lien on substantially all assets including intellectual property. Covenant: maintain ≥$40.0m of unrestricted cash in lender collateral accounts, tested quarterly. Prepayment carries a make-whole through Aug 2026 then 3% / 3% / 1% step-downs.

3.2 The share count and the dilution the screen missed by five days

Five days before the screen's as-of date, MannKind closed a $50.0m private placement: 10,440,838 shares at $3.89 plus pre-funded warrants over 2,412,632 shares at $3.88 (exercise price $0.01, no expiry). Issued at a 0.6% discount to the then-market price — i.e. no premium, straight dilution. Stated use of proceeds: funding the $45.0m contingent-value-rights payment triggered by the FDA approval of Furoscix ReadyFlow on 23 July 2026. MannKind raised $50m of equity to pay a $45m obligation. Net new capital to the business: ~$2m, for 4.2% of the company.

Cumulative dilution 2021 → today: +30.2% (247.2m → 321.8m), ~5.1%/yr. Additional potentially dilutive securities at 31 Mar 2026: 3.0m RSUs + 3.5m options/PNQs = 6.5m (10-Q, EPS note) — small.

Cross-check net income ÷ shares ≈ filed EPS: Q1 2026 net loss $(16,619)k ÷ 308,267k weighted basic = $(0.054) vs filed $(0.05). ✓ FY2025 net income $5,863k ÷ 305,639k = $0.019 vs filed $0.02. ✓ Scale confirmed; MannKind is single-class, so the dual-class caveat does not apply.

4. Product-cycle intelligence — the named drivers

4.1 Furoscix and ReadyFlow (the long case)

• Furoscix (furosemide injection) — subcutaneous loop diuretic for fluid overload in heart failure / CKD, delivered by an on-body infusor. Acquired with scPharmaceuticals, 7 Oct 2025. Underlying growth ~+70% YoY.
• Furoscix ReadyFlow — FDA approved 23 July 2026 (8-K, Item 8.01). This is the autoinjector / Self-Dose format. It is the product whose approval triggered CVR Milestone 1.
• The acquired intangibles tell you what MannKind paid for: Developed technology — on-body infusor $185.7m and IPR&D — ReadyFlow Formulation $129.6m on the 31 Mar 2026 balance sheet. Total purchase price $421.6m. $129.6m of that was for ReadyFlow, which has now been approved — the asset is de-risked.
• This is a specific, named, dated, evidenced product cycle. It is the only one MannKind owns outright.

4.2 Tyvaso DPI and Tresmi (the bear case, and it is UT's own words)

From MannKind's FY2025 10-K risk factors — MannKind reporting on its own partner:

"On its February 25, 2026 earnings call, United Therapeutics highlighted the development of Tresmi, a treprostinil solution for use in a soft mist inhaler, describing it as a "category killer" designed to significantly reduce coughing — an acknowledged side effect of dry-powder inhalers — by up to 90% based on human studies, with plans to file for approval in PAH and interstitial lung disease within the year and launch commercially in the following year. Such public statements … indicate that United Therapeutics may choose to prioritize Tresmi or other pipeline products over Tyvaso DPI."

"If United Therapeutics reduces its commercial emphasis on Tyvaso DPI, diverts resources toward Tresmi … or if Tresmi, if and when launched, displaces Tyvaso DPI in the market, our revenues could decline materially."

MannKind earns nothing on Tresmi. Tresmi is not a dry-powder product and does not use MannKind's Technosphere platform.

The deceleration is already visible in the numbers, before Tresmi launches:

Implied Tyvaso DPI net sales (royalty ÷ 10%): FY2024 $1.023bn → FY2025 $1.281bn → Q1 2026 ~$0.327bn (~$1.31bn annualised).

And the collaboration line is already shrinking: −20.0% in Q1 2026, which the 10-Q attributes to "decreased product sold to UT due to timing of manufacturing activities."

4.3 The Sagard milestone will almost certainly not be paid

$50.0m requires Tyvaso DPI net sales ≥ $1.9bn in any 12 consecutive months ending on or before 31 Dec 2026. Current run-rate is ~$1.31bn. That is +45% in under six months against a line growing 9%. The fallback $45.0m requires $2.3bn by 30 Sep 2027 — +76%. Both should be treated as not happening unless the Q2 print shows a step-change. Any model carrying a $50m Sagard inflow inside 2026 is wrong.

4.4 Pipeline (company-stated dates, from the Q1 2026 10-Q)

• MNKD-201 (inhaled nintedanib, IPF): Phase 1b top-line data expected Q3 2026; global Phase 2 first patient enrolment expected Q2 2026.
• MNKD-1501: UT-funded formulation work; $5.0m upfront received 2025, up to $35.0m development milestones (~$10.0m probable), 10% royalty if approved.
• Cipla / India: Afrezza launch supported by an initial shipment in Dec 2025; $0.1m of product sales recognised in Q1 2026. Immaterial today.

