OPENAI STATISTICS 2026

ChatGPT growth milestones

• 5 days — 1 million users (fastest ever; Netflix took 3.5 years, Instagram 2.5 months)
• 2 months — Reached 100 million MAU in January 2023
• 6 months — 1 billion website visits per month (SimilarWeb, May 2023)
• 18 months — 200 million weekly users (OpenAI, August 2024)
• 27 months — 300 million weekly users (OpenAI, December 2024)
• 39 months — 400 million+ weekly users (OpenAI, February 2026)

Usage by region

ChatGPT is a global phenomenon, but adoption varies significantly by region. The United States accounts for approximately 22% of all ChatGPT traffic, followed by India (12%), Brazil (7%), Japan (5%), and the United Kingdom (4%). Europe as a whole represents approximately 28% of traffic, with Germany, France, and the Netherlands as frontrunners.

The average session duration also varies by region: users in Japan spend an average of 11.4 minutes per session (the highest), while the global average is 8.2 minutes. Mobile usage dominates in emerging markets (68% in India, 72% in Brazil), while in Western Europe and the US, desktop still accounts for 55-60% of sessions.

ChatGPT subscription data

Of the 320 million monthly users, approximately 7% are paying subscribers. While this may seem low, it generates the lion's share of OpenAI's revenue. ChatGPT Plus ($20/month) is the most popular paid subscription with an estimated 18 million subscribers. ChatGPT Team has approximately 2.5 million seats, and ChatGPT Enterprise has approximately 1.5 million seats across more than 150,000 companies.

The churn rate for ChatGPT Plus is approximately 4.5% per month — relatively low for a SaaS product in this price range. Enterprise subscriptions have a churn rate of less than 1.5% per month, indicating strong product-market fit in the business segment.

Key facts about OpenAI's financing

• Valuation 10x'd in 18 months — from $29B (Jan 2023) to $300B (Mar 2025)
• $40B SoftBank round was the largest private funding round ever in the tech sector
• 24x forward revenue multiple — $300B valuation on $12.7B ARR, comparable to Nvidia in 2023
• Transition to for-profit announced in 2025, after years as a "capped-profit" structure under non-profit governance
• Compute costs estimated at $7-9 billion per year (2025/2026), primarily GPU clusters on Microsoft Azure
• Break-even expected by end of 2027, according to internal projections leaked to The Information
• Secondary market: OpenAI shares trade on the secondary market at valuations of $280-320B (Forge Global, CartaX)
• Total investors: more than 40 institutional investors including Microsoft, SoftBank, Thrive Capital, Tiger Global, a16z, and Sequoia

OpenAI's financial challenge

Despite impressive revenue growth, OpenAI still loses significant money. The estimated net loss of $4.5 billion in 2025 is primarily caused by three factors: compute infrastructure ($7-9B/year in GPU costs via Microsoft Azure), salaries and equity compensation ($2.5B+/year), and research & development. OpenAI spends an estimated 40% of revenue on GPU compute alone. By comparison, Google spends approximately 15% of revenue on infrastructure.

The cost structure also explains why OpenAI aggressively seeks investors. Without the $40 billion SoftBank round, the company would have run through its reserves by 2026. The expectation is that economies of scale and cheaper hardware (Nvidia Blackwell, custom chips) will significantly improve margins by 2027-2028.

API ecosystem in numbers

• Tokens processed: estimated 4.5 trillion tokens per day via the API, or ~3.4 billion pages of text (OpenAI, Bloomberg estimate)
• Average latency: GPT-4o responds with 320ms time-to-first-token, 5x faster than original GPT-4 (1,600ms)
• Uptime SLA: 99.9% for enterprise tier, 99.5% for standard tier (OpenAI API Status, trailing 90 days)
• Regions: API available in 161 countries; not available in China, Russia, Iran, and North Korea
• Fine-tuning: 180,000+ custom fine-tuned models running on the OpenAI platform (OpenAI DevDay 2025)
• Assistants API: more than 500,000 AI assistants built by developers for specific workflows and tasks
• Batch API: 50% discount on tokens for non-time-critical tasks, processing guaranteed within 24 hours
• Responses API: launched March 2025, replaces the Assistants API with better performance and lower costs

The shift toward cheaper models

A notable trend in the OpenAI ecosystem is the shift toward cheaper, faster models. GPT-4o mini — launched in July 2024 — now processes more than 60% of all API calls, despite only being available since mid-2024. The reason is simple: the model is 200x cheaper than the original GPT-4, while delivering comparable results for most tasks.

This shift has paradoxically led to higher total API revenue. Lower prices per token mean developers find more use cases economically viable, causing total token volume to grow exponentially. OpenAI internally calls this the "token inflation" strategy — comparable to how cloud computing became cheaper and therefore saw more usage.

For the average developer, an API integration with GPT-4o mini costs less than $5 per month for typical usage (10,000 requests). Even complex enterprise integrations with o3 stay under $500/month for most businesses. This makes the barrier for AI adoption historically low.

Enterprise key facts

• ChatGPT Enterprise pricing: estimated $50-60 per user per month, with unlimited GPT-4o/o3 access and admin controls
• ChatGPT Team: $25 per user per month (minimum 2 seats), most popular tier for SMBs
• SOC 2 Type II certified — enterprise-grade security and compliance
• Data is not trained on enterprise conversations, contractually established with DPA
• Microsoft 365 Copilot runs on GPT-4o — 1.3 million paying enterprise customers (Microsoft, Q4 2025)
• ROI: enterprise customers report an average 4.2x return on investment within 12 months (Forrester TEI Study, 2025)
• Implementation time: average 6-8 weeks from pilot to production for enterprise customers (Forrester)
• Custom models: enterprise customers can train fine-tuned models on their own data, with full data isolation

Enterprise use cases by department

OpenAI adoption in enterprise environments varies significantly by department. Marketing and communications are the most frequent users (78% adoption), followed by customer service (72%), product development (65%), and HR (54%). The lowest adoption is found in finance and legal (38%), primarily due to compliance considerations and high accuracy requirements.

Major enterprises such as Morgan Stanley, Stripe, Klarna, and Heineken are publicly known OpenAI customers. Morgan Stanley uses ChatGPT Enterprise to search 100,000+ research documents, Klarna replaced 700 customer service agents with a ChatGPT integration handling 2.3 million conversations per month, and Stripe uses the API for fraud detection and documentation management.

Organization and leadership

• Sam Altman — CEO since 2019. Briefly fired and reinstated in November 2023 ("boardroom coup")
• Mira Murati — Former CTO, departed September 2024. One of several high-profile departures
• 10+ senior departures in 2024-2025, including Chief Scientist Ilya Sutskever, co-founders John Schulman and Andrej Karpathy
• New board formed in 2024 with Bret Taylor (chair), Larry Summers, and Sue Desmond-Hellmann
• Headquarters: San Francisco, California. Secondary offices in London, Dublin, Tokyo, and Singapore
• 65% of employees in research & engineering, 15% in go-to-market, 20% in operations and support