OpenAI mulls making own AI chips (MSFT)
Microsoft (NASDAQ:MSFT)- backed OpenAI is thinking of manufacturing its own artificial intelligence, or AI, chips and even evaluated a potential acquisition target, Reuters reported citing people with knowledge of the matter.
The maker of ChatGPT has not yet decided to move forward. However, since at least last year the company explored several options to solve the shortage of expensive AI chips that it relies on, the report added.
These options included making its own AI chip, working more closely with other chipmakers including Nvidia (NVDA) and also diversifying its suppliers beyond Nvidia.
It is not clear if OpenAI will move forward with a plan to make a custom chip, but doing so would be a vital strategic move and a could cost hundreds of millions of dollars a year, the report added citing industry veterans.
The company had even performed due diligence on a potential acquisition target, the report noted citing on eof its sources.
The aim to get more chips is tied to two major worries CEO Sam Altman has identified — a shortage of advanced processors which power OpenAI’s software and the costs linked with running the hardware required to power its products.
Altman has made acquiring more AI chips a top priority for the AI company. He has previoulsy talked about scarcity of graphics processing units, a market dominated by Nvidia, according to the report.
Since 2020, OpenAI has developed its generative AI tools on a supercomputer built by Microsoft which utilizes 10,000 of Nvidia’s graphics processing units, or GPUs.
Running ChatGPT is costly. Each query costs about 4 cents, the report noted citing Stacy Rasgon, an analysis from Bernstein. If queries in ChatGPT increase to a 10th the size of Google search, it would need about $48.1B worth of GPUs initially and nearly $16B of chips per year to keep operational.
An attempy to make its own AI chips would put OpenAI among a small group of tech giants like Alphabet’s (GOOG) (GOOGL) unit Google and Amazon.com (AMZN) which have tried to take control over designing the chips which are vital to their businesses.
Microsoft is also developing a custom AI chip which OpenAI is testing, the report added citing news outlet The Information.
Demand for specialized AI chips has surged since the launch of ChatGPT late in 2022. AI accelerators are required to train and run the latest generative AI tech and Nvidia is one of the few companies which develops powerful chips and dominates the market.
Companies around the world have launched their own large language models, or LLMs, which can provide services such as content and image generation, to name a few. Google’s Bard, Baidu’s (BIDU) Ernie Bot, Alibaba’s (BABA) Tongyi Qianwen and Tongyi Wanxiang, OpenAI’s upcoming text-to-image AI tool DALL·E 3, Meta Platforms’ (META) AudioCraft, SeamlessM4T, and Llama 2, and Getty Images’ (GETY) model called Generative AI by Getty Images, are some of them.