OpenAI Investigating Possibility of Developing Its Own AI Processors: Report

OpenAI, the company behind ChatGPT, is looking into making its own artificial intelligence chips and has even gone so far as to evaluate a potential acquisition target, according to people familiar with the company’s plans.

The company has not yet decided to move forward, according to recent internal discussions described to Reuters. Since at least last year, however, it has been discussing various options to solve the shortage of expensive AI chips that OpenAI relies on, according to people familiar with the matter.

These options have included building its own AI chip, closer cooperation with other chip manufacturers such as Nvidia, and also diversifying its suppliers outside of Nvidia.

OpenAI declined to comment.

CEO Sam Altman has made the acquisition of new AI chips a priority for the company. He has publicly complained about the scarcity of graphics processing units. The Nvidia-dominated market controls more than 80 percent of the global market for chips best suited to run AI applications.

The push for more chips is tied to two main concerns Altman identified: the lack of advanced processors to power OpenAI’s software, and the “eye-popping” costs of running the hardware needed to power its efforts and products.

As of 2020, OpenAI has been developing its generative artificial intelligence techniques on a massive supercomputer built by Microsoft, one of its biggest backers, using 10,000 Nvidia graphics processing units (GPUs).

Using ChatGPT is very expensive for a company. According to an analysis by Bernstein analyst Stacy Rasgon, each survey costs about 4 cents. If ChatGPT queries grow to one-tenth the scale of a Google search, it would require about $48.1 billion worth of GPUs to start with and about $16 billion worth of chips per year to run.

CUSTOM CHIP TIME

The effort to develop its own artificial intelligence chips would put OpenAI among a small group of major tech players, such as Alphabet’s Google and Amazon.com, which have sought to dominate the design of chips essential to their businesses.

It’s not clear if OpenAI is moving forward with plans to build a custom chip. It would be a major strategic initiative and a heavy investment that could cost hundreds of millions of dollars a year, according to industry veterans. Even if OpenAI provides resources for the task, it does not guarantee success.

Acquiring a chip company could speed up OpenAI’s process of building its own chip — as it did for Amazon.com and its 2015 purchase of Annapurna Labs.

OpenAI had thought its way to the point where it was conducting due diligence on a potential acquisition target, said one person familiar with its plans.

The identity of the OpenAI company that investigated the purchases was not revealed.

Even if OpenAI pursues plans for a custom chip — including an acquisition — the effort will likely take several years, leaving the company dependent on commercial suppliers such as Nvidia and Advanced Micro Devices.

Some big tech companies have been building their own processors for years with limited results. According to a Reuters report, problems with Meta’s custom chips have prompted the company to scrap some of its AI chips. The Facebook owner is now working on a newer chip that covers all types of AI jobs.

Microsoft, the main backer of OpenAI, is also developing a custom AI chip that OpenAI is testing, The Information has reported. The plans may mean the two companies are further apart.

Demand for specialized AI chips has skyrocketed since ChatGPT was released last year. Certain chips or AI accelerators are necessary to train and use the latest generative AI technology. Nvidia is one of the few chipmakers that produces useful AI chips and dominates the market.