OpenAI has been promising forever that their models will keep improving the more GPUs they have (so called scaling laws). The market is anticipating that AI is just going to keep improving forever the more GPUs they buy from NVIDIA. For the past months OpenAI have been expanding their datacenters with tens of thousands more GPUs. However, theres been rumors in the AI circles for the past months that OpenAI werent meeting their expectations with their next model, and that scaling laws were failing. OpenAI just released their biggest model ever, and its TERRIBLY underwhelming. It's barely any better than their previous non-reasoning model despite havings tens of thousands more GPUs, shattering the idea that buying infinite amounts of GPUs from NVIDIA is gonna keep improving their model. And on top of that, it costs 37x more than their previous model to use, and 535x more than what deepseek costs. People are only talking about this in AI circles right now, but i was also there to watch deepseek being talked about within a small sircle to crashing the market 4 days later. Be careful