5. Criteria scoring

Types are per criteria.md for the long-only absolute-return strategy. The memo issues no position verdict.

Quality Criteria — FAIL (BINDING)

Tested against both archetypes because the screen's COMPOUNDER label is contestable.

As COMPOUNDER (already profitable, steady growth, stable margins): - ROIC vs WACC — FAIL. NOPAT ≈ TTM EBIT $14.8m (no cash tax; full valuation allowance). Invested capital = term loan $318.7m + royalty liability $150.6m + financing liability $103.2m + equity $(59.2)m = $513.3m. ROIC ≈ 2.9% against a WACC of ~11.7% (§ Valuation). A compounder must redeploy above WACC; MannKind's most recent redeployment — $421.6m for scPharma — is earning ~$62m of revenue at a company-average operating margin of 4.1%. - Steady growth / stable margins — FAIL. Operating margin FY2024 25.4% → FY2025 11.1% → TTM 4.1%. Revenue growth 43.5% → 22.2% → 15.1% → −4.7% organic. - Accruals — PASS. Operating cash flow exceeds net income in every year shown.

As INFLECTION (high gross margin, thin margin expanding fast, growth accelerating): - Gross margin LEVEL — PASS. 75.6% TTM ex-amortisation. (Though flattered: royalties are 36% of revenue and carry no cost. Own-product gross margin is 73%; collaborations/services is 43%.) - Operating margin CHANGE — FAIL. Contracting ~7.0pp TTM vs FY2025, not expanding ~+5pp. - Revenue growth ACCELERATION — FAIL. Decelerating on every measure.

It fails both archetypes on the tests that discriminate. This is a FAIL on present, verified inputs — not an INDETERMINATE for missing data.

Valuation Criteria — PASS WITH ARGUMENT (BINDING) — see MNKD_Valuation.md

Required 5-year revenue CAGR at spot: 11.8% (narrow EV, WACC 10%, terminal margin 16.2%, exit 22.6x EBIT). Demonstrated, organic TTM: +8.2%. Margin = −3.6pp. The screen's +44.8pp does not survive.

Graded PASS WITH ARGUMENT rather than FAIL because the excess is attached to a specific, evidenced, dated product cycle (Furoscix + ReadyFlow, approved 23 July 2026, growing ~70%) rather than to narrative. The argument is real but unproven, and it requires Furoscix to roughly quintuple by FY2031 while the UT block, 64% of revenue, does not shrink.

Downside Criteria — MEASURED — logged, blocks nothing

Named cause: United Therapeutics displaces Tyvaso DPI with Tresmi, and MannKind's 64%-of-revenue block stops growing and then declines, against $325m of secured bullet debt.

• Not volatility. The permanent-impairment mechanism is: royalty (36% of revenue, ~100% incremental margin) and cost-plus manufacturing (28%) both track one product controlled by one counterparty that has publicly named its replacement, with a stated 2026 filing and 2027 launch. If Tyvaso DPI net sales decline 20% from the current $1.31bn run rate, MannKind loses ~$26m of royalty and a similar order of manufacturing revenue — essentially all of TTM operating income of $14.8m and more.
• A going-concern case must be argued explicitly, so: MannKind is not at going-concern risk today. The Blackstone loan is a 2030 bullet, the only covenant is a $40.0m minimum-liquidity test against $133.9m of resources (pro forma ~$136m), and there is no near-dated maturity. The FY2025 10-K and Q1 2026 10-Q both state management believes liquidity is sufficient for the next 12 months and longer. The risk is not a 2026 default; it is that a business generating $19.3m of TTM operating cash flow carries $29.5m/yr of cash interest and a −$59.2m equity book, so any revenue shock is funded by equity issuance at whatever price the market sets. The 23 July 2026 placement — $50m raised at a 0.6% discount, 4.2% dilution, to pay a $45m bill — is the template.
• Bear-case probability: 35% over 24 months (Tresmi files 2026, launches 2027, Tyvaso DPI growth goes to zero or negative). Bear price: ~$2.30 — the 52-week low, corresponding to ~2.5x EV/Sales on a flat-to- down revenue base, which is the bottom of MannKind's own five-year multiple range. This is an estimate, and per criteria.md it is logged to be Brier-scored, not to exclude the name.

