Curiositystream - Earnings Call - Q2 2025

Executive Summary

• Q2 2025 delivered strong execution: revenue rose 53% year-over-year to $19.0M, sequentially up 26%, with record net income of $0.8M and record adjusted EBITDA of ~$3.0M, both above guidance.
• AI data/video licensing was the key driver: licensing revenue reached $9.3M, “significant new business” from AI partners; subscription revenue was also $9.3M, down YoY but up sequentially with new wholesale deals and Prime Video/DIRECTV expansions.
• Balance sheet remains solid: $30.7M cash/restricted/HTM securities and no debt; ordinary and special dividends in Q2 totaled $10.4M, and the Board declared a $0.08/share Q3 dividend.
• Guidance: Q3 2025 revenue range $15–$18M and FY 2025 adjusted free cash flow $11–$13M; Q2 results exceeded prior Q2 guidance ($16–$17M revenue; $2–$3M adjusted FCF).
• Potential stock narrative catalysts: accelerating AI licensing with repeat hyperscale partners, expanding distribution (Prime Video AU/NZ; DIRECTV SVOD/FAST), and management’s confidence in being a leading licensor of premium video for AI training.

What Went Well and What Went Wrong

What Went Well

• AI licensing scaled materially for the third consecutive quarter, powering record profitability: “Adjusted EBITDA grew by over $4M YoY… to positive $3.1M” and licensing revenue reached $9.3M, up >$8M YoY.
• Distribution footprint expanded: new/expanded multiyear wholesale agreements in Asia/LatAm; Prime Video add-on launches across Europe earlier and AU/NZ in August; DIRECTV multi-tier deal in the U.S. (SVOD + FAST).
• Management tone and strategic conviction: CEO reiterated intent to be among dominant licensors of video for AI training, citing a curated million+ hours corpus and superior data structuring capability (clipping/indexing/labeling/annotating at scale).

Quote: “We will be the or among the dominant licensors of video for AI model training.”

What Went Wrong

• Subscription revenue declined YoY by $1.7M, highlighting continued pressure in core D2C/partner distribution even as it improved sequentially.
• Cash cost of revenue increased due to revenue-share content licensing growth and storage, modestly offsetting content amortization reductions; gross margin improved only ~80bps YoY to 53.4%.
• Operating expense discipline continued, but ad/marketing + G&A still totaled $9.7M in Q2, reinforcing a need to maintain cost rationalization while scaling AI licensing and distribution initiatives.

Transcript

Speaker 4

Good afternoon. My name is Bella, and I'll be your conference operator today. I'd like to welcome everyone to the CuriosityStream Inc. second quarter 2025 earnings conference call. Please note that today's call is being recorded. All lines have been placed on mute to prevent any background noise. After the speaker's remarks, there will be a question and answer session. If you would like to ask a question, simply press the star then the number one on your telephone keypad. To withdraw your question, please press star one again. I will now turn the call over to Tia Cudahy with CuriosityStream Inc. You may begin your conference.

Speaker 2

Thank you and welcome to CuriosityStream Inc.'s discussion of its second quarter 2025 financial results. Leading the discussion today are Clint Stinchcomb, CuriosityStream Inc.'s Chief Executive Officer, and Brady Hayden, CuriosityStream Inc.'s Chief Financial Officer. Following management's prepared remarks, we will be happy to take your questions. First, I'll review the safe harbor statement. During this call, we may make statements related to our business that are forward-looking statements under the federal securities laws. These statements are not guarantees of future performance, but rather are subject to a variety of risks, uncertainties, and assumptions. Our actual results could differ materially from expectations reflected in any forward-looking statements. Please be aware that any forward-looking statements reflect management's current views only, and the company undertakes no obligation to revise or update these statements, nor to make additional forward-looking statements in the future.

For a discussion of the material risks and other important factors that could affect our actual results, please refer to our SEC filings available on the SEC website and on our investor relations website, as well as the risks and other important factors discussed in today's press release. Additional information will also be set forth in our quarterly report on Form 10-Q for the quarter ended June 30, 2025, when filed. In addition, reference will be made to non-GAAP financial measures. A reconciliation of these non-GAAP measures to comparable GAAP measures can be found on our website at investors.curiositystream.com. Unless otherwise stated, all comparisons will be against our results for the comparable 2024 period. Now, I'll turn the call over to Clint.

