OpenAI announced on July 31 (local time) that the number of active users of its AI had surpassed 1 billion, and its enterprise client base had also exceeded 2 million. This comes three years and eight months after the November 30, 2022 launch of ChatGPT. Within five days of launch, it had 1 million sign-ups; within a year, it surpassed 100 million weekly active users (WAU); it then passed 200 million in August 2024, 500 million last August, and 800 million last October. However, it reached 900 million users in February this year, and it took another five months to surpass 1 billion. Compared to Facebook, which took six years to reach 1 billion users, this pace is more than two years faster.
The order in which to read the numbers
Rather than the speed of the record, the number that those considering adoption should look at first is the 2 million enterprise clients. The figure of 1 billion individual users speaks to how widely the product has spread, but it doesn't serve as grounds for deciding whether to bring it into an organization. If you're considering internal adoption, it makes sense to separate and organize items such as contract type, scope of data processing, and access rights from individual account usage.
The five-month stretch it took to go from 900 million to 1 billion users is also worth considering together. If you're using a vendor's growth rate as grounds for adoption, it's more accurate to look not just at the numbers from the steep growth period but also at the numbers from the period when growth flattened out.
Price Cuts and the "Economics of Abundance"
This announcement came right after OpenAI cut prices for 'GPT-5.6 Terra' and 'GPT-5.6 Luna' by 20 to 80 percent. OpenAI CFO Sarah Friar introduced this price cut as an "economics of abundance" strategy, explaining that "as the cost of useful intelligence falls, the range of tasks worth performing expands." Regarding infrastructure, she emphasized that "the goal isn't simply to build the most infrastructure possible, but to deploy the right capacity at the right time in line with reliable demand."
The practical significance of this point for Korean companies is a recalculation of unit costs. If there are automation projects that were shelved because they didn't pencil out under the old pricing, the fact that the cuts range from 20 to 80 percent depending on the model gives grounds to redo that calculation. However, since the price cuts are specifically for Terra and Luna, you first need to check whether the model your company actually calls falls within that scope.
Not Assuming There's Only One Supplier
While ChatGPT's growth has somewhat slowed since late last year, Google and Anthropic's AI models have successively emerged as strong competitors. Anthropic released models such as 'Claude Mythos' and 'Fable,' at one point surpassing OpenAI in corporate valuation, while OpenAI countered with 'GPT-5.6 Sol,' followed by the coding model 'Codex' and the agent tool 'ChatGPT Work.' The cost-efficient 'GPT-5.6 Terra' and 'GPT-5.6 Luna' emerged from this same competitive flow.
So there's one thing worth checking this week. Try organizing onto a single page the names of the models actually being called within your company and their unit costs. If the list has only one line, that means your structure simply accepts whatever terms come with each price change; if it has multiple lines, you can cross-reference which items this price cut applies to.
Source: OpenAI Surpasses 1 Billion Active Users Three Years and Eight Months After ChatGPT's Launch