Jay Prasad

CuriosityStream Transforms from Streaming Service to AI Licensing Powerhouse, Boosting Revenue by 46% https://buff.ly/efQn8Wi Science-Centric Streaming Service CuriosityStream Is an AI-Licensing Firm Now In a move that’s reshaping the future of content monetization, CuriosityStream — the science and nature-focused streaming service founded by cable pioneer John Hendricks — has quietly transformed itself into a major player in the artificial intelligence (AI) licensing space. As of late 2025, CuriosityStream is no longer just a subscription video-on-demand (SVOD) platform; it’s now a leading licensor of high-quality, factual video and audio content for training large language models (LLMs) and other AI systems. A New Revenue Revolution CuriosityStream’s pivot to AI licensing has proven to be a game-changer. In its third quarter of 2025, the company reported a staggering 46% year-over-year revenue jump, reaching $18.4 million, with nearly half of that income now coming from licensing its vast library to AI developers. This new revenue stream, which didn’t exist just two years ago, has helped the company move from losses to profitability and has positioned it as a first mover in a rapidly evolving market. The company’s success is rooted in its unique position: it owns or controls access to over 2 million hours of factual content, including nature documentaries, science series, and educational programming. This content is highly sought after by AI companies because it’s not only high-quality but also ethically sourced and free from the complex rights issues that plague scripted entertainment. Why Factual Content Is Gold for AI Unlike scripted shows or movies, which are often encumbered by talent agreements, guild restrictions, and copyright complexities, factual content like that offered by CuriosityStream is much easier to license for AI training. As CEO Clint Stinchcomb noted, “It’s certainly easier when you’re in the nonfiction space than when you are in the world of scripted entertainment and you have talent to deal with and guilds. It’s fraught with lots of issues.” CuriosityStream’s library provides AI developers with hours of labeled, annotated footage of landscapes, animals, scientific phenomena, and more. This kind of data is invaluable for training video models, improving search and recommendation algorithms, and advancing the capabilities of generative AI systems. The Mechanics of AI Licensing CuriosityStream’s AI licensing model is straightforward but powerful. The company licenses its own content and also acts as a broker for third-party content owners, who can step into existing deals and share 50% of the AI revenue. This approach has allowed CuriosityStream to scale its offerings quickly and attract a wide range of partners. The company is currently working with nine major LLMs, including Google Gemini, and is in…

4

My agency, Mostec.ai, is a management-led AI company for a reason. MIT says 95% of AI projects generate zero return. BCG adds that “70% of AI transformation is people.” That matches what I’ve seen. Here is a simple example. A company struggling with lead generation installs an AI automation system that suddenly brings in 3–4x more leads. Not only do they have enough for every salesperson, but they now have twice as many as they can comfortably handle. Excitement is high, salespeople push hard… until it becomes overwhelming. Almost inevitably, instead of taking responsibility for every lead, they begin to select, picking the “best” leads and ignoring the rest. The logic sounds reasonable. Rather than giving quick, shallow attention to all leads, they picked only the most potent ones and invested more energy there. Seems smart, right? Focus energy where it matters most. Except it rarely works. When I ran the top outbound insurance sales teams, agents did the same thing. They would look at the lead data and decide to call only the ones that looked the most promising. For example, city residents instead of rural ones. Certain profiles, instead of others. Those who already have the most insurance policies. Again and again, the results showed the same: it was far better to define solid qualification criteria upfront and then systematically process every lead. This kind of situation has very little to do with AI itself. In both cases—whether employees follow through on every lead or start cherry-picking—the AI works perfectly. The difference is in management. Technology today does what it’s supposed to do. The real results are created where AI stops, and human nature takes over. That’s why at Mostec, we center our AI systems around people, not just tools. It’s people that make the difference. Always. --- In your experience, where do AI projects stumble more: the technology, or the management around it?

10

2 Comments

Is PE & VC AI Governance Theater? Many PE & VC firm leaders believe they are governing AI adoption, yet in reality, they've never transferred their own expertise to an AI system. Governance without internal AI proficiency is theater. Ops2AI helps PE & VC leaders codify their diligence, value creation, and judgment criteria, enabling Associates to learn fifty percent faster. This, in turn, allows firms to guide portfolio AI adoption for the right business value levers and build leadership succession engines. Without capturing and sharing internal intelligence at the firm and Leadership level, you're just acting. Firm-to-Portfolio AI = IP Capture >Training > Execution > The AI Lifecycle

2

Citrini Research (the Wall Street Analysts) nailed their 2025 predictions. Now they're calling #DeviceNativeAI for 2026. After accurately forecasting this year's major investment themes, Citrini has published their outlook for 2026, and "inference on device" (AI that runs on phones) sits prominently in second place among the key predictions for the next 12 months. Their analysis cuts through the hype with refreshing honesty. Yes, cloud AI is powerful. But the economics break down when you're running always-on agents that need to process every keystroke, every pixel, every moment of context throughout your day. Three forces are converging right now: 1️⃣ Mobile devices rapidly gaining AI-capable compute power (NPUs improving 2x per generation) 2️⃣ AI models are becoming radically more efficient and device-optimised 3️⃣ Always-on agentic AI requiring sub-200ms latency that cloud simply can't deliver 🟠 The bull case is compelling: when your AI agent needs to read your screen, book your Uber, and handle real-time translation, waiting for cloud round-trips isn't just slow, it's impractical. The physics don't work. 🟠 The constraint isn't compute anymore. It's memory. And that creates the infrastructure opportunity (for investors). 💡 This mirrors what we've seen before: incrementally more powerful devices meeting better compression algorithms. Remember when mobile video seemed impossible? The iPhone moment came when hardware capability met optimised delivery. Device Native AI is at that same inflection point. The shift is underway. It requires an orchestration layer. At DataSapien, we've been building exactly this, infrastructure that lets brands embed AI into existing mobile apps with zero cloud costs, 100% privacy by design, and up to 44X effectiveness improvements in customer engagement. Citrini's forecast validates what the data has been telling us: $107B in AI spend is primed to shift onto devices, and the infrastructure layer that enables this transition will capture significant value. 🤓 Worth reading their full analysis (5 min read), link in the comments below 👇 #DeviceNativeAI #EdgeAI #AIInfrastructure #InvestmentThesis #InferenceOnDevice

28

15 Comments

OpenAI has published a prompting guide for GPT-5: https://lnkd.in/e9YH_hWn “ We discuss concepts like improving agentic task performance, ensuring instruction adherence, making use of newly API features, and optimizing coding for frontend and software engineering tasks” #LLMs #GenAI

12

1 Comment

KT Genie Music launches the first conversational AI music platform in South Korea, powered by Microsoft Azure OpenAI, with a patent pending. #MusicTech #TechNews #AI #TechNews Link: https://lnkd.in/gHwQcbav 

5