Speaking at the 2025 AMD Advancing AI conference in San Jose, California, Altman said: "It [AI] has definitely been for us, and many other people, an explosion of usage over the past year.
“Models have definitely been good enough for people to build really great products- text, images, voices, all kinds of reasoning capabilities. We've also seen extremely quick adoption in the enterprise, with coding being one area where a lot of people talk about".
Emphasising that the tool is no longer just a novelty, the CEO of the AI giant, added: “What we’re hearing again and again in all these different ways is that these tools have gone from things that were fun and curious, to like truly useful in personal lives,” he said. “
“The fact that you can now, like, ask a system to go off and do some work for you autonomously is pretty remarkable.”
However, he noted that the pace of deployment and the complexity of new models are putting new strains on infrastructure, he revealed.
“Tons of change all the time,” he stated. “But one of the biggest differences has been we’ve moved to these reasoning models, these very long rollouts where a model will go up and think about a better answer or report.
“But, this is really putting pressure on model efficiency and long complex rollouts. We need tons of memory, tons of CPUs as well.
"Our infrastructure ramp over the last year, and our predictions for the next year has been a crazy, crazy thing to watch."
However, Looking ahead, Altman described the AI trajectory as something few could have predicted even five years ago.
“At the beginning of the 2020's, we didn’t kind of have AI. A bunch of other systems were out, but that was still the pre-GPT-3 era- just by a little bit,” he said.
“Now halfway mark since a decade, it’s really been remarkable progress from non-human GPT-3 model to GPT-4.5– these models that can really feel smart and helpful, and can give you real utility experiences, where people will look at this, if they could go back in time and say that feels almost impossible.
“If you went back to 2020 and said, by half of the decade we’re going to be at a system that you can talk to and it’s really smart- it’s like a smart person and can do work for you”
Initially sceptical about continuing the same rate of improvement, Altman now claimed he's confident.
"I think we are going to maintain the same rate of progress in the second half of the decade.
"I wasn't so sure about that a couple of years ago, but it looks like we would be able to deliver on that."
He concluded: "If you look forward to 2030 and the systems we can have, they will be capable of remarkable new stuff."
RELATED STORIES
OpenAI to open first international Stargate data centre in UAE
