OpenAI Misses Revenue, User Targets as IPO Looms

OpenAI missed key revenue and user growth targets, according to the Wall Street Journal. Bad timing. The company is racing toward a potential IPO, and those missed benchmarks don't help convince public market investors that OpenAI can turn popularity into profit.

Running large language models like GPT-4 costs serious money. Every query burns compute on expensive NVIDIA GPUs. Unlike regular software, where adding one more user costs nearly nothing, AI inference has real marginal costs. Users who push past subscription limits can actually cost OpenAI more than they pay in fees.

Weird position for a company about to pitch public investors.

The whole AI industry faces this problem. How do you price something that's genuinely expensive to run? OpenAI has Microsoft Azure backing its infrastructure, which helps with the bills. But that doesn't solve the underlying math. Can you charge enough to cover costs and still grow? Going public means answering that with a straight face. Right now, OpenAI's numbers suggest they can't. Not yet.