OpenAI, the company behind ChatGPT, is exploring the possibility of making its own chips dedicated to artificial intelligence. A feat revealed a Reuters from sources close to the business plans and that it already has also evaluated potential acquisitions as a solution to your needs. The reasons behind this choice (if it is confirmed) are of a different nature. This explains the moves of the giant now on everyone's lips.
Topics of this article:
Why OpenAI thinks about its AI chips
OpenAI's need to explore new avenues was born from a crisis: the shortage of expensive AI chips critical to the technologies the company develops. For at least the past year, Sam Altman's giant has been discussing various options to solve this problem, including building its own AI chips and collaborating with other chipmakers, such as Nvidia, as well as diversifying its suppliers beyond Nvidia itself .
However, it must be said that Nvidia it has made billions in the last 3 years thanks to the CPU market to power language models. This, among other things, led the company to be somewhat accused of monopoly of the market for the affected chips.
In this regard, CEO Sam Altman has declared the acquisition of more AI chips a top priority for OpenAI, having publicly complained about the shortage of graphics processing units (GPU), a market dominated by Nvidia, which controls more than 80% of the global market for chips best suited to running AI applications.
In essence Altman highlighted two serious concerns: The shortage of advanced processors to power OpenAI's software and the "staggering" costs associated with running the hardware needed to power its efforts and products.
Read also: Do AI detectors really work? OpenAI answers honestly
OpenAI's undertaking to build the chips is an expensive one
Creating a custom chip is not a task poco. It would require a major strategic initiative and a substantial investment that could amount to hundreds of millions of dollars a year in costs, according to industry experts.
And even if OpenAI decided to commit resources to the task, success would not be guaranteed. L'acquisition of a chip company it could speed up OpenAI's process of building its own chip, as happened with Amazon.com and its acquisition of Annapurna Labs in 2015. It should also be noted that some large tech companies have been building their own processors for years with limited results. The effort of Meta for a custom chip has been riddled with problems, forcing the company to scrap some of its AI chips.
At the moment, it is not yet clear whether the company will continue with its plan to build a custom chip. The realization of such a project – including a possible acquisition – will probably take several years, making the company dependent on commercial vendors like Nvidia and Advanced Micro Devices in the meantime.
The connection with Microsoft can help
Demand for specialized AI chips has skyrocketed since the launch of ChatGPT last year. Specific chips, or AI accelerators, are needed to train and run the latest generative AI technology. Nvidia is one of the few chipmakers that produces useful AI chips and dominates the market.
It is also worth considering the fact that Microsoft, one of the main supporters of Openai, is developing a custom AI chip that OpenAI is testing, as reported da The Information. The plans could signal a further estrangement between the two companies.
