The future of AI depends more on compute budgets than ideas. What does that mean for up-and-growing innovators who can’t match the trillion-dollar infrastructure game?
OpenAI is asking its investors that it now plans to expend about $600 billion on computing power by 2030. That’s the core of the latest report from Reuters and CNBC.
That isn’t a random forecast. It’s part of a broader pitch as OpenAI gears up for a potential IPO that could value the company near $1 trillion.
Here’s the first thing to grasp: $600 billion is huge, but it’s a downshift from earlier ambitions. CEO Sam Altman once spoke about spending $1.4 trillion on infrastructure. This revised figure suggests a more cautious push.
Why the reset?
OpenAI hopes to generate over $280 billion in revenue by 2030. Tying computing spending to expected revenue makes it easier to justify the capital. Investors never warm up to endless cash burn.
The math matters.
OpenAI had made around $13 billion in revenue while spending around $8 billion in 2025. These numbers show real growth. But they also show how steep the cost curve is for AI at scale.
Spending on compute isn’t abstract. It means data centres, GPUs, cooling, power, and specialised hardware that can handle training massive models. Buildouts of this scale require ongoing capital inflows- which is why investors like Nvidia, Amazon, and SoftBank are showing up with big cheques.
There’s a punch here: AI isn’t just about clever algorithms anymore.
The winner in this era is whoever can secure the infrastructure and capital to support those algorithms at scale. With rivals like Google and Anthropic also investing aggressively, the AI arms race has clearly shifted from research labs to real-world resource allocation.
This $600 billion number is a practical promise for OpenAI. It signals that the company sees massive computing as essential. But it also shows that even the most ambitious players know they can’t ignore financial discipline.