Speaker 1

Thank you, Tia. To state the obvious, we are living and operating today in an extraordinary and transformational time in media and technology. While I will talk later about what this revolution means for CuriosityStream, I'd first like to share our Q2 highlights. Quarterly revenue grew by 53% year over year from $12.4 million to $19 million, far exceeding the high end of our guidance. Revenue also grew sequentially from Q1 by 26%. Net income was again positive and improved by nearly $3 million year over year. Adjusted EBITDA grew by over $4 million year over year from negative $1 million to positive $3.1 million, also exceeding the high end of our guidance. More granularly, our subscription revenue increased sequentially, and our licensing revenue powered by video and audio for AI training grew considerably. In short, our business is strong and strengthening across the board.

Brady will provide more color around these and other key metrics in his part. In regard to our subscription revenue trajectory, we recently entered into new and expanded multi-year wholesale distribution agreements in Asia, Latin America, and the U.S., which we believe will ensure our overall subscription revenue is up and to the right for the foreseeable future. Further underscoring our confidence, we recently launched CuriosityStream subscription service and Curiosity University in new international markets with retail channel store partners like Prime Video Channels. We licensed a slate of traditional individual titles and series to both new and returning partners, including public broadcasters, pay-TV channels, and academic distributors across the U.S., Europe, Asia, and Latin America. Today, August 5, we premiere a major series on the world's most influential streaming platform, Netflix.

Titans: The Rise of Hollywood is a six-episode premium drama that chronicles the extraordinary rise of Hollywood's studio system, driven by the ambition and vision of first-generation immigrant pioneers. This world-class series blends power, scandal, greed, and profound insight into the human condition. We're eager to see how it resonates with Netflix's broad audience. The dataset licensing for AI training in the form of premium video, audio, scripts, and study guides grew substantially for the third quarter in a row. In addition to these premier ethically sourced corpuses, we also licensed about 9 million tokens of code, code for the first time ever.

While video sits firmly at the top of the dataset leaderboard in value and demand, our licensing of code is a testament to the value of controlling rights to all manner of IP for licensing and an illustration of the principle land and expand, which we are doing and will continue to do by overdelivering and delighting our partners. We are asked frequently by investors and content partners to help them better understand the lifespan and durability of data licensing for AI training and other initiatives. Is this recurring, they ask? Is this one and done? Do you know how many millions or billions of video hours are needed? What is your mode? What happens if you run out of video? Aren't you concerned about legal and regulatory issues? What will be the impact when the biggest studios move into this area? Will synthetic training data replace authentic data?

These are good and important questions, and while no one can perfectly predict the future, we have a tight team who has spent the last 16 months working on making Curi the dominant AI video licenser. This work has been done hands-on, day by day, partner by partner, shoulder to the wheel. While we have many unique advantages, four critical ones are: one, our deep and curated global premium video and audio library; two, longstanding relationships with rights-holding premium content producers around the world; three, programming services worldwide that we are simultaneously feeding; and number four, something that really sets us apart, the technical capability to now structure our data in a superior manner to our peers. We believe we've had more conversations across the licensing and model training ecosystem than our competitors, and we've translated these conversations into executed partnerships. Our perspective isn't theoretical.

It hasn't come from commissioning an overpriced McKinsey study or from spending time in the Yale faculty lounge. Our point of view is rooted in real-world experience and deed. Every new type of business model tends to go through a period of early chaos and uncertainty, but eventually the clouds clear and the blue sky presents itself. In regard to the questions I cited earlier, we know that large-scale AI models require enormous volumes of video data for training. Simply put, as they become more advanced, they need to be fed more. Scaling laws in AI show that additional video improves accuracy and generative capabilities, even with diminishing marginal returns. AI companies not generating continuous performance gains, they are losing. Further, freshness and recency are critical because cultural trends, products, and virtual references are constantly evolving. In other words, models need ongoing updates to stay relevant and avoid obsolescence.

As to synthetic data, it is incomplete. Simulated video can augment real data, but doesn't fully replicate real-world physics, actions, or the diversity and context of authentic video and data. We believe the market for high-quality, ethically sourced, rights-cleared video and audio content is incredibly durable and only growing for the foreseeable future. We're reviewing real video and audio RFPs. We do have some insight into the quantity requirements, millions of hours, and category and structural imperatives. Everyone should do their own research, but there are estimates that the industry-wide need for video could range from billions to tens of billions of hours. Some of what I just shared may sound a bit like consultant speak. I'll speak more plainly to the recurring nature of this business. Everyone wants seconds and thirds, and some already fourths and fifths.

