ElevenLabs CEO says AI audio models will be ‘commoditized’ over time

Mati Staniszewski, the CEO and co-founder of ElevenLabs, has made waves in the AI audio sector by predicting that AI models will eventually become commoditized. Speaking at the TechCrunch Disrupt 2025 conference, he shared insights into his company's vision for the evolving landscape of AI audio technology. Staniszewski emphasized that while ElevenLabs is currently focused on overcoming architectural challenges in model development, this emphasis will be crucial for the next couple of years. He asserted, "Over the long term, it will commoditize — over the next couple of years." Although he acknowledged that variations might persist across different voices and languages, he believes the overall distinctions will diminish. When questioned about the rationale behind building models amidst the expectation of commoditization, Staniszewski highlighted the immediate advantages that proprietary models offer. He pointed out that if AI-generated voices lack quality, it remains a significant issue that must be addressed. "The only way to solve it is… building the models yourself," he stated, adding that competitors will eventually rise to meet these challenges as well. Looking ahead, he anticipates a shift toward multi-modal or fused approaches, where audio and video are generated simultaneously, along with integration with large language models (LLMs) in conversational settings. He referenced Google’s Veo 3 as an example of the potential that such combinations can unlock. ElevenLabs is also planning to forge partnerships and explore open-source technologies to merge its audio capabilities with other models' expertise. Staniszewski believes that focusing on both model development and practical applications will be key to delivering long-term value. He concluded by drawing a parallel to Apple’s success with software and hardware, stating, "We think the product and AI will be the magic for the generation of the best use cases."