Can we count your visit with Google Analytics? No ads, no profiles. Details
INOD
USInformation Technology · IT Services
Each weekday: the watchlist of pullbacks inside uptrends, every new ticker teardown, and the bubble shifts that move names like this one. Free. Education, not advice.
Free when you join: the $HOOD Teardown Dossier (Robinhood Markets), the full desk file we sell in the shop. One per new reader.
No spam. One email per day max, unsubscribe in one click. Pro ($49/mo) includes every dossier and report in the shop.
Structural: pure-play vendor on the "data labeling + eval is the picks-and-shovels of the LLM stack" thesis. ~80% of revenue now flows through DDS, anchored by five hyperscaler master agreements signed 2024-2025. Small-cap proxy when private comp $SCALE (acquired by $META at ~$14B) is off-limits; thinner moat than Scale but faster operating leverage on hyperscaler ramp.
(1) Customer #1 (widely identified as $MSFT-aligned hyperscaler) revenue scaled 5x+ YoY through 2025; customers #2-5 still in early-ramp phase. (2) Margin profile expanding as repeat work shifts from greenfield annotation to higher-margin eval / red-team contracts.
(3) Operates at fraction of Scale AI's headcount cost via offshore delivery (India / Philippines / Sri Lanka). (4) Generative-AI capex cycle from $NVDA / $AMD ecosystem flows downstream into data prep as a structural %. (5) Pristine balance sheet, no debt, ~$100M cash post equity raise.
(1) Customer concentration extreme - top customer was >60% of revenue in recent quarters; loss or scope cut would crater the multiple. (2) Hyperscalers ($GOOGL, $META, $MSFT) all building in-house data ops + acquiring labelers ($META/$SCALE) - Innodata risks being a transitional vendor.
(3) Multiple has expanded with the narrative; 2026 revenue must compound >50% to stay rich. (4) Annotation is commoditizing; pricing power sits with frontier-model labs, not vendors. (5) Eval / red-team revenue still small; not yet proven to offset annotation deflation.
It reads our $INOD data (thesis, key levels, peers, a year of prices) and any chart screenshot you paste, then explains the setup in plain English. Free with an email login, 5 questions a day. Education only: it never tells you to buy or sell, and it can be wrong, so check the numbers.
High-signal only
Auto-computed retracement of the trailing 52w low-high leg. Recomputed on each refresh; not a curated level.