OpenAI relies on Pearl for live streaming

The Challenge: Ensuring 100% reliability for high-profile live events

OpenAI, the organization behind the groundbreaking AI chatbot ChatGPT, set out to start livestreaming demos and new feature announcements to YouTube. The team kicked off this initiative with the “12 Days of OpenAI” event – a high-profile series designed to showcase new developments. With millions of viewers expected to tune in live, flawless, real-time streaming was critical.

Streaming to its YouTube channel for the first time, OpenAI aimed to engage a large, tech-savvy audience. With a substantial following and high expectations, the stakes were high — and there was no room for error.

The challenge wasn’t just the complexity of live streaming, but achieving 100% reliability under pressure. As Garrett McGrath from OpenAI noted, “It’s really easy to do a big post-produced video and make everything look perfect, but that’s not very authentic.” OpenAI wanted to preserve the vitality of live video while ensuring the technical setup could handle the scale of their audience and the demands of real-time broadcasting.

Solution: Epiphan Pearl as primary encoder and streamlined workflow

To ensure smooth, high-quality streaming throughout the 12 Days of OpenAI, OpenAI chose Epiphan Pearl-2 as their primary encoder. OpenAI relied on Pearl for three main functions: primary recording, primary streaming to YouTube, and confidence monitoring. The team used Pearl’s 2 HDMI out ports to set up stream confidence monitoring as well as a backup recording.

This level of redundancy provided the OpenAI team with the assurance they needed, confident that all content would be reliably recorded and securely stored — a vital factor in maintaining the reliability required for such a high-profile event.

Results: Consistent, high-quality streams with zero downtime

By integrating Epiphan Pearl into their production setup, OpenAI successfully met the high demands of the 12 Days of OpenAI event. Despite the pressure of broadcasting live to millions, Pearl’s intuitive interface and reliable encoding capabilities enabled smooth, uninterrupted streaming throughout.

Viewer feedback was positive, with many noting the smooth delivery and authentic feel of the live experience. With Pearl handling the technical heavy lifting, the OpenAI team was free to focus on delivering meaningful content. Since then, OpenAI has continued to rely on Pearl as their go-to solution for all live streams—ensuring the same trusted performance across every broadcast.