OpenAI: ChatGPT Now Reaches 1 Billion Weekly Users and 2.5 Million Businesses

OpenAI disclosed that its products now reach more than 1 billion weekly active users and 2.5 million businesses, with ChatGPT daily message volume running roughly 50% higher six months after a user signs up than in their first month — established users attempt about twice as many distinct task types as new ones. The company positions GPT-6 Astra as its most capable model yet for computer use, web browsing, coding, and general professional work, while GPT-5.6 Sol cut end-to-end serving costs 20% and improved token efficiency more than 15%.

The infrastructure numbers underneath that scale are the more concrete story. OpenAI’s custom “Jalapeño” inference chip delivers 1.5-1.9x peak token throughput per watt and 1.7-3.6x lower latency than commercial inference systems, according to the company — efficiency gains it needs given how fast usage is compounding. That compounding matches the trajectory PwC modeled in its September Global Data Centre Outlook: cumulative AI infrastructure investment reaching $31.6 trillion through 2050, with annual capex climbing from roughly $800 billion in 2026 to $1.8 trillion by 2050. PwC names power availability — not chip access or connectivity — as the single biggest constraint on where that capital lands, and warns tighter export controls alone could cut cumulative investment by nearly a fifth.

Read together, the two reports describe the same moment from opposite ends: OpenAI is proving usage and unit economics can scale together right now, while PwC is modeling the multi-decade buildout required to keep that scaling possible. For consulting and enterprise-AI audiences advising on AI capital planning, the practical takeaway is that “infrastructure risk” isn’t abstract anymore — it’s the gap between how fast frontier labs can ship usage growth and how fast the physical capacity underneath it gets built.