OpenAI is collaborating with Broadcom (AVGO, Financials) and Taiwan Semiconductor Manufacturing Company (TSM, Financials) to develop its first in-house chip for artificial intelligence, aiming to reduce its reliance on Nvidia (NVDA, Financials) chips, according to a report from Reuters.
the corporation is also using Advanced Micro Devices (AMD, Financials) chips via Microsoft's Azure platform in addition to continuing to use Nvidia CPUs.
Originally looking at constructing its own chip manufacturing foundries, OpenAI has chosen to abandon such ideas because of lengthy timescale and expensive prices, Reuters said. According to insiders who spoke with Reuters, OpenAI would instead concentrate on developing chips in-house and depend on outside partners like Broadcom and TSMC for manufacturing. According to the reports, 2026 is when the chips are supposed to be generated.
One of the biggest consumers of Nvidia's graphics processing units, OpenAI is behind the extensively utilized ChatGPT. But with persistent shortages and growing expenses, it is trying to diversify its supply chain, according to the Reuters story. According to Reuters, more than eighty percent of the AI chip industry consists of Nvidia's GPUs.
By means of its new MI300X processors, AMD aims to carve off a portion of the AI chip market held by Nvidia. Driven by the significant computer power needed to train and execute AI models, OpenAI expects a $5 billion loss this year on $3.7 billion in sales, hence the need to diversify chip vendors comes as natural.
