Founded in the late 1980s, the IT services provider Innodata remained under the radar of investors and major companies for a long time. However, that all changed after it pivoted to large language models (LLMs) and generative AI five years ago. Since then, it has scooped up almost all of the Magnificent Seven companies as clients, while its share price has soared over 3,000% in five years.
At the forefront of digitalization
Innodata was founded in 1988 in New York by Ted Solomon. At the core of its business 30 years ago lay a simple but highly relevant idea at the time — data digitization. For example, it scanned printed documents and digitized images and medical records. The data was stored on floppy disks and CDs and then sent to clients, which included publishers who could then replicate the data in books and magazines. In 1993, the company went public on the Nasdaq. By the end of 1996, it had offices in the U.S., the Philippines, and Sri Lanka.
Gradually, the range of services offered by Innodata expanded. In 2001, it acquired Isogen International, a company specializing in sophisticated consulting and training in the knowledge-processing technologies of XML, SGML, and other standards.
Since 2007, with the rise of e-books, Innodata has been developing technology for converting print books into digital formats. In its 2007 financials, the company reported that it had digitized and converted thousands of books into an e-book format for an unnamed global electronics manufacturer. By the end of 2012, it was servicing four out of the five leading digital e-book retailers (without disclosing their names).
Five-year growth spurt
For 25 years after its IPO, Innodata was considered a slow-growth IT services and enterprise software provider, the Motley Fool noted. This changed in 2019, when the company decided to focus on AI, in particular, helping other companies train their AI models.
Innodata, in its annual filing with the SEC for 2019, pointed out that an increasing number of companies were developing AI-based applications — for self-driving cars, surveillance systems, medical diagnostics, digital assistants, and more. However, all these applications depended upon algorithms that needed to be constantly trained on large amounts of high-quality data. Preparing such data is a labor-intensive task, with most companies spending 80% of the project time on it, leaving just 20% for training.
Even major market players admitted this issue. Data-related challenges were a top reason why IBM clients halted or canceled AI projects, Arvind Krishna, then IBM’s senior vice president, told the Wall Street Journal back in May 2019.
As a result, companies increasingly looked for partners that could prepare such data, Innodata noted. Against this backdrop, it launched in 2018 a suite of task-specific microservices for training AI, which could be customized to meet a client’s needs. In the case of an error or discrepancy, the AI sends the data to a human for verification, who then checks and sends it back to the network. This way, the AI continually learns on verified data.
For Innodata, this new niche promised significant growth. Its historical core market of data digitization services is small — just $250 million — with a grim growth outlook. On the other hand, the AI training market is expected to reach $2.57 billion in 2024 and grow at a CAGR of 18% until 2034, as Innodata stated in its 2023 financials. A particularly promising area is data annotation, which involves tagging vast amounts of data based on specific features so that AI can recognize it. By 2032, this market is expected to reach $25 billion, nearly 14 times its size in 2022. This area is handled by Innodata’s research center, Innodata Labs.
Innodata’s pivot worked. From 2019 to 2023, Innodata’s revenue grew at a CAGR of 12%, according to the Motley Fool. This year, the top line added 41% year over year in the first quarter to $26.5 million, 66% in the second quarter to $32.6 million, and almost 136% in the third quarter to $52.2 million. Following the release of its third-quarter financials on Thursday, November 7, Innodata stock skyrocketed more than 75% in a single day. In the third quarter, net profit was reported at $17.4 million, up from around $1 million in the first quarter and a net loss of $14,000 in the second (Innodata attributed this to recruiting costs, among other things). Since May, the company has upgraded its 2024 full-year revenue guidance twice. Initially, it expected growth of 40% versus the $86.8 million top line in 2023, which was subsequently raised to 60% and now around 90%.
The company attributes its growth to an increasing number of clients. In the third quarter, Innodata won a “prominent social media company” and revealed, without disclosing its name, that this company had become its eighth Big Tech client. Analysts expect Innodata to continue growing at a revenue CAGR of 33% from 2023 to 2026, the Motley Fool noted.
The company’s expansion into AI training and big data services has caught Wall Street’s attention, Benzinga has pointed out. Innodata stock has soared more than 3,400% over the last five years. The Motley Fool included it in two recommendations lists: AI stocks “ready for a bull run” and stocks to invest in right now. Benzinga spotlighted it as one of three stocks that have at least doubled in value in 2024, while Yiannis Zourmpanos, an InvestorPlace contributor and the founder of the stock-market research platform Yiazou Capital Research, included it in his list of three alternatives to investing in Nvidia.
The company has agreements with Big Tech companies, including five of the Magnificent Seven. While it does not disclose their names in its financials, its website mentions partnerships with Google, Amazon, and Microsoft.
Short-seller attack
In February 2024, Innodata faced unexpected and unwelcome attention from investors: Wolfpack Research, whose founder Dan David is an activist short-seller, announced it was shorting Innodata, calling the company a “deteriorating, manual data-entry business driven by offshore labor, not innovation.” In its report, Wolfpack argued that AI model data entry was being performed not by algorithms but by people working in low-wage countries. This caused Innodata stock to lose over 30% in a single day, on February 15. However, the company’s stock has bounced back, more than doubling from that date to early November.
A week after the Wolfpack Research report was released, a lawsuit was filed in the U.S. District Court for the District of New Jersey by Innodata investors, led by David D’Agostino, against the company and several of its top executives. It alleges that investors were misled about the outlook for the company, with all the presented arguments based on Wolfpack’s research. Other investors later joined the lawsuit, which is still pending.
D’Agostino is seeking unspecified compensatory and punitive damages, while the company intends to defend itself vigorously in court, as reported in Innodata’s 2023 earnings.
The legal proceedings have not resulted in a downgrade or withdrawal of ratings for Innodata. According to MarketWatch, all three analysts covering the company have a “buy” recommendation amid an average target price of $34 per share. On Tuesday, November 12, Innodata stock closed at $45.99 per share.
