IVA — Inventiva S.A.

Company context · prepared 2026-08-12 · company sweep 2026-08-12 · USD · framework v5.6.1

How to read this. Written for a reader who has not studied pharmacology or finance. General terms are defined once in framework/04-glossary.md. Tags: [VERIFIED — source] / [UNVERIFIED] / [WEB ESTIMATE — source, date].

Program analyses for this company are the sub-folders listed in C.2.

C.0 Company-tier coverage — CLEARED

C.1 What the company is

Inventiva S.A. is a clinical-stage biopharmaceutical company headquartered in Daix, France. It develops oral small-molecule tablets, and its entire clinical effort is directed at one liver disease. It has no product revenue [VERIFIED — BPIQ fetch_company_info 2026-08-12].

It is effectively a single-asset company. The drug feed returns four program rows. Two are the same molecule, lanifibranor, and the other two are marked failed and discontinued [VERIFIED — BPIQ fetch_company_drugs 2026-08-12]. Exactly one row carries a disclosed catalyst: the Phase 3 readout of lanifibranor in metabolic dysfunction-associated steatohepatitis (MASH), a liver disease driven by fat, inflammation and scarring. In February 2025 the company cut roughly half its staff to concentrate on that single program [VERIFIED — company press feed 2026-02-11 items].

The company is dual-listed. Ordinary shares trade on Euronext Paris and American Depositary Shares (ADSs) trade on Nasdaq under IVA. Every figure in this document is the Nasdaq ADS line unless stated otherwise. The ADS line is now genuinely liquid: daily volume over the last month ran 0.4M to 1.4M ADSs, roughly $2–6M per day [VERIFIED — data/prices/IVA.json, bars 2026-07-10 to 2026-08-11]. The 2026-08-04 company document called the line thin on a reported day volume of 19,706 ADSs; the price cache shows 1.02M ADSs traded that day, so the earlier reading reproduced a BPIQ quote artifact rather than the market — see C.8.

The twelve months to this sweep contain a near-complete refit of management and balance sheet: a new chief executive in October 2025, three C-suite appointments in April 2026, and a comprehensive debt-and-equity refinancing completed across June and July 2026 [VERIFIED — company press feed; see C.3].

C.2 Pipeline — every program

The two live rows are the same molecule under two names, which is the standing connector bug that makes the integer bpiq_drug_id the only safe key. A preclinical TGF-β program in idiopathic pulmonary fibrosis appears in the company’s own business description but not as a drug-feed row at all [VERIFIED — BPIQ fetch_company_info full_description].

C.3 Financial position

The euro-to-dollar conversion used throughout is 1.1639 USD per EUR, from the company’s own June 2026 offering release (“approximately $110.8 million (€95.2 million)”). A June 2026 settlement rate, not a spot rate; every converted figure inherits that imprecision.

Dilution history. Inventiva has raised in every one of the last four half-years. Read together with C.4.

Is a raise likely before the next catalyst? A further equity raise before the Q4 2026 readout remains unlikely, and that is the point of the June refinancing: the company front-ran its own catalyst with a $120M offering and a €130M debt facility so it would not have to raise into the event, and its base runway now reaches the end of Q2 2027 — past the readout with margin. Further debt drawdowns (Tranche C, €55M) and warrant exercises are expected and are what the extended runway depends on. The dilution that matters has already happened, at $4.40 in June 2026, 7% below the current price. The residual overhang is 15.7M EIB warrants, about 6.6% of the 236.3M shares outstanding. [UNVERIFIED — judgement, built on the verified transactions above]

C.4 Valuation frame

What has changed since 2026-08-04: a run-up has started. The stock closed $3.76 on 2026-06-15, $4.51 on 2026-08-04 and $4.72 on 2026-08-11 — up 25% off the June trough and 5% over the last week, on daily volume of $2–6M [VERIFIED — data/prices/IVA.json]. It still sits at only 29.6% of its 52-week range and 41% below the 52-week high, seven weeks to five months before the readout that carries the whole company, and 7% above the June offering price of $4.40. The shares are drifting toward the event rather than priced for a positive result, and the 2026-08-04 finding of “no run-up at all” is no longer current.

C.5 Ownership and flow

Insider flow. The insider-transaction connector returned no rows on either attempt, and the EDGAR submissions feed shows no Form 3/4/5 — as a foreign private issuer Inventiva files none. No conclusion about insider buying or selling can be drawn either way. A stated gap, not a finding. [VERIFIED — BPIQ empty ×2; EDGAR submissions 2026-08-12]

Concentration, from the 13F feed — with the filing-quarter labels shifted back one quarter. The v5.6.1 connector bug is confirmed on this ticker: every line in the block labelled “2026Q2” implies a $5.55 share price, which is the cache’s 2026-03-31 close exactly, and every “2026Q1” line implies $4.65, the 2025-12-31 close. Each block therefore holds the previous quarter’s holdings [VERIFIED — value ÷ shares matched against data/prices/IVA.json quarter-end closes]. Correctly dated: eight funds held about 17.47M shares (~7.4% of shares outstanding) at 2026-03-31 — Samsara BioCapital 5.19M, Paradigm Biocapital 4.00M, ADAR1 2.17M, NEA 1.46M, Ikarian 1.43M, RTW 1.30M, Deep Track 1.22M, BVF 0.69M. Between 2025-12-31 and 2026-03-31 Ikarian added 0.50M → 1.43M and Deep Track 0.49M → 1.22M while ADAR1 trimmed 2.71M → 2.17M. The holder count jumped from two to eight at 2025-12-31, tracking the November 2025 ADS offering. There is no 13F visibility past 2026-03-31 yet; the filings holding 2026-06-30 positions are due around mid-August 2026. The 2026-08-04 company document dated all of this one quarter too late — a correction, recorded in C.8.

