Innodata Inc. (INOD) Earnings Call Transcript & Summary
November 6, 2025
Earnings Call Speaker Segments
Amy Agress
executiveThanks, Michael. Hi, everyone, and good afternoon. We’re glad to have you with us today. On the line for this call are Jack Abuhoff, Innodata’s CEO; Rahul Singhal, who is our President and Chief Revenue Officer; and Marissa Espineli, our Interim CFO. Aneesh Pendharkar, who is Senior Vice President, Finance and Corporate Development, is here as well. Jack is going to start by sharing his view of the business, then Rahul will make some comments, and after that Marissa will take us through a review of our results for the third quarter. Once that’s done, we’ll open things up for questions from analysts. Before we begin today, please remember that we will be making forward-looking statements — that means predictions, projections, or other comments about events that have not yet happened. Those statements reflect what we currently expect, assume, and estimate, and they involve risks and uncertainties. Because of that, actual results could be materially different from what those forward-looking statements indicate. The factors that could cause such material differences are described in today’s earnings press release, in the Risk Factors section of our Form 10-K, Form 10-Q, and other reports and filings we make with the Securities and Exchange Commission. We have no obligation to update forward-looking information. In addition, we may discuss certain non-GAAP financial measures on this call. You can find more details about those non-GAAP financial measures—along with a reconciliation of them to their GAAP equivalents—in the earnings release we filed with the SEC today, and in our other SEC filings, which are available on our website. That’s all I have for now. Jack, I’ll turn it over to you.
Jack Abuhoff
executiveThanks, Amy, and good afternoon, everyone. Innodata's Q3 was another record-setting quarter. We reached an all-time revenue peak of $62.6 million — that translates to 20% organic growth versus the same period last year, and a 7% sequential increase over the prior quarter. Adjusted EBITDA was $16.2 million, or 26% of revenue, up 23% from the preceding quarter, which shows margin expansion even after factoring in the growth investments I'll get into later in this call. Cash rose to $73.9 million, up $27 million from year-end and $14.1 million quarter over quarter. On the key metrics, we exceeded analyst expectations. With this strong business momentum, we're reaffirming our prior guidance of at least 45% year-over-year growth for 2025, and we anticipate 2026 growth that could be transformative; this afternoon I'll walk through where our confidence comes from, covering the considerable growth we expect from our established strategic vectors as well as the strong early returns from newly funded investment areas. Then I'll describe how we're preparing the organization for the next level. Let me start with our established strategic vectors. Since our last report, we've continued making major progress in strengthening trusted relationships with big tech customers that carry high dollar value. Our deal momentum keeps building, with significant expansion across a varied group of foundation model builders, spanning both current and newly added customers. Out of the 8 big tech customers we've recently discussed on these calls, we presently project 6 will grow next year, several of them rather dramatically. As one example, we've just gotten verbal confirmation for further expansions with our biggest customer, plus verbal confirmations for the deal we think could potentially bring in $6.5 million of revenue with a different big tech firm. On top of that, our expectations rest on our assessment of these customers' 2026 budgets for training data and evaluation, together with the growing trust we believe we're earning from them via proofs of concept, pilots and scaled deployments; furthermore, beyond those 8 customers, we closed in Q3 — or expect to wrap up soon — deals with 5 more big techs. We think each of these 5 new big techs is set to add meaningfully to our 2026 growth. From where we sit, three of those five newcomers are set up to spend potentially hundreds of millions of dollars annually on data and evaluation for generative AI, and we think we're well placed to grab a share of those dollars. Something worth noting: two of them are among the top players worldwide in commerce, cloud and AI. Let's turn now to the six fresh initiatives we're launching in 2025 — several of which I'm sharing publicly right here for the first time — and all of them are already delivering meaningful results and, in our estimation, will contribute substantially to growth in 2026. The first thing we're doing is producing pretraining data at scale, because that's what gives a model its language skills and knowledge. Until recently, our company was mainly centered on post-training data — the kind used to teach models how to reason, follow instructions, and carry