OpenAI has suspended new sign-ups and upgrades for its most expensive subscription tier, the ChatGPT Pro 20X (priced at $200 per month), citing an unprecedented surge in demand for the GPT-6 Astra model. While existing subscribers remain unaffected, the company is pausing new entries to this specific plan while it works on expanding its computing capacity.
Thibault Sottiaux, Head of Core Products and Platform at OpenAI, explained that the exceptional performance of Astra has triggered a level of demand that current infrastructure cannot sustain without compromising user experience. This follows the broader rollout of GPT-6 Astra to Plus, Business, and Enterprise users, but the Pro 20X tier has become a particular bottleneck.
The Economics of Over-Utilization
The decision to freeze the $200 plan specifically, rather than lower tiers, stems from a combination of high utilization and narrow margins. The Pro 20X plan offers the most aggressive discount on computing power: users pay double compared to the $100 Pro 5X plan but receive four times the quota (200 messages per week for Astra).
Analysis from SemiAnalysis suggests that this tier is the most financially precarious for OpenAI. Due to the high cost of Astra's inference—which consumes significantly more quota than GPT-5.6 Sol—the Pro 20X plan enters a negative gross margin once the utilization rate hits just 5.7%. This is significantly lower than the 11.4% threshold for Plus and Pro 5X plans, meaning heavy users (primarily developers running long-chain Agent tasks) are driving the system into deficit.
Infrastructure Strain and Strategic Priority
CEO Sam Altman has acknowledged that the company is currently losing money on Pro subscriptions because usage levels have far exceeded expectations. This infrastructure strain comes at a critical time, as OpenAI continues to push the boundaries of AI agents and coding capabilities.
Sottiaux had previously warned on X that the demand for Astra was "unprecedented," suggesting that every available lever was being pulled to maintain stability. The current pause is a tactical move to prioritize service quality for existing customers until capacity stabilizes, reflecting the ongoing struggle to balance cutting-edge model performance with the physical limits of GPU availability.

No comments yet. Be the first!