OpenAI, the company behind ChatGPT, is exploring making its own artificial intelligence (AI)chips and has gone as far as evaluating a potential acquisition target, according to people familiar with the company’s plans.
The company has not yet decided to move ahead, according to recent internal discussions described to Reuters.
However, since at least lay year, it has discussed options to solve the shortage of expensive AI chips that OpenAI relies on, people said.
Photo: Reuters
These options have included building its own AI chip, working more closely with other chipmakers, including Nvidia, and also diversifying its suppliers beyond Nvidia.
OpenAI declined to comment. Its CEO, Sam Altman, has made the acquisition of more AI chips a top priority for the company. He has publicly complained about the scarcity of graphics processing units (GPU), a market dominated by Nvidia, which controls more than 80 percent of the global market for the chips best suited to run AI applications. The effort to obtain more chips is tied to two major concerns Altman has identified: a shortage of the advanced processors that power OpenAI’s software and the “eye-watering” costs associated with running the hardware necessary to power its efforts and products.
Since 2020, OpenAI has developed its generative AI technologies on a massive supercomputer constructed by Microsoft, one of OpenAI’s largest backers, that uses 10,000 Nvidia GPUs.
Running ChatGPT is very expensive for the company. Each query costs roughly US$0.04, according to an analysis from Bernstein analyst Stacy Rasgon.
If ChatGPT queries grow to one-10th the scale of Google’s search engine, it would require roughly US$48.1 billion in GPUs initially and about US$16 billion in chips per year to stay operational.
CUSTOM CHIPS ERA
An effort to develop its own AI chips would put OpenAI among a small group of large tech players such as Alphabet’s Google and Amazon.com that have sought to take control over designing the chips that are fundamental to their businesses.
It is unclear whether OpenAI would move ahead with a plan to build a custom chip. Doing so would be a major strategic initiative and a heavy investment that could amount to hundreds of millions of dollars a year in costs, according to industry veterans. Even if OpenAI committed resources to the task, it would not guarantee success. An acquisition of a chip company could speed up the process of building OpenAI’s own chip — as it did for Amazon.com and its acquisition of Annapurna Labs in 2015.
OpenAI had considered the path to the point where it performed its due diligence on a potential acquisition target, according to one of the people familiar with its plans.
The identity of the company OpenAI looked at purchasing could not be learned.
Even if OpenAI goes ahead with plans for a custom chip — including an acquisition — the effort is likely to take several years, leaving the company dependent on commercial providers like Nvidia and Advanced Micro Devices in the meantime. Some big tech companies have been building their own processors for years with limited results. Meta’s custom chip effort has been besotted with issues, leading 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 would span all types of AI work.
OpenAI’s main backer, Microsoft, is also developing a custom AI chip that OpenAI is testing, The Information has reported. The plans could signal further distancing between the two companies.
TECH BOOST: New TSMC wafer fabs in Arizona are to dramatically improve US advanced chip production, a report by market research firm TrendForce said With Taiwan Semiconductor Manufacturing Co (TSMC, 台積電) pouring large funds into Arizona, the US is expected to see an improvement in its status to become the second-largest maker of advanced semiconductors in 2027, Taipei-based market researcher TrendForce Corp (集邦科技) said in a report last week. TrendForce estimates the US would account for a 21 percent share in the global advanced integrated circuit (IC) production market by 2027, sharply up from the current 9 percent, as TSMC is investing US$65 billion to build three wafer fabs in Arizona, the report said. TrendForce defined the advanced chipmaking processes as the 7-nanometer process or more
China’s Huawei Technologies Co (華為) plans to start mass-producing its most advanced artificial intelligence (AI) chip in the first quarter of next year, even as it struggles to make enough chips due to US restrictions, two people familiar with the matter said. The telecoms conglomerate has sent samples of the Ascend 910C — its newest chip, meant to rival those made by US chipmaker Nvidia Corp — to some technology firms and started taking orders, the sources told Reuters. The 910C is being made by top Chinese contract chipmaker Semiconductor Manufacturing International Corp (SMIC, 中芯) on its N+2 process, but a lack
NVIDIA PLATFORM: Hon Hai’s Mexican facility is to begin production early next year and a Taiwan site is to enter production next month, Nvidia wrote on its blog Hon Hai Precision Industry Co (鴻海精密), the world’s biggest electronics manufacturer, yesterday said it is expanding production capacity of artificial intelligence (AI) servers based on Nvidia Corp’s Blackwell chips in Taiwan, the US and Mexico to cope with rising demand. Hon Hai’s new AI-enabled factories are to use Nvidia’s Omnivores platform to create 3D digital twins to plan and simulate automated production lines at a factory in Hsinchu, the company said in a statement. Nvidia’s Omnivores platform is for developing industrial AI simulation applications and helps bring facilities online faster. Hon Hai’s Mexican facility is to begin production early next year and the
Who would not want a social media audience that grows without new content? During the three years she paused production of her short do-it-yourself (DIY) farmer’s lifestyle videos, Chinese vlogger Li Ziqi (李子柒), 34, has seen her YouTube subscribers increase to 20.2 million from about 14 million. While YouTube is banned in China, her fan base there — although not the size of YouTube’s MrBeast, who has 330 million subscribers — is close to 100 million across the country’s social media platforms Douyin (抖音), Sina Weibo (新浪微博) and Xiaohongshu (小紅書). When Li finally released new videos last week — ending what has
RSS