Analysis
OpenAI temporarily paused new sign-ups and upgrades to its $200-a-month ChatGPT Pro 20X plan starting September 10, TechCrunch reported, after demand for its new Astra model strained the company's computing infrastructure. Existing Pro subscribers keep their access, and OpenAI continues selling its $100 Pro tier, Plus, Go subscriptions, API access and enterprise accounts.
OpenAI product executive Thibault Sottiaux called demand for Astra 'unprecedented,' saying the $200 tier 'put the most strain on our systems' and that the company wanted 'to take the smallest step that allows us to continue giving the broadest access possible.' Astra's resource-intensive features, including computer use, are driving the added infrastructure load specifically on that top-priced tier.
The $200-a-month price point has functioned as a live test of how much a premium slice of consumers will pay for frontier-model access ahead of the rest of the market -- a segment Anthropic and Google have each begun testing with their own top-tier consumer plans, though neither currently prices as high as OpenAI's Pro 20X tier. Pausing that tier rather than the cheaper Plus or Go plans suggests the heaviest, most compute-intensive usage is concentrated among OpenAI's most willing-to-pay customers, not its broad base.
This is the same compute-scarcity dynamic running through nearly every story in this issue: elsewhere, Nvidia is reportedly considering anchoring Anthropic's IPO in part on the strength of Anthropic's own compute-constrained growth, and Pulse has separately tracked Microsoft's admission that it's turning away cloud and AI customers despite $175 billion in annual capex. Anthropic and Google have each managed their own capacity constraints differently this year, rationing enterprise contracts and API rate limits rather than pausing a named consumer subscription tier outright -- making OpenAI's move a more visible, consumer-facing admission of the same industry-wide shortage.
OpenAI hasn't announced when the $200 tier will reopen. The real signal will be whether that gap closes within weeks, suggesting a temporary demand spike, or stretches into months, which would be evidence that Astra's compute requirements permanently outrun OpenAI's current infrastructure rather than representing a short-term bottleneck.
