Innodata Inc. (NASDAQ:INOD) Q4 2022 Earnings Call Transcript

Innodata Inc. (NASDAQ:INOD) Q4 2022 Earnings Call Transcript February 25, 2023

Operator: Greetings. Welcome to Innodata’s Fourth Quarter and Fiscal Year 2022 Earnings Call. Please note this conference is being recorded. I will now turn the conference over to your host, Amy Agress. You may begin.

Amy Agress: Thank you, John. Good afternoon, everyone. Thank you for joining us today. Our speakers today are Jack Abuhoff, CEO of Innodata and Marissa Espineli, Interim CFO. We will hear from Jack first who will provide perspective about the business and then Marissa will follow with a review of our results for the fourth quarter and the 12 months ended December 31, 2022. We will then take your questions. First, let me qualify the forward-looking statements that are made during the call. These statements are being made pursuant to the Safe Harbor provisions of Section 21E of the Securities Exchange Act of 1934, as amended and Section 27A of the Securities Act of 1933 as amended. Forward-looking statements include, without limitation, any statements that may predict, forecast, indicate or imply future results, performance or achievements.

These statements are based on management’s current expectations, assumptions and estimates and are subject to a number of risks and uncertainties, including, without limitation, the expected or potential effects of novel coronavirus COVID-19 pandemic and the responses of governments, the general global population, our customers and the company thereto; impacts resulting from the rapidly evolving conflict between Russia and the Ukraine; investments in large language models that contracts may be terminated by customers, projected or committed volumes of work may not materialize; pipeline opportunities and customer discussions, which may not materialize into work or expected volumes of work; acceptance of our new capabilities; continuing Digital Data Solutions segment reliance on project-based work in the primarily at-will nature of such contracts and the ability of these customers to reduce, delay or cancel projects; the likelihood of continued development of the market, particularly new and emerging markets that our services and solutions support; continuing Digital Data Solutions segment revenue concentration in the limited number of customers; potential inability to replace projects that are completed, canceled or reduced; our dependency on content providers in our Agility segment; a continued downturn in or depressed market conditions, whether as a result of the COVID-19 pandemic or otherwise; changes in external market factors; the ability and willingness of our customers and prospective customers to execute business plans that could rise to requirements for our services and solutions; difficulty in integrating and driving synergies from acquisitions, joint ventures and strategic investments; potential undiscovered liabilities of companies and businesses that we may acquire; potential impairments of the carrying value of goodwill and other acquired intangible assets of companies and businesses that we may acquire; changes in our business or growth strategy; the emergence of new or growth in existing competitors; our use of and reliance on information technology systems, including potential security breaches, cyber attacks, privacy breaches or data breaches that result in the unauthorized disclosure of consumer, customers, employee or company information or service interruptions; and various other competitive and technological factors and other risks and uncertainties indicated from time to time in our filings with the Securities and Exchange Commission, including our most recent reports on Form 10-K, 10-Q and 8-K and any amendments thereto.

We undertake no obligation to update forward-looking information or to announce revisions to any forward-looking statements except as required by the federal securities laws, and actual results could differ materially from our current expectations. Thank you. I will now turn the call over to Jack.

Jack Abuhoff: Thank you, Amy. Good afternoon, everybody. Thank you for joining our call. Today, I am going to talk briefly about our Q4 and year-end results. And then I’m going to spend some time discussing recent acceleration in AI investment by large technology companies in large language models coinciding with OpenAI’s fourth quarter release of its large language model called ChatGPT and how we believe Innodata is quite well positioned to capitalize on this increased investment. So first, our results. Q4 revenue was $19.4 million, a 5% increase over the prior quarter, which annualizes roughly to a 22% growth rate. We posted positive adjusted EBITDA of approximately $250,000 in the quarter, which was a positive swing of $1.5 million from Q3.

This significant improvement resulted primarily from our September/October cost containment and efficiency initiatives. The benefits of these initiatives will be fully reflected in our first quarter 2023 results. We ended the year with a healthy balance sheet, no appreciable debt and $10.3 million in cash and short-term investments on the balance sheet. In 2022, overall, we grew revenues 13% despite the significant revenue decline from our large social media customer that underwent significant internal disruption in the second half of the year, but that we believe may normalize this year. Let’s now shift to the recent substantial uptick we are seeing in our market activity. As most everyone now knows, in late Q4, OpenAI unveiled ChatGPT. This AI large language model has since gone viral, capturing popular imaginations for its ability to write, to generate computer code and to converse at what seems like human or even superhuman levels of intelligence.

