OpenAI pauses $200 ChatGPT Pro subscriptions as Astra demand overwhelms servers

OpenAI has temporarily stopped taking new sign-ups for its premium $200-per-month ChatGPT Pro subscription, admitting that demand for its newly launched Astra model is overwhelming its servers.

The suspension, announced on Thursday, affects both brand-new sign-ups and existing users trying to upgrade to the Pro plan. OpenAI's other subscription tiers — including the lower-priced Go and Plus plans, as well as developer APIs — remain open. Existing Pro users are also unaffected by the pause.

Thibault Sottiaux, OpenAI's product lead, said the top-tier plan places the heaviest burden on the company's hardware. Sottiaux had warned a day earlier that a freeze might be imminent, describing the rush for Astra as "unprecedented" and unlike anything he had seen during previous periods of rapid growth. He added that the company is pulling every available lever to sustain the demand while prioritising service quality for existing users.

Why Astra strains the servers

Astra was launched on 3 September. Billed by OpenAI president Greg Brockman as the start of the "AGI era", the model is designed for highly complex tasks like advanced reasoning and coding. Its standout feature, known as "computer use", allows the AI to control a desktop much like a human does — filling out forms and browsing web pages at high speeds.

This level of automation, alongside the fact that the $200 Pro tier provides 20 times the usage allowance of the lower $20 Plus tier, means users are consuming vast amounts of processing power. Astra burns through these allowances much faster than previous flagship models, such as GPT 5.6 Sol, putting immense pressure on OpenAI's physical infrastructure.

A repeating bottleneck

This is not the first time OpenAI has had to turn away paying customers due to infrastructure bottlenecks. In November 2023, chief executive Sam Altman suspended sign-ups for the ChatGPT Plus tier after a surge in interest following the company's first developer conference.

Since that initial crunch, the company has aggressively expanded its hardware footprint. OpenAI said its available compute capacity grew from 0.2 gigawatts in 2023 to around 1.9 gigawatts in 2025.

To sustain this growth, OpenAI has expanded its infrastructure footprint beyond its primary partner, Microsoft Azure. The company has partnered with cloud provider CoreWeave and is embarking on Stargate, a $500 billion, four-year infrastructure buildout.

Further capacity is expected to arrive in the second half of 2026. This includes plans for at least 10 gigawatts of Nvidia systems, a 6-gigawatt agreement with AMD, and a partnership with Broadcom for custom AI accelerators. For now, however, those future pipelines cannot resolve the immediate hardware bottleneck, and the company has not said when it expects to reopen Pro sign-ups.