For us, this is de facto recurring revenue, which again comes from having the overall best corpus of video and audio in the industry and from working to treat our partners on all sides of the equation like gold. Will you run out of content to license? No chance. Just as painters are always painting the Golden Gate Bridge, always, it never stops, we are similarly continuously creating and acquiring content for our streaming services and channels around the world. Further, the hyperscalers and the many other AI companies who are, and who we believe will, license video for training, want to work with partners who control a highly reputable, critical mass of content. They tell us that this typically means a minimum of several hundred thousand hours.

The long-term durability and recurring nature of this revenue is further sustained by the additional monetizable grants of rights that are emerging and will be required. Be clear, we are granting today only a training right. In the future, we will negotiate and license additional monetizable rights. A few examples include display rights, certain derivative rights, transformative rights, certain reproduction rights, and adaptation rights. Another near certainty is that we will be asked to license rights in the future that we haven't even contemplated today. Another common question, what is your moat? Most obvious, tangible moat is our superior volume of premium rights-cleared content available for license. The largest studios have libraries of 100,000 to 225,000 hours. We have control of and access to exponentially more hours than that, and our volume is growing every day. In addition, our ability to structure our data gives us a real competitive advantage.

I'm referring to our ability to clip, index, label, and annotate at scale. Hundreds of thousands of hours into any segment length in a very short time. To index, to annotate, and again to label at scale. Think traditional metadata on steroids, but steroids wildly more powerful than the ones given to the best Eastern European javelin throwers. This structured metadata makes the content we supply that much more valuable to our licensee partners. Intangibly, it's also simply action, action, action. Enhancing our existing relationships and building new ones. As we have done and as we continue to do the hard foundational work, we are well armed to execute at a scale we think beyond potentially anyone in the space. Again, I'm referring to our ability to structure our data, clip, index, label, annotate.

We are now able ourselves to create this type of metadata for our content that we would not have thought possible even three or four months ago. This metadata makes the content we supply that much more valuable to the AI licensees. Lastly, on this topic, it's critical to be able to distinguish between signal and noise. All of us are deluged every day with viewpoints around AI from seemingly every information source, like legal and regulatory concerns. What did the president just announce? What did the AI czar, David Sacks, say on the All In podcast? Will your job be replaced? What are the threats to humanity and our personal security that we need to address? How do we keep the God in the box? Is China beating us in the race? For our purposes, 99% of this is noise, meaning irrelevant or distracting data.

Now more than ever, we will be successful by simply focusing on the signal, the meaningful information we need to detect and understand and act on in the direct service of our business objectives, namely meeting the licensing needs of our AI partners. We see and hear the signal. We will not be distracted by the noise. Over the past century, value creation in media has consistently migrated to those able to capitalize on paradigm-shifting innovation. The first TV broadcast in 1928, the Global Satellite Leak of 1967 that brought billions together to watch The Beatles, and from cable and DPH in the late 20th century to the rise of YouTube in 2005 and Netflix's streaming model in 2007, every new wave crowned new leaders and left slow movers behind. Today, as we enter the mid-2020s, we are standing at the threshold of the most profound disruption and advancement yet.

This is not an iteration. It's a redefinition and one that will surely bring about a reordering across many, if not all, industries. Looking ahead to 2026, we at CuriosityStream are confident in two dynamics. One, we will license more video and data than we did in 2025, and two, we will be the or among the dominant licensers of video for AI model training. In closing, we believe our strong balance sheet, $31 million in liquidity and no debt, and our continued double-digit growth in both top-line revenue and cash flow, our leadership in AI video and data licensing, our library of over 1 million hours of video, and our $0.32 annual dividend position us as a high-performance outlier amid a historical technological revolution. The rapid acceleration of AI is not just reshaping industries; it is redrawing the competitive landscape.

We believe CuriosityStream is uniquely positioned to capitalize on this shift. Our diversified revenues from subscriptions, licensing, and advertising, our expanding ecosystem of technology and media partners, and public market currency create powerful operating leverage and optionality. Simultaneously, our disciplined cost rationalization efforts ensure efficiency without compromising growth. Over to you, Brady.

Speaker 2

Thank you, Clint, and good afternoon, everyone. Our full results will be presented in the 10-Q that we'll file in the next day or two, but let me quickly go through some of the second quarter results that we want to highlight. Clint said in the second quarter, we reported revenue of $19 million, exceeding our guidance and a 53% increase compared to $12.4 million a year ago. Continued to generate net income in the second quarter with earnings coming in at $0.8 million or $0.01 per share and a $2.8 million improvement from 2024. Likewise, we reported another quarter of positive adjusted EBITDA, which came in at $3 million, an improvement of $4 million from a year ago, and also the highest adjusted EBITDA in company history.

