OpenAI Tops 900 Million Weekly Users, With 50 Million Consumer Subscribers

OpenAI’s audience has grown beyond the 800-million-user milestone, but the central business tension remains: the consumer subscriber base is much smaller than ChatGPT’s weekly reach. OpenAI’s February 27 scale disclosure puts ChatGPT above 900 million weekly active users, consumer subscribers above 50 million and paying business users above 9 million; it also details $110 billion in new investment.

A simple comparison of the first two figures produces roughly one consumer subscriber per 18 weekly users, or about 5.6%, but that is a scale comparison rather than a measured conversion rate. Meanwhile, Quartz’s July 3 account of audited documents and investor communications lists $13.07 billion in 2025 revenue, $34 billion in costs and expenses, a $20.92 billion operating loss and approximately $600 billion pledged for AI infrastructure through 2030.

The subscriber gap is real, but “5% pays” is too broad

The available figures support a narrow conclusion: consumer subscriptions amount to only a small share of ChatGPT’s weekly audience when the two headline totals are compared. They do not show that every other user contributes no economic value, nor do they establish a formal subscription conversion rate calculated from a matched group of accounts.

The numerator and denominator measure different things. Weekly active users count activity during a particular week, while consumer subscribers are accounts with an ongoing paid plan; the public figures do not reveal how much the groups overlap in the same measurement window.

OpenAI also identifies a separate population of paying business users. Companies can additionally purchase API access or enterprise services, so adding business users to consumer subscribers and dividing the result by weekly ChatGPT users would mix categories that may overlap and do not represent all commercial spending.

“Only 5% foot the bill” therefore captures the broad imbalance between mass free distribution and a smaller consumer subscription base, but it overstates what the disclosed data can prove. A more defensible interpretation is that OpenAI finances a vast consumer funnel through several groups: subscribers, employers, developers, commercial partners and investors.

Rapid revenue growth has not eliminated operating losses

The financial documents described by Quartz indicate that revenue more than tripled from $3.7 billion in 2024, yet the company still recorded a substantial operating deficit in 2025. The operating loss is the more useful figure for assessing the gap between revenue and ongoing expenses than the larger net-loss headline, which included a major accounting charge associated with OpenAI’s corporate restructuring.

That deficit does not prove that each free ChatGPT user is individually unprofitable. Public financial totals do not disclose the serving cost, retention, acquisition value or eventual paid conversion of a particular user cohort, and free access can support distribution for workplace and developer products.

The figures do show that audience growth alone has not made current operations self-funding. Closing the gap could involve more paid accounts, stronger enterprise and API revenue, lower inference costs or new revenue streams; continued external capital can finance expansion, but it is not evidence that the underlying business has reached profitability.

The trillion-dollar claim needs a defined scope

The newer evidence does not establish a single, present $1 trillion liability sitting entirely on OpenAI’s balance sheet. The approximately $600 billion figure covers pledged infrastructure spending through 2030, while broader trillion-dollar estimates may combine cloud contracts, chips, data-center development and partners’ capital expenditure across different periods.

Capacity has nevertheless expanded beyond the original Stargate target. OpenAI’s April 29 infrastructure update says the company had surpassed its January 2025 goal of securing 10 gigawatts of US AI infrastructure by 2029, after adding more than 3 gigawatts during the preceding 90 days.

“Secured” should not be treated as synonymous with fully constructed, powered and operating. Large data-center projects require land, electricity, transmission, permits, equipment and construction, while OpenAI’s partner-led model distributes financing and execution across utilities, cloud providers, chipmakers, investors and contractors.

This distinction changes the risk analysis. OpenAI is exposed to the cost and availability of computing capacity, but the infrastructure program is not one conventional purchase with one delivery date or one funding source; it is a network of long-term commitments whose timing and accounting treatment can differ.

What the updated figures mean for creators

For creators, the numbers help explain why broad free access can coexist with increasingly important paid plans, higher usage limits and workplace products. A large free audience gives ChatGPT distribution, while recurring consumer and business payments provide more direct evidence that users value writing, research, coding and production workflows enough to pay for them.

The weekly-user total should not be read as a count of equivalent customers. It can include occasional personal use, education and lightweight queries, while one paid business deployment may support repeated work across a team and generate a different level of revenue.

Nor does the subscriber total reveal the economics of a particular creative workflow. The public disclosures do not provide plan-level retention, average revenue, inference cost or profitability, so they cannot show whether image generation, research, writing or another activity subsidizes the rest.

The clearest update is therefore less dramatic than a claim that 95% of users contribute nothing. ChatGPT’s reach and paid consumer base have both increased, yet the wide gap between them persists; at the same time, the documented operating loss and expanding compute program show why OpenAI still depends on enterprise demand, cost reductions and outside capital as well as consumer subscriptions.

Also read:

• MrBeast Just Hit 500 Million Subscribers — And He’s Still the Only One
• Voice-Only Leaves Users Stuck: 7 Rules for Better Voice Interfaces
• Sophie Rain Says She Rejected $15 Million—But the Athlete Is Still Unknown