OpenAI says more women than men now use ChatGPT, flipping an 80-20 male split at launch

OpenAI says more women than men now use ChatGPT regularly. The company also estimates China's AI investments at up to $125 billion and sees computing power as a critical competitive advantage.

When ChatGPT launched in late 2022, around 80 percent of users had typically male first names. That picture has flipped: more than 50 percent of regular users now have female first names, according to OpenAI's own data. The shift has been in place since at least fall 2025, with a slight lean toward female users. At ChatGPT's current weekly active user count, that means nearly half a billion women worldwide now use the tool regularly. Overall, the company is closing in on one billion users, it says.

OpenAI credits the shift to ChatGPT's evolution from a technical niche product to an everyday tool. As generative AI became more familiar and socially accepted, early adoption barriers fell away. ChatGPT is moving through this cycle much faster than earlier technologies like the PC or the internet, the company claims. Still, OpenAI warns against assuming these gaps will close on their own. Inequalities still need to be tracked and addressed across income, education, company size, sector, and geography.

OpenAI puts China's AI spending at up to $125 billion

Despite the geopolitical spotlight on the US-China AI race, OpenAI says there is "surprisingly little clarity" about what China is actually spending, calling the data "opaque, fragmented, and likely incomplete."

OpenAI's Intelligence & Investigations team estimates China's total AI spending for 2025 at between $97.2 billion and $125.3 billion, spread across government, private companies, and state-owned telecom providers. The US still leads in capital expenditure with a forecast of $527 billion for 2026, but China's lower costs give it significantly more purchasing power per dollar.

The breakdown: OpenAI pegs Chinese corporate spending at $67.5 billion to $76 billion, driven mainly by Alibaba, ByteDance, Huawei, and Tencent. Direct government fiscal support lands between $16.8 billion and $36.4 billion, with large state-owned enterprises adding another $12.9 billion. Since OpenAI only counted direct subsidies, actual government spending is likely higher.

Beyond direct funding, China has a broader financing architecture built on state-controlled funds, state bank loans, and policy tools. For 2026, OpenAI expects Chinese corporate spending to grow 17 percent, reaching between $78.3 billion and $89.4 billion.

Computing power is becoming both a bottleneck and a competitive edge

OpenAI continues to stress that computing power is the defining factor in AI success. Better infrastructure enables more capable models, algorithmic advances drive down cost per token, and cheaper intelligence drives more usage, which in turn drives demand for even more capacity. The company says its available compute has roughly tripled year over year: from 0.2 gigawatts in 2023, to 0.6 gigawatts in 2024, to around 1.9 gigawatts in 2025.

But compute is increasingly a bottleneck, and OpenAI is betting big on staying ahead. Starting with a 10-gigawatt Stargate commitment in January 2025, the company signed partnerships with Nvidia, Broadcom, Oracle, and AMD over the course of 2025, totaling over 30 gigawatts on paper, though much about these deals remains unclear, starting with the biggest question: whether OpenAI can actually foot the bill.

The company now says it's targeting 30 gigawatts by 2030, with more than 8 gigawatts already "identified." That looks like a more cautious framing than last year's rapid-fire announcements. Reports also suggest OpenAI is pulling back on planned compute purchases in the EU.

While OpenAI has reportedly pitched its infrastructure lead over Anthropic to investors, Anthropic CEO Dario Amodei has called the approach high-risk, saying OpenAI is "doing stuff because it sounds cool." Anthropic is planning a more measured buildout of at least 10 gigawatts over the next few years.