📰 Key Takeaways

OpenAI has temporarily closed new sign-ups for its $200/month Pro plan after demand for its new flagship model Astra surged, citing strained system infrastructure. The news was announced on X by Thibault (Tibo) Sottiaux, OpenAI’s product lead for Codex, ChatGPT, and other core products, who said the Pro plan puts the most strain on the system, which is why new sign-ups at that tier are paused. Sottiaux emphasized this is “the smallest possible step we can take while still providing the broadest possible access,” noting that the API and the cheaper Go and Plus plans aren’t affected and remain open for subscriptions. OpenAI had already flagged this possibility on Wednesday, when Sottiaux said demand for Astra was “unlike anything we’ve seen,” and that the company was pulling every lever to keep up with the load — but the priority remains maintaining service quality for existing users, and if the pressure continues, new Pro sign-ups might need to stay paused for a while. OpenAI hasn’t said how long the pause will last, nor has it shared daily sign-up numbers to quantify the scale of demand — but it did just raise usage caps for Codex users last month, suggesting this surge is a fairly recent development. Astra launched on September 3rd and has been rolling out to Pro, Plus, Enterprise, and Business accounts, billed as a major leap in some of the most competitive areas in AI — reasoning, coding, and computer use. OpenAI has even called it the start of “the AGI era,” which has only fueled demand further.

💬 JudyAI Lab Take

OpenAI pausing new sign-ups for its $200/month Pro plan — because demand for its new flagship model Astra is “unlike anything we’ve seen” and system infrastructure is strained — is worth paying attention to, because it shows that even top-tier AI companies hit hardware ceilings during hypergrowth.

For AI builders, the lesson here is that there’s always a gap between demand growth and how fast you can scale infrastructure. OpenAI’s answer was tiered throttling: keep the API and the cheaper Go and Plus plans running normally, and only pause new sign-ups for the highest-load Pro tier, while prioritizing the experience of existing users. That “protect the installed base first, throttle new growth second” tradeoff is something any AI product could face during a traffic spike — it’s not something you can just solve by spinning up more servers. The fact that Codex usage caps were just raised last month shows how recently this pressure emerged, which says a lot about how hard it is to fully predict a demand curve when you’re doing capacity planning.

If your product also depends on an AI inference service, now’s a good time to take stock of how dependent you are on a single vendor or a single plan tier, and think about whether you need a fallback or a downgrade path.

📅 Source Info

🔗 Further Reading