Neurocrine Biosciences, Inc. [NBIX] — Tier-2 Research

Analysis date: 2026-07-29 · Framework: Criteria, 2026-07-29 (investment-memo v1.7.0) Price used: $182.13 (Alpaca IEX, 2026-07-29 intraday) This document issues no position verdict. It scores the Criteria and reports evidence.

0. The headline finding, stated first

Neurocrine completed a $2.9bn all-cash acquisition of Soleno Therapeutics on 18 May 2026, funded partly by the first drawn debt in the company's recent history — a $600m draw on a new $1.0bn secured revolver. The Tier-1 screen has no file for NBIX in this worktree and, whatever vintage it used, its net-cash and enterprise- value inputs predate the deal. Every valuation number that treats NBIX as a $2.65bn-net-cash company is now wrong.

The deal is confirmed by filings, not inferred: NBIX 8-K of 2026-05-18 (Items 2.01 and 2.03), and Soleno's own Form 15-12G deregistration filed 2026-05-28.

1. Screen-input validation

Re-derived from primary filings (SEC EDGAR XBRL companyfacts CIK 0000914475; FY2025 10-K filed 2026-02-11; Q1 2026 10-Q filed 2026-05-05; 8-Ks of 2026-04-06, 2026-05-05, 2026-05-18, 2026-05-28). Latest filing is 85 days old — current.

No fabricated share count and no skipped-quarter TTM here. NBIX's Tier-1 inputs are, on the whole, sound. The material gap is that the screen values a balance sheet that no longer exists.

1.1 The balance sheet, before and after Soleno

Pro-forma build, each component from a filing: $2,647.2m starting cash − ~$2,900m Soleno equity consideration (NBIX's own stated "total transaction equity value of $2.9 billion") + $600.0m revolver draw + $529.0m of Soleno cash and marketable securities at 2026-03-31 ($133.0m + $396.0m, Soleno 10-Q) − ~$50m Oxford Financing term loan (assumed repaid at close) = ≈ $826m gross cash against $600m drawn debt ≈ $226m net cash.

This is an estimate and is labelled as one. It excludes NBIX's own operating cash generation between 2026-04-01 and the 2026-05-18 close (Q1 2026 free cash flow was $136.7m, so roughly $70m for the stub) and any transaction costs. The Q2 2026 10-Q, due within days, replaces it with a filed number. No pro-forma figure is presented as filed.

Separately: NBIX carries $415.3m of non-current operating lease liabilities plus a current portion, largely relating to a vacated legacy campus for which it books impairment and sublease costs. Treated as debt-like, this adds roughly $470m to enterprise value. Not included in the headline EV; disclosed so a book can add it.

2. Mechanism — the specific, evidenced driver

The mechanism is a deliberate, funded de-concentration away from a single product that is still 84.6% of trailing revenue, executed across three legs, with each leg's revenue already printing.

For a decade NBIX was one drug. In FY2023, INGREZZA was 97.3% of revenue. That is now changing fast, and each leg is measurable:

Leg 1 — CRENESSITY, a first-in-class launch that is already $600m annualised

CRENESSITY (crinecerfont) was approved December 2024 as the first FDA-approved treatment for classic congenital adrenal hyperplasia. The ramp, quarter by quarter:

FY2025 total $301.2m in its first full launch year. Reimbursement: ~80% of dispensed prescriptions covered in both Q4 2025 and Q1 2026 (company-stated), with 2,048 total new-patient enrolment start forms in FY2025 and 431 in Q4 alone. Patent protection to 2035–2041, plus seven years of FDA orphan-drug exclusivity. Addressable population: "CAH affects at least 20,000 people in the U.S."

The honest caveat, stated up front: the sequential add halved in Q1 2026 (+$17.9m vs +$37.2m). Q1 is the US pharma gross-to-net reset quarter, so some of that is seasonal — but the launch curve is decelerating and this is the first quarter it has done so.

Leg 2 — INGREZZA re-accelerated in Q1 2026, and it is a volume story with a price offset

INGREZZA growth collapsed from +26.0% to +8.7% across FY2025 and then printed +20.5% in Q1 2026. The company attributes it to "double-digit prescription volume growth in TRx and record NRx driven by strong patient demand, partially offset by a lower net price compared to the first quarter of 2025."