out tasks. But early this year, we noticed researchers finding stronger and stronger links between LLM benchmark scores and how good their pretraining data is. When pretraining corpora were of higher quality, the resulting models reliably did better at grasping nuance, context, and intent across languages and domains. Once we came across that research, we reached the conclusion that customers would be looking more and more for better-quality pretraining data sources. So we put roughly $1.3 million into developing new capabilities for producing best-in-class pretraining corpora. That money has turned out to be a really solid investment. Since then, we've signed deals we project could generate around $42 million in revenue, and we anticipate closing more agreements soon that we think might contribute another $26 million — adding up to $68 million in potential revenue connected to these efforts, including both the signed ones and those likely to be signed before long. These initiatives involve 5 customers. They've only been going for a few months and are still gaining steam. As we see it, most of that anticipated revenue ought to arrive in 2026, even though we've already taken in more than we spent. Pretraining data is increasingly viewed as a strategic advantage for next-generation LLMs, and we feel we're well placed to take advantage of this early trend. Today we announced the launch of Innodata Federal, a dedicated government-focused business unit designed to deliver mission-critical AI solutions to U.S. defense, intelligence and civilian agencies. We expect this business unit to be a material revenue generator for us in 2026 and beyond. We're also announcing today that the business unit has won an initial project with a new high-profile customer. We think this first project will bring in roughly $25 million, with the bulk of that landing in 2026. There are more projects we're currently talking about with the same customer, and we believe those will be sizable. This new partnership matters a great deal strategically — not just because of how big it could become, but also because of the recognition and market leadership position we think it will give us. Before the year is over, we plan to put out a joint press release covering this relationship. For our next stage of growth, we see it as something that could truly change the game. We've also seen some encouraging early traction in the market. That includes our very first direct award from a big defense agency, talks that could lead to work with other well-known defense tech firms, and proposals we've turned in across the DoD, the intelligence community, and civilian agencies. The thing that sets Innodata Federal apart is our ability to handle the whole AI life cycle — we're not just doing data annotation or some single-point fix, but real end-to-end effort that runs from collecting data clear through to putting models into production and keeping them running out in the field. Leading technology companies around the globe, along with Fortune 1000 businesses, already depend on our platforms and our expertise. What we're doing now is bringing that proven track record into federal missions, delivering the security, compliance, and speed that government work demands. From where we sit, the moment is ideal. Agencies across government are making ambitious moves to put artificial intelligence into practice. Back in July, the White House put out its AI strategy for the country and signed 3 executive orders meant to smooth out purchasing and speed up deployment. At this very moment, the General Services Administration, or GSA, is overhauling how it buys things so that agencies can obtain AI solutions more easily. In the past, federal procurement was sluggish and buried in red tape, yet that is shifting quickly, and we plan to tackle both that demand and that opportunity directly. As we shared earlier today, General Richard D. Clarke — a retired four-star Army General who once led U.S. Special Operations Command — has come onto the Innodata Board; his expertise and relationships will help guide where Innodata Federal goes, and we couldn't be happier about it. Pushing further into the expanding sovereign AI market has been another key focus for us this year. That means government-led efforts around the world aimed at independently building, deploying, and governing AI systems as a matter of national interest. These programs seek to keep national control across the full AI technology stack, beginning with the semiconductors that train models and reaching all the way to the data that gives those models their intelligence. To our thinking, this stands as one of the most significant structural shifts in the global technology landscape. Sovereign capability building has set off enormous investment pushes directed by governments, in effect generating state-backed demand guarantees for the entire AI sector — from chipmakers and cloud providers through to data engineering companies such as ours, a market worth $million.
