AI data growth proves the second customer is real
- Innodata is a picks-and-shovels supplier for generative AI, making complex data used to train and improve large language models.
- Q2 2026 revenue hit $92.1 million, up 58 percent year over year, with an adjusted gross margin of 49 percent.
- The new second major customer ramped up successfully, accounting for 34 percent of total Q2 revenue.
- Customer concentration risk is dropping rapidly, with the top buyer falling to 37 percent of revenue.
- The stock asks investors to pay for fast growth, so any slowdown in AI spending could hurt the case.
Diversification is working
The bull case is playing out exactly as planned. Innodata posted $92.1 million in Q2 2026 revenue, growing 58 percent year over year. Adjusted gross margin hit 49 percent.
The biggest news is that customer diversification is working. The second major customer grew to 34 percent of quarterly revenue. That means Innodata now has two true anchors in Big Tech, answering the market's loudest worry.
The bear case is weaker now. The largest customer dropped to 37 percent of revenue from 58 percent a year ago. The main remaining question is how the older software platforms perform, since the company now reports everything as one segment.
Growth and margins look excellent, but the market expects perfection. To keep the momentum, Q3 and Q4 need to show steady gross margins near the high 40s and continued sequential growth from the second largest customer.
Selling data to the AI buildout
Innodata makes money by creating high-quality data sets and services that help companies train, test, and tune AI models. A large language model learns patterns from data so it can write, reason, code, or answer questions. Better data can make the model safer and more useful.
The core customer base is large technology companies building foundation models. These customers need custom data for supervised fine-tuning, reasoning, pretraining, and special use cases. Innodata pitches that if AI labs race to build better models, they need suppliers that deliver accurate data at scale.
The second layer is enterprise AI work. This includes fine-tuning models and building RAG applications, which stands for retrieval augmented generation. In plain English, RAG lets an AI system look up trusted company information before answering.
The top layer is Innodata’s own software platforms. Agility serves public relations teams, and Synodex extracts and processes medical records. The problem for investors is that Innodata now reports as one segment, so the public filings no longer show how these platforms perform on their own.
What Innodata sells
AI Data Services
This is the core business. Innodata builds custom data sets for generative AI models, mainly for large technology companies.
Advanced LLM Training Data
The company engineers specialized data for long-context reasoning and other hard model-training tasks. This work is tied to the race to make AI models reason better.
Agentic AI Evaluation Data
Innodata is building data and testing tools for autonomous AI agents. These tools help check whether agents can handle real-world tasks and resist bad prompts or edge cases.
Physical AI Data
This work supports robotics and machines that act in the physical world. It includes data about first-person views and what actions objects allow.
Agility Platform
Agility is software for public relations teams. Its PR CoPilot feature adds generative AI to media monitoring and PR workflows.
Synodex Platform
Synodex extracts and structures medical record data. It has been used in life insurance underwriting and is expanding toward clinical use cases for hospitals and doctors.
One segment, two major customers
As of Q1 2026, Innodata reports one business segment. In Q2 2026, revenue was split between the top customer at 37 percent, a second major customer at 34 percent, and all others making up 29 percent.
What could break the story
Two-buyer concentration
High impact · Medium oddsThe top customer was 37 percent of Q2 2026 revenue, and the second was 34 percent. If either major AI lab cuts spend, delays projects, or changes vendors, Innodata could lose a large slice of revenue fast.
Margins fall back
Medium impact · Medium oddsQ2 adjusted gross margin was 49 percent, well above the company’s historic norms. That level may not hold if the mix shifts toward lower-margin services or if hiring and delivery costs rise.
Less reporting detail
Medium impact · High oddsIn Q1 2026, Innodata moved to one reportable segment. That matches how management says it runs the business, but it lowers outside visibility. Investors can no longer see standalone growth or margins for Agility and Synodex.
Legal and regulatory overhang
Medium impact · Medium oddsThe company remains subject to a putative securities class action filed in February 2024. It has also previously disclosed SEC and DOJ investigations. An adverse result could cost cash, distract management, or hurt investor trust.
Future dilution
Medium impact · Low oddsInnodata maintains a universal shelf registration. That gives it flexibility to raise equity or debt. If the company uses equity while the share price is weak, existing shareholders could be diluted.
In one breath
What does Innodata actually do?
Innodata creates and manages complex data used to train and improve AI models. It also runs Agility for PR teams and Synodex for medical record processing.
Why did the INOD thesis improve in 2026?
The company added a second Big Tech customer that successfully ramped to 34 percent of Q2 2026 revenue. That significantly reduces the risk that Innodata depends too much on one single buyer.
What is the biggest risk for Innodata stock?
Customer concentration remains the main risk. Even with diversification, the top two customers still make up 71 percent of Q2 2026 revenue, meaning two companies hold massive sway over results.
Does Innodata still report Agility and Synodex separately?
No. Effective Q1 2026, the company reports as one segment. The products still exist, but investors no longer get a separate public financial split for them.
Sources and research notes
This page combines Finn's company research with public filings and other cited materials. The thesis is reviewed when material company information changes; Finn Scores use the latest available scoring data.
- Thesis reviewed
- August 23, 2026
- Score data
- September 27, 2026
- Reviewed by
- Shivam Bharuka
Comparable Information Technology Services companies
Companies near Innodata Inc. in Finn's Information Technology Services industry ranking.
