$INOD: 550 Jobs Cut Mid-Shift, a 25% Crash, and Earnings in Eight Days. The Filings Tell a Different Story
Innodata cut ~550 workers mid-shift with no press release and the stock lost a quarter of its value on fears it lost its biggest AI customer. The concentration math, the real cash net of customer prepayments, the $300M ATM, and scenario zones into the November 5 print.
On Monday, October 26, employees at Innodata's India operation logged in, recorded their attendance, and started an ordinary workweek. By lunchtime, rumors were circulating that the company's Philippines team, including raters, quality experts, team leads, and operations managers, had been terminated on a Zoom call. Then, one by one, Indian employees watched their laptops log out and their company IDs deactivate, mid-shift, before any termination email arrived. By evening, nearly 550 people in India had lost their jobs, according to India Today, which reviewed termination letters and spoke to affected workers. The notices cited "a reduction in the business assignment" the employees had been engaged for.
There was no press release. There was no 8-K. As of this writing, Innodata (Nasdaq: INOD) has not publicly confirmed the layoffs, the number affected, or which customer program was involved. The market filled the silence itself: the stock fell 25.3% on Tuesday, closing at $44.74 against a prior close of $59.91, on roughly 9.7 million shares of volume. A shareholder-suit shop had an "investigation notice" out the same day, which tells you more about how these things work than about Innodata.
If you read our September 23 briefing on Innodata, you know why the market reacted this violently to a workforce action the company itself hasn't even acknowledged. Innodata is a data-engineering and AI-services firm, the human layer behind frontier-model training: data labeling, model evaluation, reinforcement-learning environments, safety testing. The bull case and the bear case are the same fact. One unnamed customer, widely argued to be Meta after investigative outlet Hunterbrook Media connected the dots in September, generated 58% of Innodata's $251.7 million in 2025 revenue. When you are a people-intensive services vendor and your largest customer's program winds down, the first visible symptom is exactly what workers in Delhi and Manila described: raters and quality experts locked out of their machines on a Monday.
So the fear is coherent. Meta paid $14.3 billion for 49% of Scale AI last year and has been expected to in-source more of this work ever since. A sudden, unannounced cut of several hundred rating and QA staff looks, from the outside, like a major program ending. The stock had already round-tripped this exact anxiety once in 2026: $34 in March, above $120 in June, back below $50 in the summer on concentration fears, $70 in late September when the Hunterbrook report flipped the narrative, and now $44.74.
Here is what makes the next eight days unusually clean as a setup: Innodata reports third-quarter results after the close on Thursday, November 5, with a call at 5:00 p.m. Eastern. Management will have to address the layoffs, the state of the largest customer relationship, and full-year guidance, on the record, under Reg FD, with analysts asking the questions the termination emails didn't answer. This is a dated, binary information event, and the stock just repriced 25% on a story with no company-confirmed facts in it.
Which raises the question the market is currently answering with panic: do the actual filings, the most recent ones, from August, support the "they just lost Meta" read? Some of what's in the 10-Q cuts against the crash narrative. Some of it cuts the other way, including a detail about the headline cash number that most of the coverage this week has missed entirely.
A free membership gets you access to every free article at AlphaBriefing.com: Join AlphaBriefing.
The rest of this briefing is for paid members: the customer-concentration math quarter by quarter, the real cash position once you net out customer prepayments, the $300 million ATM nobody is talking about, what 550 heads actually means for a 10,000-person services model, and scenario-by-scenario price zones into the November 5 print.
AlphaBriefing Paid gets you every investment thesis, scenario framework, and catalyst brief we publish: the analysis private intel clients pay four figures for, at a fraction of that.