The mechanistically important detail is what did not happen: the usual Q1 step-down. Q1'25 fell 11.4% sequentially from Q4'24 ($615.2m → $545.2m). Q1'26 was flat on Q4'25 ($657.5m → $656.9m). On a two-year stack, Q1'26 vs Q1'24 is +29.8%, or +13.9% annualised — which is the honest run rate, materially below the headline +20.5% and materially above FY2025's +8.7%.

Price is going the wrong way and volume is more than covering it. FY2025's deceleration was explicitly "a lower net price due to new formulary access investments to support long-term growth" — i.e. NBIX bought formulary position with gross-to-net. That investment appears to be converting: Q1 2026 volume growth is double-digit with record new prescriptions while price remains a drag.

Leg 3 — VYKAT XR, bought for $2.9bn on 18 May 2026

Soleno's VYKAT XR (diazoxide choline ER) was FDA-approved 26 March 2025 for hyperphagia in Prader-Willi syndrome — the first approved therapy for that indication. Its full commercial history:

FY2025 total $190.4m; trailing twelve months to Q1'26 $285.0m. Soleno was profitable — net income of $31.4m in Q1 2026 on $94.6m of revenue.

NBIX paid ~10.2x trailing sales, and the last observable data point before it closed showed sequential growth of 3.2%. That is the finding, and §3.3 treats it as an accounting-quality and capital-allocation matter rather than a product one.

What ties the three together

All three are rare/orphan or under-penetrated CNS-endocrine franchises sold to a specialty prescriber base that NBIX already covers, and all three are small-molecule, high-gross-margin (FY2025 cost of revenues was $52.1m on $2,860.5m — 98.2% gross margin). The commercial infrastructure is the reusable asset; management expanded the INGREZZA and CRENESSITY sales teams in Q1 2026 and named the expansion as the SG&A driver.

Behind them sits the largest R&D build in the company's history — see §4.

3. Accounting quality

3.1 Is the reported growth real? Yes — this is the cleanest revenue in the batch.

• 98.2% gross margin, no manufacturing, no channel.
• Collaboration revenue is trivial and shrinking as a share: $26.6m in FY2025 on $2,860.5m of total revenue (0.9%), being AbbVie royalties on elagolix and Tanabe royalties on valbenazine. Revenue is 99.1% product sales.
• No milestone, settlement or one-time revenue anywhere in the top line. Checked explicitly against the 10x Genomics precedent named in the brief: NBIX's revenue note disaggregates into INGREZZA, CRENESSITY and "other revenues" only, and other revenues are the two named royalty streams.
• No distributor concentration on the NBIX side. (Soleno, by contrast, "contracts with one customer, its specialty pharmacy" and "accounts receivable, net are solely from sales of VYKAT XR, and are from this sole customer" — a single-customer structure that now sits inside NBIX. That is normal for a rare-disease launch, but it is a real concentration and it is recorded here.)
• Accruals: FY2025 operating cash flow $782.7m against net income $478.6m. Cash exceeds earnings by $304.1m. Clean.

3.2 The real finding is not revenue quality — it is that operating leverage went negative in FY2025

Revenue grew +21.4% in FY2025. Non-GAAP diluted EPS grew +0.9%.

GAAP EPS rose 42% while non-GAAP EPS was flat, because FY2024 GAAP carried a $138.4m charge on convertible senior note settlements that did not recur. The GAAP series overstates the improvement; the non-GAAP series is the one that describes the business, and it says NBIX spent an entire year of revenue growth on R&D.

R&D at 35.5% of revenue in FY2025 (from 31.0%) is a deliberate, disclosed choice — two simultaneous Phase 3 programmes (osavampator in MDD, direclidine in schizophrenia) plus $65.4m of development milestones and $17.4m of acquired IPR&D. FY2026 guidance takes GAAP R&D to $1,200–1,250m and SG&A to $1,375–1,400m: combined opex of ~$2.6bn against an INGREZZA guide of $2.7–2.8bn. The company is running a build, and the build is funded out of the P&L rather than the balance sheet.

Q1 2026 is the first quarter that suggests the leverage turns: non-GAAP EPS $1.94 vs $0.70, non-GAAP operating income $226.4m vs $78.8m. But Q1'25 was depressed by a $37.5m osavampator Phase 3 initiation milestone, so the comparison is against an unusually weak base. Milestone expense was $22.6m in Q1'26 vs $45.4m in Q1'25.

