DevOps & Platform Eng

OpenAI's 900M User Surge: Ory Solves Identity at Scale

Forget login boxes. When OpenAI blew up, their identity layer became the bottleneck. Enter Ory.

{# Always render the hero — falls back to the theme OG image when article.image_url is empty (e.g. after the audit's repair_hero_images cleared a blocked Unsplash hot-link). Without this fallback, evergreens with cleared image_url render no hero at all → the JSON-LD ImageObject loses its visual counterpart and LCP attrs go missing. #}
Diagram showing OpenAI's user growth from 100M to 900M weekly active users, with Ory as the foundational identity layer.

Key Takeaways

  • OpenAI scaled to 900 million weekly users by implementing Ory for identity and access management.
  • The company avoided traditional IAM solutions, opting for a flexible, self-hosted Ory-powered system to handle 'success disaster' growth.
  • Ory provided enhanced control, scalability, and security, enabling OpenAI to focus on AI innovation rather than identity infrastructure maintenance.

Did you ever stop to think about the plumbing behind 900 million weekly logins? Probably not. You just want your AI fix. But for OpenAI, that sheer volume wasn’t just a success story; it was a potential catastrophe.

Scaling to 900 million weekly active users. That’s not a typo. It’s a number that dwarfs many nations. And when your product is a high-compute, stateful AI environment, that’s not just about usernames and passwords anymore. It’s the digital equivalent of managing a planet’s worth of people trying to get into the same highly sensitive building. System stability? Data privacy? Global access? All hang on that gatekeeper.

OpenAI didn’t just launch ChatGPT. They detonated a shift in computing. And the ensuing “success disaster” — a glorious, terrifying tidal wave of demand — required an identity layer that could stretch, not snap. Horizontally. Securely. Flexibly.

Here’s the thing: they didn’t go with the usual suspects. No monolithic ‘Identity-as-a-Service’ vendor was going to cut it. Instead, they opted for Ory. Why? Because they wanted a bespoke, hardened, infinitely scalable identity system. Not something off-the-shelf that would buckle under the strain.

The “Success Disaster” Identity Crisis

The launch of ChatGPT was, to put it mildly, a nuclear event in user acquisition. One million users in five days. That’s a speed run that makes traditional Identity and Access Management (IAM) solutions weep. They choke on rigid database schemas, buckle under global distribution demands, and hemorrhage resources with even minor latency spikes in token validation.

And let’s not forget control. OpenAI wanted to A/B test. They wanted to observe. They wanted to optimize. Not be locked into a provider’s roadmap. Plus, the need for self-hosting in a chosen environment was non-negotiable. A standards-based, agile IAM approach, then, wasn’t just a preference; it was a survival strategy. No vendor lock-in. Just a smooth, globally scalable user experience they could actually steer.

“OpenAI wanted a partner that could help enable our vision for owning our identity processes, data, and success. We have a lot of partners, and Ory is one of our best.” – Benjamin Billings, Engineering Manager, OpenAI

Building Infrastructure for the Next Billion Users (and Beyond)

Navigating the explosive growth from 200 million to over 400 million weekly active users in just months—what Benjamin Billings so aptly called a “success disaster”—was made possible by Ory. This wasn’t just about keeping the lights on. It was about resolving the classic “SaaS paradox”: needing enterprise-grade security without the shackles of a single, rigid provider.

OpenAI got to own its user interface. They got to dictate the roadmap. They could build a tailored experience for everyone, from casual users to enterprise clients. This partnership wasn’t just about identity; it was about accelerating engineering velocity. Instead of months wrestling with a homegrown identity stack, they use Ory’s expertise and support to actually build AI. Identity became the invisible, rock-solid foundation.

Massive Scalability Achieved: From 100 million to over 900 million weekly active users. Check.

Zero-Downtime Migration: The legacy system vanished without a ripple.

Enhanced Observability: Deep, in-house control over identity data. No more guesswork with third-party black boxes.

Engineering Velocity Boost: Faster feature rollouts thanks to Ory Hydra’s battle-tested engine.

OpenAI’s journey isn’t just a case study in scaling; it’s a masterclass in architectural foresight. They sidestepped the SaaS vendor lock-in and the DIY maintenance quagmire. The result? A lean operation. A secure one, built on globally trusted open-source. And lightning-fast delivery at the edge.

As the 1 billion user mark looms, the Ory-powered identity layer remains the unsung hero. The critical, invisible backbone of the most significant tech story of our era.

“Overall, working with your team has been very satisfying. You’ve been very responsive. You’ve done changes that we’ve asked for in a very timely manner. So it’s been overall great working with you all.”

– Benjamin Billings, Engineering Manager, OpenAI

Does this mean identity management is suddenly easy?

No. It means choosing the right tools and architecture for the specific demands of hyper-scale AI services is paramount. Ory offers a flexible, open-source approach that companies like OpenAI found necessary.

Will Ory replace traditional IAM providers entirely?

Unlikely. Traditional providers still serve many use cases. However, OpenAI’s success highlights the growing need for more flexible, self-hostable, and scalable identity solutions for cutting-edge applications.

How does Ory Hydra contribute to this scale?

Ory Hydra is an OAuth2 and OpenID Connect server. Its design prioritizes performance, extensibility, and security, making it suitable for handling complex authentication and authorization flows at massive scale, which is exactly what OpenAI needed.


🧬 Related Insights

Alex Rivera
Written by

Developer tools reporter covering SDKs, APIs, frameworks, and the everyday tools engineers depend on.

Worth sharing?

Get the best Developer Tools stories of the week in your inbox — no noise, no spam.

Originally reported by The NewStack

Stay in the loop

The week's most important stories from DevTools Feed, delivered once a week.