SAN FRANCISCO/WASHINGTON (Reuters) – OpenAI, the company behind ChatGPT, is exploring manufacturing its own artificial intelligence chips, and has 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. However, since at least last year, it has been discussing various options to solve the shortage of the expensive AI chips that OpenAI relies on, according to people familiar with the matter.
These options included building its own AI chip, working more closely with other chipmakers including Nvidia as well as diversifying its suppliers beyond Nvidia (NVDA.O).
OpenAI declined to comment.
CEO Sam Altman has made getting more AI chips a top priority for the company. He has complained openly about the scarcity Of graphics processing units, a market dominated by Nvidia, which controls more than 80% of the global market for chips best suited to running artificial intelligence applications.
The effort to obtain more chips is tied to two main concerns identified by Altman: the shortage of advanced processors that run OpenAI’s software and the high costs associated with running the hardware needed to power its efforts and products.
Since 2020, OpenAI has been developing its generative AI techniques on a massive supercomputer built by Microsoft, one of its biggest backers, that uses 10,000 Nvidia graphics processing units (GPUs).
Running ChatGPT is very expensive for a company. Each query costs approximately 4 cents, according to an analysis by Bernstein analyst Stacey Rasgon. If ChatGPT queries grow to a tenth of Google’s search volume, it will require approximately $48.1 billion worth of GPUs initially and about $16 billion worth of chips per year to keep running.
The era of custom chips
The effort to develop its own AI chips would place OpenAI among a small group of large technology companies such as Alphabet’s Google (GOOGL.O) and Amazon.com (AMZN.O) that have sought to dominate the design of core chips. To their business.
It’s not clear whether OpenAI will move forward with a plan to build a custom chip. Doing so would be a major strategic initiative and a huge investment that could cost hundreds of millions of dollars annually, according to industry experts. Even if OpenAI devotes the necessary resources to the task, it will not guarantee success.
Acquiring a chip company can accelerate the process of building OpenAI’s own chip — as happened with Amazon.com and its acquisition of Annapurna Labs in 2015.
OpenAI has thought all the way to the point where it has done due diligence on a potential acquisition target, according to one of the people familiar with its plans.
The identity of the company checked by OpenAI cannot be identified during the purchase.
Even if OpenAI goes ahead with its plans for custom chips — including an acquisition — that effort will likely take several years, leaving the company dependent on commercial providers like Nvidia and Advanced Micro Devices (AMD.O) in the meantime.
Some major technology companies have been building their own processors for years with limited results. Meta (META.O)’s custom chip efforts have run into problems, prompting the company to scrap some of its AI chips, according to a Reuters report. The Facebook owner is now working on a newer chip that covers all types of AI work.
OpenAI’s main backer, Microsoft (MSFT.O), is also developing a custom AI chip that OpenAI is testing, The Information reported. The plans could indicate further divergence between the two companies.
Demand for specialized AI chips has soared since the launch of ChatGPT last year. Certain chips, or AI accelerators, are essential for training and running the latest generative AI technologies. Nvidia is one of the few chipmakers producing useful AI chips and dominates the market.
(Reporting by Anna Tong, Stephen Nellis and Max A. Cherny in San Francisco; Reporting by Muhammad for the Arabic Bulletin) Editing by Kenneth Lee and Stephen Coates
Our standards: Thomson Reuters Trust Principles.
“Infuriatingly humble alcohol fanatic. Unapologetic beer practitioner. Analyst.”