3.3 Two non-recurring items in Q1 2026 GAAP net income — both correctly excluded from non-GAAP

Q1 2026 GAAP net income of $197.9m includes: - +$28.6m pre-tax gain, net of transaction costs, on the sale of Neurocrine Group Limited to Immedica Pharma AB for $63.2m cash, completed 21 January 2026. "The sale did not qualify for discontinued operations presentation." - +$25.3m unrealised gain on equity investments (vs a $30.6m loss in Q1'25 — a $55.9m swing).

Together $53.9m, or 27% of GAAP net income. Management excludes both from non-GAAP, correctly, so non-GAAP diluted EPS of $1.94 is clean of them. A screen reading GAAP EPS of $1.91 (+2,288% YoY) is reading a divestiture and a mark-to-market.

3.4 The Soleno purchase — the capital-allocation question

NBIX paid $2.9bn for an asset with $285.0m of trailing revenue (10.2x sales) whose last three sequential growth prints before close were +102%, +39%, +3.2%. The 3.2% quarter (Q1 2026) was filed on 2026-05-07 — eight days before the tender offer expired and eleven days before close.

Two further facts, both from filings: - On 7 April 2026 — one day after the merger was announced — Soleno "voluntarily withdrew its marketing authorization application (MAA) for VIOKAT prolonged-release tablets" from the EMA, a decision that had been expected mid-2026. The ex-US opportunity was removed from the asset the day after the deal was announced. - 88.9% of Soleno shares tendered; the merger closed under DGCL §251(h) without a shareholder vote.

Whether this is a good deal is not determinable from the file — Q1 is the gross-to-net reset quarter for US pharma, and a rare-disease launch fifteen months in is early. But the price paid, the deceleration visible at the time of payment, and the EU withdrawal are all facts, and they are the reason the pro-forma implied-path test in NBIX_Valuation.md is run as the primary case rather than the pre-deal one.

Purchase accounting has not yet been filed — no allocation, no goodwill figure, no intangibles. The Q2 2026 10-Q will carry it. Watch specifically for the size of the inventory fair-value step-up, which is exactly what distorted GMED's FY2024 margins in this same batch, and for the intangible amortisation schedule.

4. Product-cycle intelligence — the pipeline is the second half of the thesis

Two structural points a screen cannot see:

1. The osavampator economics changed materially in January 2025 and it went the wrong way. The original 2020 Takeda agreement made osavampator a 50/50 profit-share. The Restated Agreement converted it to a royalty-bearing licence with tiered royalties in the mid-to-upper teens in the US, and returned Japan rights to Takeda. NBIX now bears all US development cost and pays a mid-to-upper-teens royalty on a Phase 3 asset in the largest CNS indication there is. That is a worse deal than a profit share if the drug works, and a better one if it does not. No memo should model osavampator at full economics.
2. NBIX terminated the DAAO programme (luvadaxistat, NBI-1065846) in October 2024, effective April 2025 — a disclosed pipeline failure, and the reason R&D reallocated toward AMPA and muscarinic.

INGREZZA's patent position is the single most important fact in the file for a five-year test. 22 Orange Book-listed US patents expiring 2027–2040; US Patent 8,039,627 covering valbenazine extended by 552 days to 2031. Critically: "In 2023, we entered into settlement agreements resolving all patent litigation brought by us against the companies that filed ANDAs... such companies have the right to sell generic versions of INGREZZA in the U.S. beginning March 1, 2038, or earlier under certain circumstances."

A five-year implied path (to 2031) sits entirely inside settled exclusivity. The live exposure is a different filer: Zydus filed a Paragraph IV ANDA against INGREZZA SPRINKLE in March 2025, is not party to the 2023 settlements, and is in litigation in D. Delaware. That is the named downside cause in §7.

5. Transcript mention-frequency (required core metric)

Corpus: SEC EDGAR 8-K Exhibit 99.1 quarterly earnings releases, 13 consecutive quarters, 2023Q1–2026Q1. Alpha Vantage EARNINGS_CALL_TRANSCRIPT was unavailable (shared 25/day quota exhausted on 2026-07-29; the API returned its rate-limit notice). Per references/mention-frequency.md the first-party EDGAR release corpus is the designated substitute — one source, used consistently across the whole series. Word counts range 2,583–3,516, so normalisation matters; all figures are per 10,000 words with raw counts shown where the signal is load-bearing.

