{"slug": "openai-pauses-200-chatgpt-pro-plan-as-astra-strains-compute-capacity", "title": "OpenAI Pauses $200 ChatGPT Pro Plan as Astra Strains Compute Capacity", "summary": "OpenAI has paused new subscriptions and upgrades to its $200-a-month ChatGPT Pro 20X plan after demand for its Astra model strained computing capacity, OpenAI product executive Thibault Sottiaux said. Existing Pro 20X customers keep access, and OpenAI continues selling its $100 Pro plan, Plus and Go subscriptions, API access and enterprise accounts, with no announced date for reopening the $200 tier. Astra, launched Sept. 3, consumes user allowances faster than GPT-5.6 Sol and includes a computer use feature, while OpenAI's available compute grew from 0.2 gigawatts in 2023 to about 1.9 GW in 2025.", "body_md": "TL;DR — Key Takeaways\n\n- OpenAI has paused new subscriptions and upgrades to its $200-a-month ChatGPT Pro 20X plan after demand for Astra strained available computing capacity.\n- Existing Pro 20X customers keep access, while OpenAI continues offering its other subscription tiers, API services and enterprise products.\n- Astra’s resource-intensive features, including computer use, are contributing to higher infrastructure demands.\n\nOpenAI has temporarily stopped new subscriptions and upgrades to its $200-a-month ChatGPT Pro 20X plan after heavy demand for its new Astra model put pressure on the company’s computing infrastructure.\n\nExisting Pro 20X customers retain access, while OpenAI continues selling other offerings, including its $100 Pro plan, Plus and Go subscriptions, API access and enterprise accounts. OpenAI has not announced when it will reopen the $200 tier.\n\nThe move highlights a core challenge facing AI model developers: even vast investments in computing infrastructure may not keep up with demand for a popular new model. For enterprise users, the pause also highlights that companies deploying frontier models may need to consider compute availability alongside model performance when planning deployments.\n\nAstra launched Sept. 3, with OpenAI initially limiting access to a small group of enterprise customers while it expanded capacity. The model subsequently became available to paying users, generating demand that OpenAI product executive Thibault Sottiaux described as unprecedented.\n\nOpenAI chose the $200 subscription for the cutoff because Pro 20X customers receive 20 times the usage available on the Plus tier, making them particularly compute intensive. Sottiaux said the company selected the option that would cause the least disruption while preserving access for current customers.\n\nAstra’s advanced capabilities contribute to its heavy resource demands. Its computer use feature can run desktop interfaces, navigate websites and complete forms. OpenAI also says Astra consumes user allowances more quickly than GPT-5.6 Sol, its previous flagship model.\n\n### **More Capacity Planned**\n\nOpenAI has substantially expanded its infrastructure. The company says its available compute grew from 0.2 gigawatts in 2023 to about 1.9 GW in 2025, or 9.5 times its 2023 capacity. Yet Astra’s launch shows how quickly new workloads can absorb added resources.\n\nMore capacity is planned, but the timing is not clear. OpenAI has agreements covering at least 10 GW of NVIDIA systems, 6 GW of AMD GPU infrastructure and 10 GW of custom AI accelerators through Broadcom. The company has also broadened its infrastructure relationships beyond Microsoft Azure to include CoreWeave and the Stargate project, a planned $500 billion, four-year AI infrastructure buildout.\n\nOpenAI previously suspended new ChatGPT Plus subscriptions in November 2023 when usage following its first developer conference exceeded available capacity. The current limit is occurring despite the massive infrastructure expansion since then.\n\nEuropean enterprises face an additional challenge. Astra can be accessed through Microsoft Foundry in European regions, including France Central and Germany West Central, but those are global deployments, so processing is not guaranteed to remain within the EU. Astra’s six data-zone deployments are all in the U.S., leaving European customers without a paid option that guarantees processing stays inside the EU data boundary.", "url": "https://wpnews.pro/news/openai-pauses-200-chatgpt-pro-plan-as-astra-strains-compute-capacity", "canonical_source": "https://techstrong.ai/articles/openai-pauses-200-chatgpt-pro-plan-as-astra-strains-compute-capacity/", "published_at": "2026-09-11 23:09:56+00:00", "updated_at": "2026-09-11 23:54:40.790507+00:00", "lang": "en", "topics": ["ai-products", "large-language-models", "ai-infrastructure", "ai-chips", "ai-agents"], "entities": ["OpenAI", "ChatGPT Pro 20X", "Astra", "Thibault Sottiaux", "GPT-5.6 Sol", "NVIDIA", "AMD", "Broadcom"], "alternates": {"html": "https://wpnews.pro/news/openai-pauses-200-chatgpt-pro-plan-as-astra-strains-compute-capacity", "markdown": "https://wpnews.pro/news/openai-pauses-200-chatgpt-pro-plan-as-astra-strains-compute-capacity.md", "text": "https://wpnews.pro/news/openai-pauses-200-chatgpt-pro-plan-as-astra-strains-compute-capacity.txt", "jsonld": "https://wpnews.pro/news/openai-pauses-200-chatgpt-pro-plan-as-astra-strains-compute-capacity.jsonld"}}