OpenAI isn’t thinking like a software company anymore.
The company behind ChatGPT is internally projecting it could generate more than $280 billion in annual revenue by 2030, according to reporting citing people familiar with its financial outlook.
At first glance the number sounds unrealistic until you look at what OpenAI is turning into.
This isn’t a startup selling an app or service. It’s becoming a utility.
Today the company already produces over $20 billion in annualised revenue, a massive jump from roughly $6 billion just a year earlier as businesses begin paying for AI tools at scale.
Its flagship product, ChatGPT, is still growing rapidly, with internal messages indicating more than 800 million weekly users and double-digit monthly growth.
And that’s the key to understanding the $280B projection: people are no longer using AI occasionally they are running work on it.
Companies now rely on models for coding, research, customer support, document processing and automation. Instead of buying software once, they pay continuously as the AI performs tasks. OpenAI itself describes this model as a business that scales with “the value of intelligence,” meaning revenue rises the more work the system does.
In simple terms, AI is shifting from product to infrastructure similar to cloud computing or electricity.
To support that demand, OpenAI plans to spend roughly $600 billion on computing capacity by the end of the decade, building massive data-centre networks and acquiring specialised chips needed to run future models.
The company is also pursuing enormous funding rounds that could value it near or above $800 billion.
The economics are unusual: AI companies burn extraordinary capital upfront but generate recurring revenue once businesses depend on them. That’s why OpenAI’s forecast is less about predicting profits and more about predicting how much global work will be handled by machines.
If AI truly becomes embedded into everyday operations from writing software to running departments spending on intelligence scales with economic activity itself.
In that scenario, the addressable market stops being the software market.
It becomes the labour market. The $280 billion figure, then, isn’t just a financial milestone.
It’s a statement about where computing is headed: toward systems companies pay not to use, but to think and work for them.
Whether OpenAI reaches the number is uncertain. But the fact it’s even plausible explains why every major tech company and increasingly entire governments are racing to build or control AI infrastructure.
Because whoever operates the intelligence layer of the economy doesn’t just run an app.
They run the modern workplace.