The generative reads — run before the valuation work, as the method requires

1. schizophrenia is the cleanest emerging signal in the corpus. 3.9 → 12.2 per 10k, and the step change is sharp and sustained: 6.2 in 2024Q4 → 10.2, 12.6, 12.8, 12.5, 12.2 across the next five quarters. Raw counts double from 2 to 4 and hold. This is not a one-quarter spike. Independent corroboration exists and it is a filing, not a mention: direclidine entered Phase 3 in schizophrenia, NBI-1117570 entered Phase 2 with first patients dosed in March 2026, and a $22.5m milestone was expensed to R&D in Q1 2026 specifically on that Phase 2 initiation (10-Q Note). Management attention, clinical stage and cash all moved together.

2. Soleno, VYKAT and Prader-Willi are first-ever mentions in 2026Q1 — an entire franchise appearing from zero. Raw counts 8 / 3 / 2 in a single release. Corroborated by the merger agreement 8-K, the tender offer and the closing 8-K. In a 13-quarter series that is the largest single-quarter vocabulary change in the corpus.

3. INGREZZA's share of management attention is at its 13-quarter low exactly as CRENESSITY, Soleno and schizophrenia rise. INGREZZA 42.7 per 10k in 2026Q1 against a 13-quarter mean of 53.4 and a peak of 66.2. This is the de-concentration thesis (§2) showing up in the language before it shows up fully in the revenue mix.

4. A new commercial vocabulary appears from 2025Q2 and persists: net price, volume, prescriptions, NRx, TRx — and AUSTEDO, the direct competitor, named for the first time in 2025Q4. A company that starts reporting script metrics and naming its competitor is defending a franchise on volume because price is going against it. Corroborated directly by the filings: FY2025 INGREZZA growth of +8.7% is explained as "lower net price due to new formulary access investments", and both 2025Q4 and 2026Q1 carry head-to-head INGREZZA-vs- AUSTEDO clinical data in the release. This is a genuine competitive-intensity signal, and it is the one that cuts against the long case.

5. obesity and biologics are first-ever mentions in 2025Q4 / 2026Q1 — corroborated by the disclosed Phase 1 initiation of NBIP-'2118, a CRF₂ peptide agonist, and by the appointment of a Chief Technical Operations Officer to lead "expansion beyond small molecules into biologics and device-based therapies." Two independent records for a term that appeared from nothing.

What is NOT a signal, flagged because it looks like one. crinecerfont collapses from 25.6 to 3.0 per 10k across 2024Q3→2024Q4. That is not de-emphasis — it is the switch from the molecule name to the brand name CRENESSITY at approval, and the two series are near-perfect mirror images. Reading it as a decay signal would be exactly the error references/mention-frequency.md warns about.

Honesty constraints. Raw counts for the emerging terms are 1–8. Only Soleno (8), INGREZZA (14–19), CRENESSITY (9–13) and schizophrenia (4, sustained across five quarters) carry robust shape at raw-count scale. Everything above is a hypothesis with an independent corroborating record cited alongside it; nothing rests on the counts alone. Window is 13 quarters — not "all history." The corpus is press releases, which are curated and carry no Q&A.

6. Criteria scorecard

7. Downside Criteria — the permanent-loss case with a named cause

MEASURED. Does not block. Logged to the ledger.

Named cause: Zydus wins or settles early on the INGREZZA SPRINKLE ANDA, breaking the 2038 generic wall years ahead of schedule, while the $2.9bn spent on Soleno and the ~$2.6bn/yr opex build leave no balance-sheet cushion to absorb it.

The mechanism, specifically. INGREZZA is 84.6% of TTM revenue ($2,625.4m of $3,102.4m). Its generic exposure is settled to 1 March 2038 — for the original ANDA filers only. Zydus filed a Paragraph IV ANDA on INGREZZA SPRINKLE in March 2025 and is not a party to those settlements; NBIX sued in D. Delaware in April 2025 and the case is live. An adverse ruling, or a settlement at an earlier date, does not merely dent one presentation — it establishes the entry date for the franchise. This is the single largest binary in the file and NBIX has no obligation to disclose settlement terms until they exist.

