OpenAI’s newest model is so popular that the company hit the brakes on new $200-a-month ChatGPT Pro signups. TechCrunch’s Sarah Perez reports that product lead Thibault Sottiaux said Pro puts the most strain on OpenAI’s systems, so the company disabled new subscriptions for that tier to protect service for people already on it.
Astra launched on September 3 and has been rolling out across Pro, Plus, Enterprise, Business, and the API. OpenAI pitched it as a leap in reasoning, coding, and computer use, and even called it the start of an AGI era. That marketing worked. Demand spiked hard enough that Sottiaux warned on Wednesday that a Pro pause might be coming if the load kept climbing.
Existing Pro users keep access. The API stays open. Lower-cost Go and Plus plans remain available. OpenAI has not said how long the pause will last or how many people try to join Pro each day. Last month the company even raised Codex usage limits, which makes this crunch feel recent and tied to Astra’s rollout rather than a long-running shortage story.
If you were about to drop $200 a month for the top ChatGPT tier, you are stuck waiting. If you already pay for Pro, this is OpenAI choosing current customers over growth. That is rare enough in consumer tech that it is worth noticing. Capacity is the quiet bottleneck behind every unlimited AI pitch.
I care about this as a developer because the same model wave that makes coding agents feel magical also melts GPUs. When inference demand outruns data centers, the first thing that breaks is the premium plan that burns the most tokens. The positive read is that OpenAI is adding capacity as fast as it can and taking the smallest cut it can invent, a pause on new Pro rather than throttling everyone.
If Plus or the API covers your day-to-day work, you may not feel this pause at all. If your team standardized on Pro for heavy Astra sessions, plan for a waitlist mindset and keep a backup model ready for peak hours. For now, shop the plans that are still open, watch for OpenAI’s reopen announcement, and treat Astra hype as a reminder that compute is finite. The tools are getting better. The queue to get in just got longer.
