Overview
OpenAI data reveals a significant demographic shift in ChatGPT usage, noting that more women now use the platform regularly than men. This trend represents a dramatic reversal from the platform’s launch, when usage was heavily skewed toward male users, reportedly around an 80-20 male split. The shift, which has been gradual since at least late 2025, suggests that generative AI is maturing from a technical novelty into a mainstream, socially accepted utility.
The company estimates that nearly half a billion women worldwide now utilize the tool on a weekly basis, contributing to a total user base approaching one billion. OpenAI attributes this rapid adoption curve to the technology's increasing familiarity, claiming the adoption cycle is moving faster than previous technologies like the internet or the personal computer.
However, the narrative of user adoption masks deeper structural concerns regarding the underlying infrastructure. While the consumer side shows remarkable growth, the core battleground remains compute power and geopolitical investment, areas where the US and China are locked in a high-stakes race for dominance.
The Maturation of AI and User Adoption Dynamics
The Maturation of AI and User Adoption Dynamics
The shift in user demographics signals more than just a change in user base; it reflects the successful transition of AI from a specialized tool into an everyday productivity layer. The initial adoption barriers that plagued early tech cycles—such as the steep learning curve or perceived technical difficulty—have largely dissipated.
OpenAI’s analysis suggests that as generative AI becomes integrated into daily workflows, its utility transcends niche technical applications. This mainstream acceptance has allowed the platform to shed its early, specialized user profile and achieve a broader, more diverse global reach.
The company, while celebrating the user growth, remains cautious, issuing warnings that technological adoption alone does not close all societal gaps. It emphasizes that systemic inequalities related to income, education, corporate size, sector, and geography still require deliberate tracking and intervention, regardless of the user count.
Geopolitical AI Spending and Compute Bottlenecks
Beyond user numbers, the operational reality of the AI sector is defined by capital expenditure and the race for compute. OpenAI provided an estimate of China's total AI spending for 2025, placing it between $97.2 billion and $125.3 billion, spread across government, private entities, and state-owned telecom providers.
While the US continues to lead in total capital expenditure, forecasting $527 billion for 2026, the data highlights a critical nuance: China's lower per-dollar costs afford it significant purchasing power. The estimated corporate spending in China alone, driven by giants like Alibaba, ByteDance, Huawei, and Tencent, falls between $67.5 billion and $76 billion.
The core constraint, however, remains compute. OpenAI stresses that computing power is the defining factor for AI success, enabling better models and driving down the cost per token. The company reports that its available compute capacity has roughly tripled year-over-year, moving from 0.2 gigawatts in 2023 to an estimated 1.9 gigawatts in 2025.
The Race for Gigawatts and Infrastructure Commitment
The accelerating demand for compute has transformed infrastructure into both the primary bottleneck and the ultimate competitive advantage. To stay ahead, OpenAI has committed to massive scaling efforts, initially announcing a 10-gigawatt Stargate commitment in January 2025.
The company has since solidified partnerships with hardware titans like Nvidia, Broadcom, Oracle, and AMD, totaling over 30 gigawatts on paper throughout 2025. These deals underscore the sheer scale of capital required to sustain advanced AI development.
More recently, the company has adjusted its public targets, now aiming for 30 gigawatts by 2030, with over 8 gigawatts already identified. This more measured framing suggests a pivot toward sustainable, verifiable scaling rather than the rapid-fire announcements of the previous year.