We believe the release of ChatGPT has been broadly seen as a watershed event potentially heralding a fundamental advancement in the way AI can drive changes in business communication, processes and productivity. Our market intelligence indicates that many large tech companies are accelerating their AI investments as they compete for domination in building and commercializing large language models and that an arms race of sorts is now forming. We believe that the significant investment that will likely result from this competition could dramatically accelerate the performance of these large language models. As a result of this dramatic increase in performance, we expect almost every industry will face fundamental reinvention. We believe that the opportunity for Innodata in all of this is significant and that it is now upon us.

We believe our opportunity is actually threefold: first, to help large technology companies, both existing customers and new customers, compete in this large language model arms race; second, to help businesses incorporate large language models into their products and operations; and third, to integrate these technologies into our own platforms. Let’s take each of these in turn, starting with what I just laid out as our first opportunity, helping technology companies, both existing customers and new customers, compete in the large language model arms race. While ChatGPT and a host of lesser-known but equally impressive large language models are for sure amazing, our view is that it’s still early days. We believe these large language models have room for significant improvements in output quality, in the languages they serve, in the domains they support and in terms of safety.

These are all challenges that we believe we can help with. We expect to help by collecting large-scale, real-world data for training; by creating high-quality synthetic data when real-world data training is hard to come by; by annotating training data; and by providing reinforcement learning from human feedback, or RLHF, to fine-tune model performance and eliminate hallucinations, which is the tendency of these models to make things up on the fly. In addition, we expect to help by minimizing the risk that models generate unsafe or biased results, and we expect to help by hyper-training generalized models for specialized domains. High-quality data is at the root of addressing all of these challenges, and this is and has been Innodata’s bread and butter specialty for 30 years.

We believe that the arms race to which I’m referring has likely already begun. In just the past few weeks, it seems, activity for us has dramatically surged. We are now either expanding work with, beginning work with or discussing working with 4 of the 5 largest technology companies in the world. I am going to share some examples of the surge in activity we’ve seen in just the past few weeks. A major cloud provider, whose AI needs we began serving 24 months ago, engaged us to help them build a new large-scale generative AI model for images. We started the initial phase of this just this week. In addition, the customer asked us just last week to kick off a pilot to support their generative AI large language model development. We started the pilot this week.

With the same customer, also in the last few weeks, we expanded our synthetic data program to support its large language model development. We believe high-quality synthetic data is likely to be a key ingredient to performing €“ to high-performing large language models of the future. Synthetic data is entirely new data that we generate through a machine-assisted process to match real-world data and maintain all of the statistical properties of real-world data, which is especially useful for capturing rare cohorts and outliers of interest. Synthetic data is also helpful to correct for data bias, to improve algorithmic fairness and to avoid having to retrain proprietary or confidential data. We started working with synthetic data back in 2022 and we have been continually improving our capabilities and technologies for synthetic data creation since then.

With this customer, we’ve gone from serving just one of their product lines to now being firmly engaged with three product lines and we are in pilots with three additional product lines. Also in the past couple of weeks, with another of the world’s largest tech companies, this one a company that would be a new customer for us, we’ve gotten a verbal commitment to assist them on projects relating to large language models. They have told us that they are in the final stages of putting in place a statement of work. While there can be no assurance that the SOW is put in place, based on our current estimations and assumptions, value of this program could potentially approach approximately $1.8 million per year run rate in its initial phases and could ramp up to approximately $6 million per year as it gains momentum.

In addition, 2 weeks ago, one of the world’s largest social media companies, another potential new customer for us, reached out to discuss how we might potentially support its large-scale model development. It has been referred to us by one of our existing customers who apparently said that we could be helpful in unlocking value, unlocking scale and by bringing a consultative approach to a partnership. We believe that the opportunities I’ve just mentioned individually and in the aggregate are potentially very large. I want to underscore that several of these are pipeline opportunities and at various stages of pipeline from early stage to late stage. Pipeline opportunities are inherently difficult to forecast and often do not close. That said, I’ve offered them here in support of two beliefs: the first belief that there is building momentum among big tech companies for AI innovation generally and large language model specifically; and the second belief that Innodata’s reputation for high-quality work with high-quality outcomes is becoming firmly instantiated in a dynamic market that is viewing us as a potential partner in one of our generation’s greatest innovations.

