Goldman Sachs29 min
Epidemic Sound’s Oscar Höglund on the future of the music industry
Oscar Höglund, Clif Marriott, Oskar Haugland
Conference Presentation, Fireside Chat, Interview
Epidemic Sound's Origin and Core Model
- Founded 15 years ago by Oskar Haugland to "soundtrack the internet" as it transitioned from text to video-centric platforms.
- Operates as a B2B SaaS subscription platform providing royalty-free music to content creators, with a catalog covering approximately 20% of all YouTube content.
- Reaches 2.5 billion people daily, serving hundreds of thousands of customers ranging from individual creators (e.g., MrBeast, PewDiePie) to major corporations (e.g., Netflix, Fortune 500 companies).
- Differentiates itself by acquiring 100% of IP rights for all tracks in its catalog, eliminating the "tragedy of the commons" caused by multiple stakeholders with veto rights in the traditional music industry.
Financial Structure and Artist Economics
- Employs a 50/50 revenue split between the platform and artists on royalty income generated from streaming platforms (Spotify, Apple, etc.), contrasting sharply with the traditional industry standard where artists receive only 10–15% after labels and publishers take 80–85%.
- Generates a gross margin of 92% by avoiding traditional licensing complexities and leveraging a unified rights ownership model.
- Created a "flywheel" effect where content creators drive massive distribution, generating data signals that lead to increased artist remuneration and further artist acquisition.
Artificial Intelligence and the Creator Economy
- Macro Impact: AI is expected to expand the Total Addressable Market (TAM) by lowering barriers to content creation and increasing efficacy across the entire creator economy.
- Strategic Advantage: Epidemic Sound leverages three key AI ingredients:
- Data: Ownership of the world's largest depository of unencumbered, high-quality music IP (words, stems, vocals) which legacy rights holders cannot fully utilize due to fragmented ownership.
- LLM Access: Utilization of accessible Large Language Models and skilled workforces for fine-tuning.
- Reinforcement Learning: Daily feedback loops from 2.5 billion interactions to train models on user preference (e.g., skips, engagement).
- Application Priority: The company prioritizes "recommendation" and "adaptation" over pure generative music to solve immediate creator pain points:
- Recommendation: AI analyzes billions of data points to suggest specific tracks that historically perform well with specific video contexts (e.g., specific audience demographics or content length).
- Adaptation: AI tools automatically adapt full-length tracks to specific video durations (e.g., 15-second TikTok clips) without requiring manual audio engineering.
Industry Trends and Future Outlook
- Content Renaissance: The last 3–5 years are characterized as a renaissance for storytelling, with innovation in both short-form and long-form content formats.
- Format Convergence: Short-form platforms are introducing long-form content, while long-form platforms are adopting short-form, and Western ad-driven models are merging with Eastern fan-commerce models.
- Democratization of Production: While distribution was democratized by smartphones and social platforms, AI is now expected to democratize production, significantly altering the ratio of viewers to creators.
- Challenges: The industry faces ongoing concerns regarding intellectual property guardrails, child safety, hate speech, and the potential for reduced attention spans due to content abundance.
- Market Sentiment: Haugland maintains a "hardcore optimistic" view, expecting the sector to solve these challenges while continuing to grow.