Adjusted free cash flow came in at $2.9 million, near the high end of our guidance range, and an increase of $0.4 million compared to last year. This also represented our sixth quarter in a row of positive adjusted free cash flow. Revenue for the second quarter was led by content licensing, which came in at $9.3 million, an increase from last year of over $8 million, driven by significant new business from AI content licensing. Our subscription revenue, which we consider our D2C, partner direct, and bundled distribution revenues, was also $9.3 million in the second quarter. This was a $1.7 million decline from last year, but a sequential increase from Q1, a trend that we believe will continue. Second quarter gross margin was 53%, a slight improvement from 52% a year ago.

While we're seeing continued reductions in content amortization, our cash cost of revenue increased slightly, a result of growth we're seeing in the licensing of content that we have acquired through revenue share arrangements and associated storage costs. Regarding other operating expenses, combined costs for advertising and marketing plus G&A were down 8% compared to last year as we continued to benefit from our ongoing cost rationalization. Excluding stock-based compensation, G&A declined 10% from a year ago. As I mentioned earlier, adjusted EBITDA was $3 million in the second quarter compared to a loss of $1 million a year ago, and adjusted free cash flow was $2.9 million in the quarter compared with $2.5 million a year ago. In June, we paid dividends of $10.4 million, including our ordinary Q2 dividend of $4.6 million, as well as a $5.8 million special dividend.

At yesterday's closing price, our shares currently provide for about a 6.5% dividend yield. We ended the quarter with total cash and securities of $30.7 million and no outstanding debt. We believe our balance sheet remains in great shape and that this provides us significant operating flexibility. Looking forward for the third quarter, we expect revenue in the range of $15 to $18 million, and for 2025, we expect adjusted free cash flow in the range of $11 to $13 million for the full year. With that, we can hand it back to Bella and open the call to questions.

Speaker 4

At this time, I would like to remind everyone in order to ask a question, press star, then the number one on your telephone keypad. We will pause for just a moment to compile the Q&A roster. Your first question comes from the line of Laura Anne Martin with Needham & Company. Please go ahead. Your line is now open.

Speaker 3

Okay, great results. Congratulations, you guys.

Speaker 2

Thanks, Laura.

Speaker 1

Thank you, Laura.

Speaker 3

My first question is, financial theory tells us we must ignore soft costs. My question, Clint, is why are you in the media business, the core media business?

Speaker 1

Oh, why are we in the core media business? Yeah, it's a great question, Laura, and I appreciate you asking that. I think we're in the core media business because we have a subscription video on demand business that is strong, that is global, that is durable, and I think represents the core of who we are as a company. I think you may be asking, like, okay, if you're doing this type of technology licensing, why do you continue to do this? Our answer is all of this works hand in glove. All of this works closely together. The vision that we have for CuriosityStream over the near, mid, and long term, Laura, is that we will have three solid revenue pillars: a subscription business, a licensing business, and an advertising business. We believe our subscription business will grow steadily.

We've done a lot of work to enter into certain wholesale agreements and other agreements that we believe really firm up the foundation and put us in a position to grow steadily. We believe subscription will grow steadily. Licensing we believe will grow rapidly and is certainly a high-growth area. Advertising, which for us is still nascent, I think holds significant steady and high-growth opportunity over time. I say everything works together because the reason that we've been able to build such a large library for AI content licensing is because of all the existing relationships that we have around the world with producers and distributors. We're constantly acquiring content for our three subscription services, for our linear pay-TV channels, for AVOD, for FAST.

Because we're doing this, because we have the people in place, the processes in place to do this, it's not a huge extra lift to add on the ability to license AI rights or other rights as well. We're in it because it all works. It all works hand in glove. It all works together. I think it'd be very difficult to be in just one of these today.

Speaker 3

Okay. What are the extra costs? To the extent you start aggregating more and more content and licensing to a broader swath of people, they all have different needs. They all have different tech stuff. What is, what is, should we expect to see the cost increases happen as you pivot more towards this high-growth licensing for inference business?