Liquidity Criteria — PASS (BINDING)

• Equity: ample. Median consolidated volume over the last 63 sessions (Alpaca SIP feed) 4.08m shares / ~$15.1m per day. Any position this book would take is a fraction of one day's volume. (Note: the Alpaca IEX feed reports ~202k shares/day for the same period — 5% of consolidated. If any tool in this pipeline reads the IEX feed for liquidity screening it is understating volume ~20x.)
• Options: a real chain exists and was pulled. See MNKD_Trade_Construction.md for strike-level open interest and quoted size. Headline: Aug-21-2026 total OI 37,606; Jan-15-2027 total OI 31,891. The $4.00 Jan-2027 call shows OI 1,704 with 15×262 quoted size; the $7.00 shows OI 4,456 but only 1 contract on the bid. Fillable at small size on the wings only.

Momentum Criteria — MEASURED — timing input only

12-1 momentum +11.5% (price 12m ago $3.82 → 1m ago $4.26). 52-week range $2.28 – $6.21; spot $3.89 sits at the 41st percentile of that range. Realised vol 252d 77.0%, 63d 71.3%. Not a momentum leader and not broken.

Catalyst Criteria — MEASURED — see MNKD_Catalyst_Calendar.md

Dated events exist that would confirm or refute the implied path, principally the Q2 2026 print (Furoscix sequential run rate; UT royalty growth) and MNKD-201 Phase 1b top-line in Q3 2026.

Consensus Criteria — INDETERMINATE

Alpha Vantage EARNINGS_ESTIMATES returned the 25/day rate-limit message on 2026-07-29. No consensus was obtained. Per the brief this blocks nothing. No consensus figure has been invented anywhere in this memo; the NTM revenue base in the valuation is an explicitly-labelled house build.

Peer Spread Criteria — INDETERMINATE

No named peer in the same end-market was underwritten in this run. United Therapeutics (UTHR) is the counterparty, not a comparator, and comparing MannKind's multiple to UTHR's would be comparing a royalty recipient to the royalty payer. Declared unidentified rather than substituted.

Short Mechanism Criteria — PARTIAL PASS (MEASURED, acted on by nothing on this fork)

Decelerating growth: yes, unambiguously (43.5% → 22.2% → 15.1% → −4.7% organic). Exhausted margin runway: no — operating margin is 4.1%, not high-and-already-expanded; it is falling because of acquisition SG&A, which is a different thing. The classic short setup requires both; only one is present.

Sub-sector Criteria

Pharma (SIC 2834). Secondary tag: SMID Growth. Note for concentration purposes that MNKD's dominant exposure is to United Therapeutics' Tyvaso franchise, not to diabetes or to cardiometabolic broadly.

6. What is unsupported, missing, or an estimate

Stated explicitly per the brief's "never fabricate" rule.

1. No analyst consensus. Alpha Vantage quota exhausted. The NTM revenue figure in the valuation is a house build from disclosed run rates, labelled as such. It is not consensus and must not be read as one.
2. The expiry date of the 10% UT royalty is not disclosed in the FY2025 10-K or Q1 2026 10-Q. The Sagard 1% runs to 31 Dec 2042; the underlying MannKind royalty term is unknown to this memo. A licence term is material to a name where the royalty is 36% of revenue, and its absence is noted rather than assumed.
3. No Q2 2026 financials. The last filed balance sheet is 31 March 2026. Everything "pro forma" for the July placement and the $45m CVR payment holds the 31 March balance sheet otherwise constant and therefore excludes Q2 2026 operating burn, which is unknown. Q2 2026 results have not been filed as of 2026-07-29.
4. The Q2 2026 earnings date is not company-confirmed. MannKind filed its Q2 Item-2.02 8-K on 6 Aug 2025, 7 Aug 2024 and 7 Aug 2023. "Early August 2026" is a pattern inference, marked as such in the calendar. No date is asserted.
5. The ~$16m CVR remeasurement inference. Contingent consideration on the 31 Mar 2026 balance sheet was $29.0m ($23.877m current + $5.146m long-term) against a now-triggered $45.0m obligation. A fair-value remeasurement charge of roughly that difference should appear across Q2/Q3 2026. This is derived, not disclosed — the company has not stated it.
6. Afrezza / V-Go / Furoscix TTM splits are derived, not separately tagged. They are reconstructed from the MD&A's stated gross revenue changes, gross-to-net percentages and dollar amounts, and cross-checked to sum to the tagged Commercial product sales total ($33,907k in Q1 2026; $114,137k in FY2025). Individual product figures carry ±$1m of rounding.
7. The exit multiple of 22.6x EBIT is the screen's, retained only so the correction to the inputs can be read cleanly against the screen's own output. Its "GROWTH_MATCHED" basis over 130 peers is not verified by this memo and its growth-bracketing has not been checked. The valuation's own preferred anchor is MannKind's own EV/Sales history — see MNKD_Valuation.md.
8. Alpaca IEX vs SIP. All price, volume and volatility figures here use the SIP feed. The IEX feed returns ~5% of true volume for this name.

7. Sources

All primary. No third-party estimates were used anywhere in this memo.