Jensen Huang, the CEO of Nvidia, has made a bold prediction. According to him, the processing power of artificial ielligence will increase one million times in the next decade and we will see more powerful models. Of course, experts have a differe opinion and say that maybe in the future, with the increase in computing power, we won’t have more advanced artificial ielligence.
Jensen Huang said in his rece speech: “Processing power will increase fourfold annually and (according to the law of scaling, this figure will exponeially) reach (one) million times in a decade.” This growth rate makes the resources needed to develop artificial ielligence much more powerful in the next 10 years.
Nvidia CEO’s prediction about future processing power
Nvidia has become the most valuable company in the world thanks to the advanceme of artificial ielligence. Technology companies need this company’s chips to provide the processing power needed to train artificial ielligence models.
Processing power obeys “laws of scalability,” according to which large artificial ielligence language models, or LLMs, will perform more advanced as they increase in size and access more processing power and data, says Jensen Huang.
Maybe Jensen Huang’s opinion is right; Because uil now, with the increase of educational and processing resources of models, we have seen better artificial ielligence. But experts say that we may not see extraordinary leaps in the developme of more advanced artificial ielligence models for the next few years.
For example, Ilya Sutskever, one of the founders of OpenAI, says that the 2010s was the era of scaling, where with increasing processing resources and more training data, significa improvemes were made in later models, but now the peaks of this era have been conquered. Of course, Jensen Huang says that Nvidia will move faster over the next decade to help other companies explore new heights of artificial ielligence.
