Meta’s Muse Could be Big Win for Innodata – HUNTERBROOK

Ask Meta's new AI agent, Muse, whether Innodata helped train it, and the app won't say.

Instead it points to Meta's privacy materials, which note that user conversations may be reviewed and used to improve AI unless people opt out. That non-answer sits at the center of a report on Innodata, a Ridgefield Park, New Jersey document-processing company turned AI-data supplier, whose largest customer, likely Meta, accounted for 58% of 2025 revenue before pulling back sharply in the most recent quarter.

Ticker: INOD (Innodata Inc.)
Research Firm: Hunterbrook Media
Report URL: https://hntrbrk.com/breaking-news/innodata-muse?ref=shortreport.fyi
Position Disclosure: Not disclosed in the source materials.

Thesis

Hunterbrook Media examined Innodata's dominant customer relationship and its pivot into agentic AI training, concluding that investor fears of Meta abandoning the company for Scale AI may be overstated, while several key facts remain unconfirmed.

• Single-Customer Dependence: Innodata's Form 10-K for 2025 discloses that one unnamed customer generated approximately $146 million, or 58% of total revenue, for the year.
• Meta Identification: The report ties that customer to Meta based on Innodata's customer concentration, its described AI-training work, and former-employee accounts, though Innodata never names Meta in its filings.
• Scale AI Overhang: Meta's $14.3 billion investment for a 49% stake in Scale AI, paired with a Q2 2026 drop in Innodata's top-customer revenue and a share slide from a $121.50 June high to roughly $57, fed fears Meta would redirect work elsewhere.
• Vendor Work Reportedly Growing: A former Innodata employee told Hunterbrook that Meta had not cut outside help after its Scale investment, saying "if anything, I would say it's more"; Innodata's CEO separately said the company kept growing with its largest customer by expanding into "more organizations and more AI workflows."
• Muse-Shaped Assignment: Innodata's newly disclosed largest-customer work involves personalizing agents to handle "long, complicated tasks," a description that closely tracks Meta's public description of Muse, but neither company confirmed Innodata's involvement when asked.
• Broadening Customer Base: A second unnamed "Big Tech customer" more than doubled its spending to roughly $31 million in Q2, and Innodata separately disclosed winning "one of the fastest-scaling frontier labs" as a client, though the durability of that new spending is unclear.
• Undisclosed Deal Terms: Innodata has not revealed the size, duration, or profitability of its new agentic-AI assignment, and its CEO acknowledged that quarter-over-quarter revenue deceleration is "certainly possible" in the business.
• Prior Skepticism and Closed Inquiries: Short sellers Wolfpack Research and J Capital have separately challenged Innodata's AI capabilities and customer-agreement claims; DOJ and SEC inquiries disclosed in 2025 were closed without enforcement action that June, and executives sold shares during 2026 amid Glassdoor reports of layoffs.

Catalysts

• CEO transition on September 30, 2026: Rahul Singhal formally takes over as Innodata CEO amid unresolved concentration and growth questions.
• Next quarterly filing details: Upcoming disclosures could reveal for the first time the size, duration, and profitability of the new largest-customer agentic-AI assignment.
• Top-customer revenue trend: Future earnings may show whether the roughly one-third Q2 decline in top-customer revenue continues or reverses.
• Muse-role confirmation risk: Any confirmation or denial from Meta or Innodata regarding Innodata's role in training Muse would directly address a central open question.
• Second Big Tech customer durability: Subsequent-quarter disclosures will indicate whether the roughly $31 million run rate holds up.
• Frontier-lab contribution: Coming-quarter revenue disclosures may clarify the financial impact of Innodata's newly announced "fastest-scaling frontier lab" customer.

Company Response

Hunterbrook sent questions to both Innodata and Meta about Innodata's possible role in training Muse; neither company responded. When a Hunterbrook reporter asked Muse itself whether Innodata helped train it, the agent did not confirm or deny any involvement and instead directed the reporter to Meta's privacy documentation, which states that user conversations may be reviewed and used to improve Meta's AI unless users opt out.