Jack Abuhoff
executiveHaving traveled to a number of nations across the Far and Middle East, what has really stood out to us is how much interest there is in what we offer. In most of those places, you won't find a domestic enterprise along the lines of Innodata that has demonstrated success in supporting generative AI and LLM efforts; we found ourselves quickly moving into advanced talks with sovereign AI organizations spanning multiple regions, and we anticipate making one or more strategic partnership announcements within the coming months. Their economic strength and their eagerness to move fast are genuinely remarkable, and our excitement about this newer growth area for the company could hardly be greater. At the same time, our enterprise AI practice is building momentum and looks promising for 2026. It delivers full stack support that helps enterprises bring generative AI into their products and operations. To illustrate, this practice is assisting a major social media platform in automating content monitoring and monetization workflows through generative AI, and it is helping a hyperscaler embed generative AI into data center operations for real-time analytics; typically, we expect such projects to begin somewhere between $1 million and $2 million, with considerable room to expand and repeat. We are likewise holding discussions regarding strategic relationships that might push our enterprise AI practice ahead in 2026. Agentic AI is the next initiative I want to address. As I've mentioned previously on these calls, our view is that Agentic AI will make generative AI genuinely useful within the enterprise, and that autonomous agents will before long be as commonplace as human employees carrying out many of their tasks. Agentic AI remains at a very early stage; alongside big tech model builders, we are evaluating and refining autonomous agents across numerous real-world use cases, developing evaluation models and human-in-the-loop systems built to measure, interpret and guide how agents behave. Our starting point is assessing task success — did the agent reach the goal? After that, we examine why the agent acted as it did and profile its general behavior in order to inform additional fine-tuning. These capabilities — diagnostic judge, task success judge and profiling judge — are being used more and more within RLHF and RLHA frameworks for Agentic systems, in which agents operate autonomously through multistep real-world workflows. We have additionally been creating agents inside our agility platform as a means of improving the product and advising a number of enterprise customers on bringing agents into their environments. That leads me to model safety, our sixth area of 2025 investment. As agents acquire autonomy, businesses need to figure out how to monitor them and keep making them better. Our aim is to turn into a trusted partner for software companies and other enterprises, assisting them in benchmarking for safety, reliability and ethical behavior. Here is a single illustration of the work we are doing right now. Not long ago, we started working with a leading chip company to stress test its multimodal AI products, simulating real-world risks such as data exfiltration, privilege escalation, instruction manipulation and multimodal injection attacks. So after we pinpoint weaknesses, we create focused remediation data, retrain the model, and verify outcomes using reproducible benchmarks. What we're aiming for is stronger model safety without any loss of capability resulting from that retraining. In our view, model safety represents a massive opportunity—so much so that we've brought in a top-tier global consulting firm to assist us in sharpening both our offering and our market approach in this space; that covers briefly the 6 areas where we've made investments during 2025, and a number of them are being disclosed publicly today for the very first time. Across all of them, we've put in relatively small amounts, yet the payoffs have been disproportionately large and product market fit arrived rapidly; in our view, some start-ups have brought in tens of millions at lofty valuations pursuing a number of these very same opportunities. Even so, we accomplish more, more quickly, while putting far less capital at risk. During this year, we expect to spend roughly $9.5 million on capability building investments across these and other comparable efforts; that figure includes $8.2 million in SG&A plus direct operating costs along with $1.3 million in CapEx; additionally, we are carrying costs tied to significant unused capacity inside the company, expecting business we believe will soon be won. Although we might have chosen to avoid these expenses and instead show a higher adjusted EBITDA, in our view these outlays are attractive short-cycle investments that set us up for faster growth in markets; we think we are ready to serve, and we think they will produce substantial gains during 2026 and afterward. We've likewise reinforced our leadership team and operating base for the scale we expect ahead. I'm happy to share that Rahul Singhal has been named President and Chief Revenue Officer. Rahul came to Innodata during 2019 and has played a key role in shaping our strategy while cultivating strong ties with our biggest customers. We're also bringing on 2 excellent new Board members, Don Callahan, who contributes extensive digital transformation knowledge from Citigroup and Time plus strong connections to Silicon Valley and Enterprise CEOs via Bridge Growth Partners; and retired four-star Army General Rich Clarke, former Commander of U.S. Special Operations Command, who offers exceptional defense perspective and robust federal relationships. Their backgrounds match our emphasis on big tech, defense and enterprise markets, and I'm certain they will help steer us through the coming phase of transformative growth. Last, I'd like to thank Nick for 5 years serving on the Board. Nick has been enormously valuable both to me personally and to the company. He's departing to focus his efforts on a fresh opportunity beyond our markets, and we wish him all the best. With that, I'll hand the call to Rahul.
