Nearly a billion people use ChatGPT, Codex weekly: OpenAI president

OpenAI President and cofounder Greg Brockman said nearly one billion people now use the company’s AI chatbot ChatGPT and coding tool Codex each week.

He shared the figures in a lengthy post on X about the future of AI-powered computing.

Brockman said the milestone reflects a broader shift in how people interact with computers, rather than just product growth.

The usage is rising as people apply the tools to software, spreadsheets, presentations, workflows, and science.

OpenAI also confirmed it is working to merge ChatGPT, Codex, and its Atlas browser into a single desktop “super app,” with Brockman leading the overhaul.

ChatGPT remains the leading consumer AI product, with over 900 million weekly active users and around 50 million subscribers.

🔗 Source: Greg Brockman

🧠 Food for thought

Implications, context, and why it matters.

The real bottleneck for AI is raw electricity

Rising computing needs are driving electricity use, with U.S. data center demand projected to reach 35 gigawatts within five years 1.
Power now limits hyperscale expansion. Tech companies have lined up more than 40 gigawatts of behind-the-meter power generation, which is produced on-site or nearby rather than drawn from the main grid, plus colocated generation as grid buildouts and interconnection backlogs fail to keep pace with demand 2.
Grid limits have already halted projects. Developers canceled a planned US$1 billion data center in West Jefferson, Ohio after concluding the local system could not support the facility without overloading infrastructure 2.
The planning model has shifted. Data center builders increasingly choose sites where scalable power already exists 2.

Enterprise AI adoption is wide, but productivity gains remain shallow

OpenAI said staff at more than 80% of Fortune 500 firms used ChatGPT in some form within its first year. Only 1% of companies call their AI deployment mature, which signals limited integration 3.
Many organizations still struggle to earn strong returns. Efficiency gains of up to 40% tend to come from embedding AI into core business processes when tools match real work patterns rather than sitting off to the side 4.
Further gains depend on rebuilding handoffs between business functions, since that is where efficiency often leaks away. This work can require custom AI systems rather than generic off-the-shelf tools 4.