Notable Details

• Meta's Muse reached No. 1 on Apple's U.S. free iPhone app chart just 10 days after launch, carrying a 4.9 rating, and Meta added more than $100 billion in market value in a single day, per Business Insider and the report.
• Innodata's stock climbed from roughly $4 a share in early 2023 to a June closing high of $121.50, the run-up that set the stage for its later pullback.
• AI-training startup Mercor served as the backup bidder, at $7.5 million, for Spirit Airlines' bankruptcy-estate data, including old emails, Teams messages, and spreadsheets.
• Mercor's valuation rose from $250 million in its 2024 Series A to $10 billion in its 2025 Series C, a 40-fold jump, with a reported $20 billion round under discussion, according to TechCrunch.

"bruh this is all me out here rawdogging these tweets. i don't let the corposlops near my tweet cannon."

Meta AI chief Alexandr Wang wrote this while claiming he personally ran the social-media push behind Muse's breakout launch.

FAQs

What is Innodata (INOD) and why is its stock so sensitive to one customer?

Innodata is a Nasdaq-listed, New Jersey-based company that supplies training data, human feedback, and AI testing services, work it has performed since shifting from its original document-digitization business. Per its Form 10-K, one unnamed customer generated approximately $146 million, or 58% of 2025 revenue, making Innodata's results highly dependent on that single relationship.

Who is Innodata Inc.'s largest customer, and has the company confirmed it?

Innodata Inc. has not named the customer in its filings. Hunterbrook Media's report concludes, based on customer concentration, the nature of the disclosed work, and former-employee accounts, that the customer is likely Meta, though this identification remains unconfirmed by either company.

What did Hunterbrook Media find about Innodata's ties to Meta's Muse?

Hunterbrook Media found that Innodata's newly disclosed largest-customer assignment, personalizing agents to handle "long, complicated tasks," closely resembles the capabilities Meta describes for Muse, which remembers preferences, works across apps, and continues tasks after a user closes the app. The firm could not confirm Innodata's direct role, since neither Meta, Innodata, nor Muse itself would answer the question.

Why did Innodata's stock fall in 2026?

Innodata shares dropped from a June closing high of $121.50 to roughly $57 amid investor concern that Meta's $14.3 billion investment for a 49% stake in Scale AI could shift AI-training work away from Innodata. That concern was reinforced when Innodata's Q2 2026 results showed its largest customer's revenue falling from an estimated $50.5 million to $34.1 million.

Is Meta actually cutting back on outside AI-training vendors like Innodata?

A former Innodata employee told Hunterbrook that Meta's reliance on outside vendors had not declined after its Scale AI investment, saying "if anything, I would say it's more." Innodata's CEO separately said on an earnings call that the company continued growing with its largest customer by expanding into additional organizations and AI workflows.

Does Innodata have other major customers besides its presumed Meta account?

Yes. Innodata disclosed that a second unnamed "Big Tech customer" more than doubled its spending to roughly $31 million in Q2 2026, nearly matching the top account, and separately said it had won "one of the fastest-scaling frontier labs" as a client. The report notes it is unclear whether the second customer's spending level will hold in future quarters.

What regulatory or governance issues has Innodata faced?

Innodata disclosed Department of Justice and Securities and Exchange Commission inquiries in its 2025 filings; both were closed without enforcement action as of June 2025. Separately, short sellers Wolfpack Research and J Capital have published reports questioning Innodata's AI capabilities and the scale of its customer agreements, and Glassdoor reviews cited in the report describe layoffs throughout 2026.

Who is taking over as Innodata's CEO, and when?

Rahul Singhal, currently Innodata's Chief Revenue Officer, is scheduled to become CEO effective September 30, 2026, succeeding Jack Abuhoff. The transition comes as the company faces open questions about customer concentration and the durability of its newer agentic-AI revenue.

Disclaimer: This summary is not primary research and does not constitute investment advice. It is a brief overview of a detailed equity research report authored by the firm, organization, or source referenced in this article or at https://hntrbrk.com/breaking-news/innodata-muse, which contains extensive evidence, regulatory filings, and analysis; readers are encouraged to review the full report there for a comprehensive understanding. The content provided in this publication is not authored or originated by us — we act solely as a distributor and do not endorse, verify, or take responsibility for the accuracy, completeness, or reliability of the information presented. This publication is for informational purposes only and should not be construed as legal, business, investment, or tax advice. Always conduct independent due diligence and consult qualified professionals before making any decisions based on the information contained herein. We disclaim all liability for any loss or damage arising from reliance on third-party content, and the views expressed are solely those of the respective source and do not necessarily reflect our own.