Rahul Singhal
executiveThanks, Jack. It means a lot to me to take on this bigger responsibility. A lot of you probably noticed that Time Magazine just placed Innodata at number 24 on its first-ever America's Top 500 Growth Leaders list for 2026 — a list that honors businesses which "Capture trends and stay ahead of time." That way of thinking — spotting what comes next and moving on it quickly — sits at the heart of who we are today. You're witnessing the payoff right now; we're strengthening ties with both current and newly added Silicon Valley clients, while notching fast wins across the 6 investment areas Jack described a moment ago, and a growing count of the biggest tech firms and enterprises on earth are recognizing what we deliver. As we look beyond 2026, across the medium and long term, we expect our engagement with frontier model builders to grow broader and more intricate. Tomorrow's models won't simply require additional data. They'll require data that's smarter, drawn from simulation labs, massive synthetic generation, and RL gems that encode human judgment, context, and values. Beyond that, the AI enterprise services market — which we're now successfully positioning ourselves against — will probably end up 10 or more times bigger than the model builder market; our belief is that Innodata was purpose-built for this wide-ranging enterprise transition. Working side by side with frontier model builders gives us distinctive insight into the way large models get trained, tuned, scaled, and evaluated. And we're pulling off the trick of packaging that insight into solutions that deliver value to enterprises. Take one example: we've recently started offering model safety and remediation solutions that draw on what we've done hand in glove alongside engineering teams at leading AI hyperscalers over roughly the last year. As of today, we're extending those capabilities to one of the top SaaS software companies in the world and one of the top generative AI chip designers in the world. In brief, my belief is that we're only at the opening stage of the generational technology shift that Innodata sits at the center of and is positioned to capitalize on. When I survey the competitive landscape, you won't find even a handful of firms capable of handling $50 million, $100 million, or bigger order sizes in our space. And that's precisely what hyperscalers need today, along with sovereign entities. On top of that, they lack the demonstrated capacity to scale the organization, deliver flawless data accuracy, and stay highly nimble in responding to shifting client needs within a very dynamic environment. What an incredible moment to be alive, as the world undergoes a seismic change powered by AI, and to occupy such a privileged spot, helping lead a company that plays a critical role in catalyzing that change. I'll hand the call over to Marissa now. Once she's finished speaking, we'll be ready to take your questions.
Marissa Espineli
executiveThanks, Rahul and Jack, and good afternoon to all of you. For the third quarter of 2025, revenue came in at $62.6 million, which is a 20% rise compared with the same period last year. Looking at it sequentially, revenue grew 7% over the $58.4 million we posted in Q2. Adjusted gross profit during Q3 2025 totaled $27.7 million, which is $4.8 million higher, or 21% up year-over-year, with an adjusted gross margin of 44%. Adjusted EBITDA came to $16.2 million, equal to 26% of revenue, marking a 23% increase quarter-over-quarter, which shows the powerful operating leverage in our business. Net income in Q3 2025 was $8.3 million, versus $17.4 million during the same quarter last year. That decline stemmed largely from the tax benefit that came out of utilizing net operating loss carryforward in Q3 2024. We closed the quarter holding $73.9 million in cash, which is higher than the $60 million we had when the previous quarter ended and the $46.9 million at year-end 2024, and we made no drawdown on our $30 million credit facility with Wells Fargo. As Jack noted, given the momentum we currently have, we are reaffirming the guidance we gave earlier of 45% or greater year-over-year growth for 2025, and we expect growth in 2026 that could potentially be transformative. Thanks to all of you for being with us today. Operator or Michael, please open the line for questions.
Allen Klee
analystNice work on the quarter. I was just toting up the — you brought up several possible contract wins and what those might be worth. And if I — the ones where you gave dollar figures came to roughly $100 million; but something I couldn't tell is these — a number of them might be deals spanning multiple years. Do you have a sense of how much of that might land in 2026?
Jack Abuhoff
executiveAllen, that's a terrific question. I'd say the contracts that we — when we discuss annualized recurring revenue, those tend to be the ones we expect will sort of continue at the figure we cite as a year's value from that. Other contracts we mention, we're going to attempt some ramping up of certain ones during this quarter, but after that the revenue would mostly be landing in the following quarter — pardon me, the following year.
Allen Klee
analystAll right. Then regarding — you noted an additional spend, I believe the figure was $8.2 million of incremental SG&A. Could you clarify across what span that occurs? And how should we frame it — against which sort of base?
Jack Abuhoff
executiveThat's on a year-over-year basis, and it's incremental for 2025 compared with 2024.
Allen Klee
analystUnderstood. Then regarding your biggest client, I believe you've said repeatedly there's room to grow that relationship, and it might end up being quite substantial. Still, is there anything you can share about the current business with them? Should we view it as basically steady?
Jack Abuhoff
executiveThe connection there is solid, and that business is steady. As you'll notice, it climbed quarter over quarter. And like we talked about only a few minutes back, we received a verbal concerning what could be an extremely big new program that would arrive with that client. We haven't actually factored that into anything so far, since we don't know what the ramp-up might look like, yet it's definitely quite meaningful when measured against next year.
George Sutton
analystThat's quite a lot of news, and congratulations to both Jack and Rahul on taking on bigger responsibilities. Regarding what you said verbally, Jack, about your biggest client, my guess is that this would simply go through a statement of work that's already in place, which means you'd be able to pick up that business fairly fast?