European holders, from the press feed. Sofinnova funds reported an 8.6% voting stake (February 2026), Great Point Partners 3.49%, Deep Track 3.76% including warrants (May 2026). In July 2026 an NEA-led investor group crossed below the 5% threshold — still the one clearly directional flow on record, and it is a reduction [VERIFIED — press feed items 2026-02-15, 2026-02-17, 2026-05-15, 2026-07-24; underlying AMF and 13G filings not read directly].

C.6 Options chain and its readability

Plain takeaway: the chain still cannot price this event and must not be used to. Strikes sit $2.50 apart on a $4.72 share, bid-ask spreads exceed the contract mids, and the deep put quotes print the documented BPIQ floor artifact. Nothing here has improved since 2026-08-04.

Several put contracts (the $5.00 across expiries) return implied volatility of exactly 0.01488, the documented BPIQ floor artifact, and are treated as null throughout.

C.7 Reaction to past catalysts

The historical-catalyst feed is byte-identical to the 2026-08-04 sweep: six rows, 2022-10-31 to 2025-04-01, none of them a Phase 3 efficacy readout [VERIFIED — BPIQ fetch_company_historical_catalysts 2026-08-12].

Not one of these is a Phase 3 efficacy readout. The largest recorded move is 8.1%. There is no precedent in this company’s history for how it trades on a registrational result, so the program document leans on external peer precedents. The feed still begins 2022-10-31, so the June 2020 NATIVE Phase 2b topline — the one comparable data event — remains unrecoverable from it. The price cache now reaches back to 2021-08-06, which also predates nothing relevant here.

C.8 Company data-quality flags

• The connector reports euro figures in dollar-labelled fields. cash 230,870,000 matches the company’s 2025-12-31 total of €99.3M cash + €131.6M deposits (€230.9M) to three significant figures; ttm_burn −91,311,000 is consistent with a full-year 2025 euro operating outflow (the sign flipped to negative since the 2026-08-04 sweep; the magnitude is unchanged). Every cash and burn figure in C.3 is restated from company releases with explicit currency and date. Still the largest data trap on this ticker (rules 6 and 11).
• The BPIQ quote block lags one session. On 2026-08-12 it reports last 4.72 / high 4.84 / low 4.72 / volume 382,924 — the price cache’s 2026-08-11 bar. The same lag explains the 2026-08-04 sweep reading $4.26 (the 2026-08-03 close) as that day’s spot. Spot in this document is dated to the session it belongs to.
• The BPIQ volume field is unreliable on this ticker. The 2026-08-04 sweep reported 19,706 ADSs traded; the cache shows 1.02M that day. The “thin ADS line” conclusion in the previous company document was an artifact. Liquidity is read from the cache.
• finance_updated_at is null, so nothing on the company record can be dated from the connector itself. The cash figure is demonstrably 2025-12-31.
• Enterprise value ($954.5M) is stale on both legs — 2025-12-31 cash and a debt picture that predates the June 2026 refinancing. Recomputed in C.4 as about $930.3M.
• max_52_week_position reproduces the previous-close bug exactly: reported 30.22% recovers from the $4.75 previous close; the correct figure on the last price is 29.6%. C.4 derives the range and position from the price cache per the C.4 instruction, not from this field.
• The 13F filing-quarter labelling bug (v5.6.1) is confirmed on IVA: the “2026Q2” block’s value/shares ratios equal the 2026-03-31 close ($5.55) and “2026Q1” equals the 2025-12-31 close ($4.65). C.5 shifts every label back one quarter. The 2026-08-04 company document took the labels at face value and dated the fund flows one quarter too late; C.5 above is the corrected reading.
• The price cache’s 2026-08-11 bar carries a null close (o/h/l/v present, c/a null) — a Yahoo artifact on the most recent bar. The $4.72 close for that session is taken from BPIQ’s (lagged) quote, which carries the matching o/h/l. Cache-derived figures use the cache’s own last non-null close ($4.75, 2026-08-10) where the library requires one.
• An unexplained block of option open interest persists: 20,350 contracts at the $2.50 Jan-2027 call strike and 20,224 at the $10.00 strike (was 20,348 / 20,223 on 2026-08-04), against 102 at $5.00, on near-zero volume. Two matched ~20,000-lot blocks at the extreme strikes look like one institutional call spread or a data artifact. Flagged, not interpreted.
• short_float units remain unconfirmed per the standing connector documentation. C.5 uses the FINRA count over the filed share count instead, whose units are known.
• The runway contradiction flagged on 2026-08-04 is now largely resolved. The 6-K/A of 2026-07-30 states scenario runways (base: end Q2 2027; extended: start Q1 2028) rather than a bare going-concern sentence. The naive 28-month arithmetic still overshoots because it ignores debt service and pre-launch spend; the company’s scenario framing is used. The remaining unverified piece is the exact composition of the gap.
• The share count is resolved: 236,280,202 as of 2026-07-31 from the company’s monthly statement (via a press-feed re-report — the statement itself was not read). The 2026-08-04 document’s 235.67M was implied from market cap ÷ price; the filed EDGAR XBRL count (193.2M) is annual-only and predates the June 2026 offering.
• A possible duplicate financing entry in the press feed (2026-05-07 re-report of the 2025-05-05 €116M tranche) — carried forward, still flagged, still unresolved.
• Drug names in the pipeline feed are dirty (“Lanifibranor (Pan-PPAR)” vs “Lanifibranor + empagliflozin”). Programs are keyed on bpiq_drug_id throughout.
• A preclinical TGF-β program in idiopathic pulmonary fibrosis appears in the company’s business description but not as a drug-feed row.