The reason it bites harder now than a year ago: the balance sheet no longer absorbs it. Net cash went from $2,647.2m to roughly $226m in a single transaction, with $600m of newly drawn secured debt carrying a maximum total net leverage covenant of 3.75:1.00 and a minimum interest coverage covenant of 2.00:1.00. NBIX granted a security interest in substantially all of its assets. A company with $2.6bn of net cash and one product can absorb a patent surprise; a company with $600m of secured debt and a $2.6bn annual opex run rate has less room.

Quantification. Re-solving the pro-forma reverse DCF at a 12.0x exit multiple (below every pharma comparator, appropriate for an asset with a visible cliff) and holding the terminal margin at 30% requires a +19.0% revenue CAGR — which INGREZZA-impaired NBIX would not deliver. The price that clears at a 10% CAGR under those terms is roughly $115–125 per share, −32% to −37%.

Probability assigned: 20%. Logged to the recommendation ledger for Brier scoring.

Second named cause, lower severity: the Soleno deal was mistimed. VYKAT XR's sequential growth was +3.2% in the last quarter filed before close, against +102% and +39% in the two prior quarters, and the EU application was withdrawn the day after announcement. If VYKAT annualises near $380m rather than compounding, NBIX has paid 7.6x forward sales for a decelerating single-product single-customer franchise and will carry the goodwill. Watch the Q2 2026 purchase price allocation for the size of the goodwill and the inventory step-up. Probability that VYKAT FY2027 revenue lands below $400m: 35%. This is a value-destruction case, not a permanent-loss case — it costs roughly 10–15% of equity value, not 35%.

Third, disclosed without a probability because it is not sizeable from the file: the DOJ civil investigative demand received in August 2025 regarding INGREZZA sales and marketing. NBIX states it is cooperating and that "no assurance can be given as to the timing or outcome." No accrual is disclosed. No figure is estimated here.

Going concern: not applicable. $748.7m of FY2025 free cash flow, 98% gross margins, $600m of drawn debt against a $1.0bn facility, and every dollar of opex discretionary at the R&D line. There is no solvency path to permanent loss.

8. Disclosed limitations

1. The pro-forma post-Soleno balance sheet is an estimate, built from five separately filed components and labelled as such at every point of use. The Q2 2026 10-Q replaces it with a filed number, within days.
2. No Soleno purchase accounting exists yet. No goodwill, no intangibles, no inventory fair-value step-up. Given what the same step-up did to GMED's reported margins in this batch, this is the single most important unfiled number in the file.
3. No Street consensus. Alpha Vantage quota exhausted 2026-07-29. Every forward figure is company guidance (labelled) or a house estimate (labelled). No consensus number is quoted or invented.
4. Mention-frequency corpus is press releases, not transcripts (§5). Same method, different instrument, one source used consistently.
5. NBIX guides INGREZZA only. There is no company guidance for CRENESSITY, VYKAT XR or total revenue. The FY2026 and FY2027 total-revenue figures used in the 12-month target are house builds and are shown line by line in NBIX_Valuation.md §3.1 so the assumption is auditable rather than asserted.
6. Q2 2026 has not been reported. The latest hard data is Q1 2026, filed 2026-05-05. Given NBIX's history of reporting Q2 in late July, this memo is being written within days of, and possibly hours of, a print that changes several of its inputs.
7. Consolidated ADV is not available from the IEX feed; only the partial-tape figure is reported, labelled.

Sources

• SEC EDGAR XBRL companyfacts, CIK 0000914475 (NBIX), 0001484565 (Soleno), 0001070494 (ACADIA)
• NBIX FY2025 Form 10-K, filed 2026-02-11 (accession 0000914475-26-000007)
• NBIX Q1 2026 Form 10-Q, filed 2026-05-05 (accession 0000914475-26-000026)
• NBIX Form 8-K 2026-04-06 (merger agreement, accession 0001193125-26-142904); 2026-05-18 (closing + credit agreement, accession 0001193125-26-228013); 2026-05-28 (annual meeting / share count, accession 0000914475-26-000029)
• NBIX Form 8-K Ex-99.1 earnings releases, 13 quarters: 2026-05-05 through 2023-05-03
• Soleno Therapeutics Q1 2026 Form 10-Q, filed 2026-05-07 (accession 0001193125-26-212205); Form 15-12G filed 2026-05-28
• ACADIA Pharmaceuticals Forms 10-K/10-Q through Q1 2026
• Alpaca Markets daily bars and options chain/snapshots