Small cap with big ambitions: Data services firm Innodata up nearly 300% in LTM

It is not just tech giants capitalizing on surging interest in AI. Innodata, a small cap that once specialized in digitizing books, has refocused its business and today helps train third-party AI models. The company has been spotlighted as one of the fastest growing for AI investments. Its partners include Google, Amazon, and Microsoft. Innodata’s stock has soared nearly 300% over the past year, and Wall Street predicts it could rise another third.

What the company is working on now, the risks it faces, and what makes it interesting for investors – in this Oninvest deep dive.

Innodata has existed for more than 30 years, most of them in content digitization – converting documents, books, and records into digital formats. For a long time, the company remained off most analysts’ radar and was seen as a slow-growth business. In 1992, Innodata went public, expanded globally, and opened major centers in the Philippines, Sri Lanka, and India to handle large-scale data and content processing. Its clients during this period were mainly media, publishing, and other industries that require management of large content libraries.

In 2019, Innodata shifted its focus to AI. That pivot drove an acceleration in growth. The company now develops generative-AI services and supports the training of third-party models.

As Barron’s notes, Innodata cleans inaccurate, incomplete, or irrelevant data, uploads it to its platform, and then has it reviewed by more than 6,000 expert consultants. This human-in-the-loop infrastructure provides specialized services – fine-tuning and preference optimization – to help large language models control tone, avoid bias, and reduce errors. Partners include Google Cloud, Amazon, Microsoft, Apple, Deloitte, Snowflake, and Boeing.

On February 15, 2024, Innodata shares plunged 30% after Wolfpack Research disclosed a short position. Wolfpack called Innodata a “a deteriorating, manual data-entry business driven by offshore labor, not innovation," despite the management’s claims about “delivering the promise of AI to many of the world’s most prestigious companies."

Job postings provide some color: Innodata actively hires “generative AI specialists.” The descriptions indicate that candidates should speak languages other than English (for example, Arabic or Polish), and will handle quality control, content editing, and translation.

Wolfpack also argued that Innodata does not operate like a technology company, and that its Goldengate platform is rudimentary software. Citing a former employee, the report said the company provides data structuring and labeling services rather than inventing new technologies.

The report had consequences. Shortly afterward, investors filed a class action lawsuit, and the U.S. Securities and Exchange Commission opened an inquiry. The complaint alleges that Innodata’s operations “were powered by thousands of low-wage offshore workers, not proprietary AI" and that the company “did not have a viable AI and was not effectively developing the technology.”

In late 2024, Innodata first addressed the matter in its financial statements, denying the allegations, and stating that the class action contains “false and misleading statements regarding the company’s AI technology and services.”

The legal overhang has not halted growth. In May, Wedbush analyst Daniel Ives listed Innodata among 30 technology companies poised to benefit from AI over the long term – alongside Nvidia, Palantir, Microsoft, and Alphabet.

In September, Barron’s published an article recommending buying Innodata shares and describing the company as a high-quality small cap. On October 10, BWS Financial maintained its "buy" rating while raising its target price from $74 per share to $110 per share. That implies roughly 35% upside from the October 23 close.

BWS argues that enterprise AI adoption is still early, and that data quality is a key bottleneck for AI agents – areas where Innodata is well positioned.

Barron’s notes that revenue has tripled over the past three years, and profitability has risen sharply. In its most recent quarterly update in July, Innodata reporteda 79% year-over-year increase in revenue to $58.4 million, and a 10 percentage point increase in the gross margin to 39%. The next quarterly report will be released in early November.

Zacks consensus estimates call for Innodata sales to rise 42.8% in 2025 and 23.6% in 2026. Profit is seen as declining 6.7% in 2025 before growing 38% in 2026.

Meanwhile, peers such as CSG Systems International, CACI International, and CGI are expected by Zacks to post 2025 sales growth of 4.5%, 8.1%, and 7.2%, respectively.

Despite strong results, the management has acknowledged challenges that could weigh on performance, Zacks noted. Chief among them is customer concentration – particularly reliance on the largest client for a substantial share of revenue. The company has not disclosed that client. Barron’s also flagged dependence on two clients.

Another factor is aggressive investment. Management is increasing operating expenses in sales, delivery, and product development to capture opportunities. While these costs are framed as long-term growth drivers, they compress current profitability, and may continue to do so – a concern for investors focused on near-term margins.

Finally, competition is intense, with pricing pressure and shifting partnerships. The management has also underscored the realities of a fast-moving AI ecosystem, where enterprise adoption cycles, and technical hurdles in model safety and performance can delay revenue. Even with strong long-term potential, these uncertainties translate into execution risk.

Innodata’s stock has already rallied strongly as demand for generative-AI services grows, and the company expands into new markets. The shares are up 87% since the start of 2025, and 285% in the last 12 months. Innodata's market capitalization is now $2.35 billion. According to MarketWatch, five Wall Street analysts cover the stock, and all assign "buy" or "overweight" ratings.

However, Zacks notes the stock trades at a substantial premium to peers, limiting near-term upside for new investors. Innodata’s forward price-to-earnings ratio is 77.7, versus 16.48 for the Zacks computer-services peer group, and 28.43 for the broader computer and technology sector.

At the same time, the risks around customer concentration, rising operating costs, and high competition remain.

Zacks suggests current shareholders may continue to hold to capture long-term AI-driven growth, while new buying should be approached cautiously until the valuation cools, and earnings visibility improves. Barron’s believes that for investors convinced the AI cycle is still in its early stages, Innodata is a quality small cap to buy and hold for the long term.

The AI translation of this story was reviewed by a human editor.