On this page of StockholderLetter.com we present the latest annual shareholder letter from INNODATA INC — ticker symbol INOD. Reading current and past INOD letters to shareholders can bring important insights into the investment thesis.

 •  4/25/2025 Letter Continued (Full PDF)
 •  4/26/2023 INOD Stockholder Letter
 •  4/25/2024 INOD Stockholder Letter
 •  More "Business Services & Equipment" Category Stockholder Letters
 •  INOD Benford's Law Stock Score = 80

INOD Shareholder/Stockholder Letter Transcript:

Fellow Shareholders,
Innodata had a fantastic 2024, almost doubling our revenues organically year-over-year
and solidifying our position at the forefront of the generative AI revolution. We believe we
are now an integral part of the ecosystem enabling the push toward artificial general
intelligence or AGI and within a few years the progression to    artificial superintelligence    or
ASI.
On the Precipice of Transforming the World
Although it seems like hyperbole, we truly believe generative AI has placed humankind at
the precipice of wholesale transformation of our world. Consequently, despite
experiencing exceptional growth last year, Innodata finds itself at the very beginning of a
global trend that we expect will propel our growth trajectory forward for the foreseeable
future. The evidence of this unprecedented transformation is all around us. In less than
two and a half years since its launch, 10% of the world   s population is now using ChatGPT
and the adoption rate of large language models (LLMs) is growing at a geometric rate.1
As early as last year, we believed the goal of the big techs    generative AI innovation race
was to create AI models that could perform in a manner indistinguishable from humans
(sometimes referred to as    artificial general intelligence    or AGI). At that time, this goal
seemed elusive and far in the future. Now we anticipate the release of very strong
generative AI models that have the potential to exhibit AGI perhaps as soon as this year.
We now believe the AI built over the next five years will achieve superhuman capabilities. In
the coming world of    artificial superintelligence    (ASI), humans will have the capability to
make decisions with the help of a seemingly divine intelligence and innovation could
happen at lightning speed.
Today   s AI generates text and images in response to human (or system) prompts.
Tomorrow   s AI will be context-aware, infused with domain/expert knowledge,
independently reason, and develop and execute end-to-end plans of action both in
isolation and in concert with other AI. This is often referred to as agentic AI.
Our Strategy and Progress
We find ourselves in the enviable position of having the expertise, credibility and
relationships to help enable the world   s largest technology companies and leading
enterprises to create and capitalize on the AI future described above.
Paris, M. (2025, April 12). ChatGPT hits 1 billion users?    Doubled in just weeks    says OpenAI CEO. Forbes.
www.forbes.com
1
1

Innodata now plays an essential role in the generative AI ecosystem as a result of a
strategic decision we made in 2023 to pivot our organization   s focus to servicing generative
AI development and deployment. We have developed a broad range of advanced services
designed to support the full lifecycle of large-scale AI systems. These include data
collection, red teaming, benchmarking, and evaluation, as well as more advanced
multimodal fine-tuning and human preference optimization solutions. We also launched
development of our Generative AI Test and Evaluation Platform, which is intended to
provide enterprise-grade continuous model evaluation and benchmarking.
In parallel, we deepened our linguistic and dialect coverage and expanded our network of
global domain experts - bringing additional subject-matter expertise verticals. We believe
these investments further strengthen our ability to serve as a trusted partner to
organizations deploying cutting-edge, scalable, and safe AI solutions in increasingly
complex real-world settings.
We expanded our    AI for enterprise    program capabilities, developing solution
architectures that can help organizations across industries reimagine core workflows.
From vision workshops and AI strategy consulting to implementation and beyond, we
believe we are positioned to support enterprises at various steps of their AI journeys. This
includes building GenAI-powered copilots, enterprise knowledge assistants, automated
document processes, and other digital transformation solutions that drive measurable
ROI, productivity, and operational transformation.
We also expanded our capabilities to support more advanced use cases   positioning
ourselves to help customers deploy AI in contexts that demand specialized multimodal
training data, deep domain expertise, rigorous safety and quality frameworks, and more.
These efforts bore fruit as we landed a significant customer, one of the    Magnificent 7   ,
who is at the forefront of developing frontier generative AI models. From an initial contract
value of up $8.0 million in 2023, we grew this significant customer to approximately $135
million in annualized run rate revenues as of our February 2025 earnings call. This
assignment allowed us to showcase the quality of our work, breadth of our capabilities and
sophistication of our expertise in the generative AI space.
Now, we are supporting some of the world   s leading companies, including    ve of the
   Magni   cent 7,    because of our ability to deliver enterprise-grade training data with
astonishing accuracy, scale, and domain expertise. Enterprises can turn to Innodata when
their in-house resources or competitor vendors fall short   especially for high-volume, highcomplexity projects that require domain expertise, rapid scaling, or stringent quality
standards. But our value doesn   t stop at traditional and generative AI development. For
enterprises looking to harness AI across their operations, Innodata provides full end-to-end
support   from early-stage vision workshops and strategic consulting to implementation,
deployment, and post-launch evaluation.
2

