Troveo’s Platform Expansion: What’s New?

On April 28, 2026, Troveo announced the expansion of its AI training data platform into five new categories, broadening its offerings beyond video to include:

• Audio: 4 million hours of single- and multi-channel audio across dozens of languages and dialects, tailored for voice-based models like automatic speech recognition and conversational AI.
• Text: Billions of words sourced from publishers and rights holders, structured for training, fine-tuning, and evaluation.
• Enterprise Workflows: Proprietary data from business systems, enabling AI models to understand and automate complex organizational processes.
• Gaming: Datasets designed to improve AI performance in virtual environments, including esports and interactive media.
• Robotics: Task demonstration clips and sensor data to train AI for physical automation and real-world applications.

This expansion builds on Troveo’s existing video library, which now exceeds 8 million hours of licensed footage. The company’s datasets are annotated and training-ready, eliminating the require for AI labs to preprocess raw data—a significant time and cost savings.

Why Licensed Data Matters

The AI industry has long relied on publicly available data scraped from the internet, often without permission. This practice has led to lawsuits, reputational damage, and models trained on low-quality or biased data. Troveo’s model offers a legal alternative: datasets sourced directly from rights holders with clear licensing agreements. As the company’s CEO, Marty Pesis, stated in 2025, “If you have public content out there, your content is already being used to train models. The question is, do you seek to opt in and extract that value?”

Troveo’s approach addresses two key pain points for AI developers:

1. Speed: Training-ready datasets can be deployed immediately, helping labs meet aggressive deadlines.
2. Quality: Licensed data is often higher quality than public scrapes, leading to better model performance. For example, one Troveo client sought 50,000 hours of dog videos after their AI-generated dogs kept resembling cats—a problem rooted in insufficient training data.

$20 Million in Payouts: A New Revenue Stream for Creators

Troveo’s expansion coincides with a milestone: over $20 million paid to content owners since its launch. This figure underscores the growing demand for licensed data and the platform’s role as a bridge between creators and AI labs. Content owners—including filmmakers, YouTubers, talent agencies, and media companies—grant Troveo exclusive rights to use their footage for AI training in exchange for compensation.

How Payouts Work

Troveo’s payout structure is based on several factors:

• Content Type: Video footage typically sells for $0.75 to $3 per minute, with higher rates for visually diverse or niche content.
• Demand: Specialized datasets (e.g., robotics demonstrations or multilingual audio) command premium prices.
• Exclusivity: Content owners retain rights but grant Troveo exclusive AI training licenses, ensuring their data isn’t resold or reused without permission.

One notable participant is Peter Hollens, a YouTuber and a cappella singer with 3 million subscribers. Hollens shared in 2025 that for every minute of content he uploads, he scraps 50 minutes of unused footage—material that now generates passive income through Troveo.

Featured Data Providers

Troveo’s network includes over 7,000 data providers across 150 countries, spanning 60+ languages. Key partners include:

• Barstool Sports (Chicago, USA): A leading sports and pop culture media brand contributing video, audio, and text datasets.
• POPS Worldwide (Ho Chi Minh City, Vietnam): Southeast Asia’s largest digital entertainment company, with a library serving 700 million+ fans.
• 100 Thieves (Los Angeles, USA): An esports and lifestyle brand providing gaming and video data.
• Savage Ventures (Nashville, USA): A digital media company behind viral content brands, including Vice.

Why AI Labs Are Turning to Troveo

The AI training data market is projected to grow exponentially as models become more sophisticated. However, three challenges have hindered progress:

1. Legal Risks: Scraping public data without permission exposes AI companies to copyright lawsuits. Troveo’s licensed model mitigates this risk.
2. Data Scarcity: Public datasets are often noisy, biased, or outdated. Troveo’s proprietary data—sourced from broadcast archives, studio vaults, and private collections—fills this gap.
3. Quality Control: Raw data requires extensive preprocessing. Troveo’s annotated datasets are ready for immediate use, reducing development time.

As Pesis noted, “Access to training data is a key bottleneck for developing next-generation frontier models.” Troveo’s ability to deliver high-quality, rights-cleared data at scale positions it as a critical partner for AI labs aiming to build more accurate, ethical, and commercially viable models.

Key Takeaways

• Troveo has expanded its AI training data platform to five new categories: audio, text, enterprise workflows, gaming, and robotics.
• The company has paid over $20 million to content owners, reflecting strong demand for licensed data.
• Troveo’s library now includes 8 million+ hours of video and 4 million+ hours of audio, sourced from 7,000+ providers worldwide.
• Content owners earn passive income by licensing unused footage, with payouts ranging from $0.75 to $3 per minute.
• AI labs benefit from training-ready datasets that reduce legal risks, improve model quality, and accelerate development timelines.

FAQ

How does Troveo ensure data quality?

Troveo’s datasets are annotated and structured for AI training, eliminating the need for preprocessing. The company works directly with rights holders to source high-quality, diverse content.

Who can contribute data to Troveo?

Content owners—including filmmakers, YouTubers, media companies, and enterprises—can license their unused footage or proprietary data to Troveo. The platform currently has over 7,000 approved providers.

What are the benefits for AI labs?

AI labs gain access to licensed, high-quality data that reduces legal risks, improves model performance, and accelerates development. Troveo’s datasets are training-ready, saving labs time and resources.

How are payouts calculated?

Payouts depend on content type, demand, and exclusivity. Video footage typically earns $0.75 to $3 per minute, with premium rates for specialized datasets.

The Future of AI Training Data

Troveo’s expansion and $20 million payout milestone reflect a broader industry shift toward ethical, licensed data sourcing. As AI models grow more complex, the demand for high-quality training data will only increase. Platforms like Troveo are poised to play a pivotal role in shaping the future of AI development—one where creators are fairly compensated, and labs have access to the data they need to build the next generation of intelligent systems.

For content owners, Troveo offers a way to monetize unused assets while maintaining control over how their data is used. For AI labs, it provides a scalable solution to the industry’s most pressing challenge: access to reliable, rights-cleared training data. As the AI landscape evolves, expect to see more companies adopt Troveo’s model—or risk falling behind.