Now, let’s shift to our second significant market opportunity. We believe that our second significant market opportunity is to help businesses harness the power of these foundational generative AI models. Most enterprises have tasks that generative AI can make easier. As the technology improves, and we expect it will, we believe that businesses will see incorporating the technology as a must-have rather than a nice-to-have. Analysts are predicting that this year, the most forward-thinking business leaders will be actively putting time and money into reimagining their products, service delivery and operations based on what AI can do for them, leading to widespread deployments over 2024 and 2025. What we are also hearing, especially from CTOs, is that their biggest roadblock to deploying AI is finding the right engineers and data scientists to help them get there.

We believe our opportunity will be to do just that to help them get there. We anticipate that this will take the form of fine-tuning existing pre-trained large language models on specific tasks within specific domains, bringing expertise in prompt engineering, the art of prompting large language models to produce the appropriate results, and helping with large language model application integration. Early in the first quarter of 2023, a large financial technology company expanded scope with us to leverage our proprietary AI models more fully and reengineer their technology for the cloud to drive operational efficiencies. Our proprietary AI engine, Goldengate, uses the same underlying encoder-decoder transformer neural network architecture as GPT.

While GPT is trained broadly, Goldengate is trained narrowly on specific tasks and domains. We have experimented with coupling GPT and Goldengate, and this seems to result in even higher orders of performance. This is the third scope expansion we’ve had with this company over the course of the past 6 months, again, providing further validation of our land-and-expand strategy. We believe our third opportunity is to harness GPT and other large language models in our own AI industry platforms. Just last month, we announced PR CoPilot, a new module within our Agility PR platform that combines proprietary Innodata technology and GPT to enable communications professionals to generate first draft of press releases and media outreach in record time.

With our release of PR CoPilot, we became, we believe, the PR industry’s first integrated platform to incorporate large language model technology. The implementation was significant for Innodata, and we received a supportive write-up in PR weekly for it. The start-up named Jasper vaulted to unicorn status when it implemented something very similar to PR CoPilot for creating blogs and social media postings. Their efforts got them a $125 million Series A round on a healthy $1.5 billion valuation. With respect to our Agility platform, we are seeing positive momentum in key performance indicators, which we think PR CoPilot and our newly integrated social media listening product will help to further accelerate. In Q4, Agility platform sales grew 6% over Q3, which annualizes to a roughly 26% growth rate.

In 2022 overall, our direct sales new logo bookings increased by 83% year-over-year and our direct sales net retention increased to 100%. In 2022 overall, approximately 83% of our Agility revenue came from direct sales, and 17% of our revenue came from channel partners. In Q4, our conversion from demo to win in direct sales increased to 33%, up from approximately 18% at the beginning of the year. We believe the notion that customers who use us love us is also very much apparent in our Synodex platform. Synodex grew by 71% in 2022 with a net retention of 168%. We announced in Q4 that one of our large Synodex customers had expanded its recurring revenue program with us. In the announcement, we stated that the expansion was valued at approximately $600,000, but we now believe the value of the expansion is actually closer to $1.2 million.

This is now our second-largest Synodex customer with an estimated annual recurring revenue base of $2.3 million. This year, we will be focused on product development to expand our addressable market for medical data extraction. We’ve got new products currently being evaluated by charter customers in disability claims processing, personal injury claims processing and long-term care claims processing as well as in clinical medical data annotation and fully automated life underwriting. Integrated AI will be a feature in all of these products. We are more enthusiastic than ever about our market opportunity and the intrinsic value of our business. In our last call, we said we anticipated expanding our adjusted EBITDA to $10 million or more in 2023 and, at the same time, capturing significant growth opportunities.

We believe the activity we are now seeing in our markets will likely enable us to achieve this and potentially more. I will now turn the call over to Marissa to go over the numbers, and then we’ll open the line for some questions.