Looking to 2025 and Beyond
Building on our success, in 2025 we will be investing in new solutions and capabilities that
we believe will enable us to continue to be the partner of choice for the world   s leading
technology companies. Our strategic priorities for 2025 include:
(1) Creating progressively more robust capabilities to feed the data requirements of
models designed to deliver AGI and ASI.
We believe the road to AGI, and eventually ASI, will be paved with data. We believe data
requirements will be increasingly diverse and complex: meta-learning and reasoning
data; computer use, agentic and operator data; domain-specific data; industry-specific
data; and data for world modeling and simulation.
These data requirements fall squarely within Innodata   s core competency to source,
create, and deliver across a wide range of modalities, languages, and specialized
domains. Our services span the full data lifecycle   from annotation and collection to
demonstration data creation and human preference optimization and beyond   in
addition to AI technical consulting   equipping clients to build, fine-tune, scale, and
adopt advanced AI systems with confidence. Innodata leverages a global workforce of
over 6,000 full-time professionals across domains, including experts with advanced
academic credentials, to ensure the integrity, relevance, and accuracy of the data we
provide. Our workflows are tailored to meet the demands of high-complexity sectors
such as medical, finance, legal, and others.
Innodata   s solutions are designed to integrate seamlessly into our clients    AI
development pipelines, including applications involving supervised fine-tuning,
reinforcement learning from human feedback (RLHF), AI red teaming, and more. Our
quality assurance processes, and human-in-the-loop practices, are structured to
reduce error rates, improve quality, support model alignment, and enhance
performance across a wide range of use cases.
We believe we are well positioned to solidify and extend the relationships we now have
with our eight Big Tech customers this year by serving their needs for complex data,
delivered at high quality, economically, and on-time.
At the same time, we are targeting dozens of potential new customers we believe are
also investing significantly over the next few years in building high-performing
foundation models of the future.
It is generally accepted that the most highly performing models will result from highquality, large-scale, highly consistent, and diverse supervised fine-tuning data. Our
customer base is spending billions on their generative AI initiatives and would not trust
3

just anybody for this critical task. Our track record of proven success makes us an ideal
partner.
(2) Building new solutions around Agentic AI for our Big Tech customers as well as our
enterprise customers.
Agentic AI refers to artificial intelligence systems that can autonomously initiate and
carry out complex tasks in pursuit of specified goals, with minimal ongoing human
input. These systems go beyond reactive execution   they exhibit goal-oriented
behavior, make decisions, adapt to changing contexts, and take initiative to achieve
outcomes.
In contrast to traditional AI, which typically responds to prompts or instructions,
agentic AI is designed to operate with a degree of independence, managing multi-step
processes, reasoning through uncertainty, and dynamically adjusting actions based on
feedback. It represents a shift from AI as a tool to AI as a collaborator   one that can
understand objectives, plan strategically, and act accordingly.
We believe human knowledge, captured as data and used to train models, will be the
key to building agentic AI ecosystems over the next several years. Much of this
uncaptured, but useful, data does not exist explicitly today, such as how to execute a
multi-step process using a series of websites or how to reason through complex
domain-specific problems.
We intend to develop tools to broadly capture human knowledge across an everexpanding suite of diverse difficult tasks and to perform this service on behalf of Big
Tech customers, AI labs, independent software vendors (ISVs), infrastructure
companies, governmental entities and vertically oriented companies.
Enterprise digital data is widely considered the world   s    new natural resource   , and
within this we believe lies the potential to augment human productivity and enhance
complex decision-making (as well as defeat cancer, reverse climate change, manage
the complexities of the global economy, to name a few). Within the enterprise,
however, much of this growing body of digital data is not yet suitable for fine-tuning
models or agents or for incorporating into LLM prompts. It exists only in unmanaged
data arrays and silos. We believe that models introduced within the next year will
increasingly be open source, so enterprises can improve upon them and customize
them     provided the data is ready. In conjunction with improvements in the cost of
inferencing, we believe this will catalyze innovation. We plan to assist enterprises in
everything from preparing and refining their raw data to helping them build (and monitor
the performance of) AI agents.
4

(3) Providing innovative solutions for trust and safety for Big Techs and enterprises
that anticipates and addresses the challenges of agentic AI workflows.
We believe the rapid adoption of agentic and multi-agent systems will push us to a new
phase of complexity when it comes to trust and safety. Unlike today   s more
predictable, single-agent systems, these new architectures will be made up of highly
autonomous, often collaborative, agents that can interact dynamically across
organizational and domain boundaries and with a wide array of tools, APIs, databases,
and LLMs. The attack surface grows exponentially, introducing new and nuanced risks.
At the heart of the shift is autonomy - agents making decisions independently, often
chaining together or delegating tasks across networks of other agents. This introduces
emergent behaviors that are harder to anticipate and monitor in real time. These multiagent systems will demand a new layer of safeguards that understand intent, enforce
proper tool usage, and keep agents on-task and contextually grounded. These are
challenges we intend to address.
(4) Aggressively invest in our organization to accelerate our growth
We believe we have a blue-sky growth opportunity ahead of us. We intend to broadly
and aggressively invest in our organization to add talent, technologies and capabilities
to capitalize on the growth possibilities ahead of us. We expect that the capital light
nature of our business model and scalability of our organization will allow us to show
strong growth in earnings and cash flows, despite these investments, and should
position us for continued growth for the foreseeable future.
Giving Back
In 2024 we continued to extend our i-Hope Program building computer labs at mostly
primary and secondary schools across India, the Philippines, and Sri Lanka. The idea of
this program, launched in 2016, is to harness the energies and talents of our staff to help
provide a technology education to children in economically disadvantaged communities
so that they may thrive in an increasingly AI-driven world.
We achieved our original goal of providing computer literacy to 25,000 children by 2025
ahead of schedule in the third quarter of 2023, bringing the total beneficiaries to
approximately 40,400 children, well above our original target. We continued our efforts in
2024, and since launching our i-Hope Program have contributed over 3,700 person-days to
this and other CSR programs. As a result of our work, approximately 42,500 children have
now become more technology-proficient and better equipped to seize opportunities in the
AI era. Our efforts have been widely recognized, and we have received many prestigious
awards.
5



 4/25/2025 Letter Continued (Full PDF)