Jack Abuhoff
executiveYes, that's right. In terms of how it works, it would go through the master services agreement that already exists and would likely be a fresh statement of work. Still, you're right that onboarding this new requirement will be extremely simple and smooth.
George Sutton
analystI was really pleased to learn about the win in the federal market; and that raises a question, which I believe you covered in what you said regarding GSA. But usually, to handle substantial business of this kind, being involved in a FedRAMP program is necessary. Could you explain how you're managing this through the GSA approach? Or how it differs from the standard FedRAMP approach?
Jack Abuhoff
executiveYes; so what we were trying to convey, I think, is that this is a perfect moment for us to launch this practice. The strategic focus that the federal government is placing on AI and enabling AI has been made clear, across the DoD, the IC and even civilian agencies. So that's there — and beyond that, they acknowledge that the programs and processes for procurement and acquisition are burdensome, and they're going to hold back the AI advancement they aim to achieve. And as a result, executive orders have been issued. I believe there might even be some fresh announcements on that topic coming out tomorrow. So if you consider these 2 factors together, the priority the government gives to AI, once more, covering all of federal on one hand, and then the relaxations they're implementing regarding acquisition and procurement, this truly couldn't be a better moment for us to be in that market.
George Sutton
analystGeorge, thanks — and then last, Rahul, you said something quite fascinating: the services opportunity might be 10 times larger than the model builder space. Could you flesh that out a bit more? And how much of it do you figure you've begun to observe so far?
Rahul Singhal
executiveYes, George. When you look at today's enterprise market alongside the frontier models, those models are increasingly being woven into workflows that are changing things, mostly for cost reduction right now. Before long, transformative workflows will emerge that create fresh business models and bring in revenue. As we mentioned, with one big social media company, we managed to save them over $24 million in costs. So it's early days. We're beginning to reach the point where Gen AI solutions are being rolled out across our customer base, and we're hoping to grow this service going forward.
Jack Abuhoff
executiveThanks. Yeah, I'd say Innodata is really performing from a place of strength right now. We posted yet another quarter that broke records. Revenue has never been higher. Profitability is on the rise, and the results came in above what our analysts had projected. As we look toward 2026, we can see the possibility of ongoing transformative growth, driven by relationships with the Mag 7 and other Silicon Valley leaders that keep getting deeper. That growth, we believe, will come from 2 places. First, there's the ongoing expansion we're generating with current and new clients; and second, the solid returns we're starting to see from investments we've made recently. Today, I covered 6 particular areas of investment. And in each of those areas, without exception, we're showing the results that come from doing exactly what earned us recognition from Time Magazine — spotting what's coming and moving quickly. So let me briefly recap a handful of these initial successes. To start, we've landed $68 million in fresh pretraining data deals, with $42 million already under contract and another $26 million that we expect to have signed in the very near future. We recently secured a $25 million deal with a new federal strategic client—we hope to reveal their name shortly—and we see this as potentially the first among numerous initiatives together. We also have another expansion lined up with our biggest customer, pending verbal confirmation, plus a $6.5 million verbal agreement for a deal with yet another major tech company. On top of that, fresh alliances are forming with prominent AI and sovereign AI organizations, and we anticipate sharing those details in 2026. So, we appreciate everyone being here today. Our enthusiasm for the future couldn't be greater. Thanks.
Operator
operatorEveryone, that wraps up the call for today. We appreciate you joining us. You're free to hang up at this point.
Read the full transcript via the API
You're viewing the first half of this call. Get the complete Innodata Inc. transcript — plus 132,000+ transcripts from 7,000+ companies, speaker segments, AI summaries and full-text search — through the EarningsCalls.dev API.
Get the API View API docs →This call discussed
For developers and AI pipelines
Programmatic access to Innodata Inc. earnings transcripts and 132,000+ others is available through the
EarningsCalls.dev REST API. Plans from $39.99/month — full transcripts, speaker segments,
full-text search, and the recently-added /api/v1/transcripts/recent polling endpoint for ETL pipelines.
More tools for stock research
OptionsBell: an unusual options activity scanner that tracks options flow across all US stocks and emails you when large trades appear in Innodata Inc. or any ticker on your watchlist.
StockMarketScan: a stock screener with alerts: 24 pre-built screens refreshed every 5 minutes, watchlists and email notifications. Both offer API and MCP access and fair monthly pricing.