Marissa Espineli: Thank you, Jack. Good afternoon, everyone. Let me recap the fourth quarter and fiscal year 2022 financial results. Our revenue for the quarter ended December 31, 2022, was $19.4 million compared to revenue of $19.3 million in the same period last year. Our net loss for the quarter ended December 31, 2022, was $2 million or $0.07 per basic and diluted share compared to a net loss of $1.2 million or $0.04 per basic and diluted share in the same period last year. The total revenue for the year ended December 31, 2022, was $79 million, up 13% from revenue of $69.8 million in 2021. Net loss for the year ended December 31, 2022, was $12 million or $0.44 per basic and diluted share compared to a net loss of $1.7 million or $0.06 per basic and diluted share in 2021.

Adjusted EBITDA was $2 million in the fourth quarter of 2022 compared to adjusted EBITDA of $0.3 million in the same period last year. Adjusted EBITDA loss was $3.3 million for the year ended December 31, 2022, compared to adjusted EBITDA of $3 million in 2021. Our cash and cash equivalents and short-term investments were $10.3 million at December 31, 2022, consisting of cash and cash equivalents of $9.8 million and short-term investment of $0.5 million, and $18.9 million at December 31, 2021, consisting of cash and cash equivalents. So thanks, everyone. John, we are now ready for questions.

See also 12 Most Promising New Tech Stocks To Buy and 12 Most Promising Low Cost Stocks.

Q&A Session

Operator: Thank you. The first question comes from Tim Clarkson with Van Clemens. Tim, please proceed.

Tim Clarkson: Hi, Jack. Apologize, if you head noise in the background, we had a major snowstorm in Minneapolis and the snow is coming off our buildings, so anyhow. The first question I have is, what exactly is the technology experience that makes Innodata successful at moving forward artificial intelligence in these chatbots?

Jack Abuhoff: Sure, Tim. Well, it’s not limited to chatbots. Artificial intelligence, people believe and I firmly believe, is at a kind of a fundamental inflection point. We’re now seeing the kinds of technologies that people have dreamt about probably since the 1950s. And when you think about building these technologies and think about what goes into them, it’s not programming in the traditional sense, it’s data. It’s high-quality data and data that can help to address some of the fundamental problems that these technologies have. They need to improve their output quality. They need to improve languages they are supporting. They need to be customized for particular domains. They need to improve what we think of as safety, the kinds of responses, the kinds of things that they tell us.

So what needs to be done for that? The things that need to be done are the things fundamentally that we’ve done for a very long time very, very successfully for some of the largest companies out there. When we were being retained by the large engagements we’ve had in the past, things like Apple, what we were doing for them was fundamentally building large quality data or high-quality data but for products and for publishing. Here, we’re building it because data is the programming language of AI and the programming language of large language models.

Tim Clarkson: Right, right. And what typically are the gross margins on the revenues you’re getting from this? Are these high gross margin products?

Jack Abuhoff: I think from a gross margin perspective, I would continue to expect a range of gross margins from our different capabilities. In the services sector, I think mid-30s to mid-40s gross margins are achievable, and they’ll get better over time. As we introduce automations in technology, they tend to drift higher. When we’re starting up new projects, they tend to drift a little bit lower. On the platform side, they are higher than that. Incremental gross margins especially can be very substantial. And as we build scale and start to scale on our fixed costs, we can start to see the kinds of gross margins that will emerge from those kinds of business models.

Tim Clarkson: Right, right. Well, just one last comment. I’ve been all excited about Innodata lately. And the analogy I use is it’s what’s happening with artificial intelligence is before it was sort of like da Vinci seeing a picture of an airplane. But it’s one thing to see a picture of an airplane, it’s another one to see one fly by you and go from Minneapolis to New York. And when you actually see this artificial intelligence stuff work, you no longer have to be sold on the value. I mean it’s magical. So that’s, for me, that’s the difference as people are really excited about the end product, and emotion is what drives ultimately decision making, and there is real excitement behind this. So with that, I’m done.

Jack Abuhoff: Thank you, Tim.

Operator: The next question comes from Dana Buska with Feltl and Company. Please proceed.

Dana Buska: Hi, Jack. How are you today?

Jack Abuhoff: Dana, I am doing great. Thank you, welcome to the call.