Speaker 1

One of the things we're very proud of is we've amassed a library of over a million hours of content, primarily on a rev share basis. What that means is, as we license out content, obviously, if it's our content, there's 100% margin. If it's our partner's content, we're paying out a piece to them. It's maybe 40 to 50% margin, something in that frame. Our approach, because as we got into this, we knew a couple of things. We knew that we were going to have to amass a lot of content in order to make it a meaningful business. We knew that we need to work with a lot of partners. At the same time, we also have told you and many other people that we're going to run a profitable business.

We needed to do all of that in a way that didn't layer on costs to our business today. We can continue to amass content like this over time. Would it make sense for us to pay outright for certain large libraries that are available? We'll look at that on an ongoing basis. Essentially right now, the only cost that we see going up would be storage and delivery costs to some extent. In the grand scheme of things, those are relatively de minimis.

Speaker 3

Hosting costs and R&D costs and hiring tech geniuses, that doesn't add a lot to the cost structure?

Speaker 1

It does not. It does not, Laura. We pride ourselves on being a tight company of 42 full-time people. I think over the course of this year, we'll be at $1.5 million to $2 million per employee in revenue.

Speaker 3

Excellent. Thank you.

Speaker 1

Thank you, Laura. Much appreciated.

Speaker 4

Again, if you would like to ask a question, press star one on your telephone keypad. Your next question comes from the line of Kris Tuttle with IPO Candy. Your line is now open. Please go ahead.

Speaker 0

Hi, thanks a lot. You guys did a great job covering everything, so I don't have a lot of questions. One thing you did say that surprised me, Clint, is you said something about code. I guess I wanted to understand a little bit more about what that is and how that, you know, if that's something that we should be looking for more of in the future. That's a pretty interesting comment.

Speaker 1

Thank you for asking, Chris. As we began licensing data for AI training, we took a look at what we had, and we took a look at the industry. Obviously, like video, as I said, that sits at the top of the leaderboard in terms of value, in terms of demand. Audio is kind of underneath that. Images, study guides, scripts, those are valuable as well. We did also have code. In the early days, the earliest days of AI training licensing, there was a lot of code that was scraped from internet sites like GitHub. There was also some proprietary code that was not there that people licensed out. I never thought in a million years, to be honest with you, that we would be able to license our code. We have included it in our sales materials with a number of different partners.

As we build relationships with these companies and as their needs evolve and change, it is something that we have. I cannot really speak to the long-term value of it, the long-term nature. I just think it goes to the value in owning and controlling good IP. That is, I think, the big point here. That is why I think if you control millions of hours of video and other data, there will always be ways to monetize that. I just point this out as something that is kind of unique, unsuspected, and representative of something else that I said, like talked about these additional rights that we will monetize. I think it is, in addition to potentially data that we are not aware is valuable, I think there will be additional licensing rights that we have not even contemplated today. Hopefully, that is helpful.

Speaker 0

Yeah, I mean, one slightly expanded follow-on question there, which I guess gets a little bit into what Laura was bringing up. You know, there's entertainment content, and then, you know, there are, I can't even tell you how many millions of hours of training video content has been built at very high levels of quality, for instance, in the hospitality industry and things like that. Are those areas that you're currently involved in? How significant, I mean, is that on the roadmap? I'm just trying to gauge where that is on the timeline.

Speaker 1

I think that's a really good question, Chris. We're not actively looking at that right now. What we're primarily focused on amassing in the video space is obviously continuing to build out our massive factual entertainment library. As part of this, we've definitely increased our scripted content, meaning general entertainment, movies, series, animation, kids. We've increased significantly our sports content, and we've increased significantly some of our audio content as well. As it relates to those types of videos, whether they're promotional or instructional, I think they can absolutely have value, particularly if they've, the challenge comes, have they been in front of the paywall or behind the paywall? A lot of that content, if it's free running around on the internet, there's a good chance that it's already been trained on.

At the same time, there is a whole lot of video, to your point, outside of what traditional media companies own that has value in the licensing space. If you believe that these models will need billions or tens of billions of hours of video, and it certainly can't all be synthetic, and it can't all be stuff that's been freely available on the internet, it could definitely have value.

Speaker 0

Yeah, all that stuff is, you know, Four Seasons doesn't put that stuff, you know, outside of the paywall.

Speaker 1

Right. Yeah, it does become of value. You just need a lot of it.

Speaker 0

Right. All right. Listen, thanks a lot. I will stay tuned for the Q, and after I go through that, I'll circle back with you guys in a few.

Speaker 1

Thank you for the question, Kris.

Speaker 4

That concludes our Q&A session, ladies and gentlemen. Thank you all for joining in. Everyone, have a great day.