OpenAI could produce its own AI chipset soon

Image from Financial Times

OpenAI, the company behind ChatGPT, is reportedly set to launch its first artificial intelligence (AI) chip as early as next year. So, what should you know about it?

Will OpenAI join the market soon?

For your information, this move is aimed at reducing dependence on Nvidia, which currently dominates the AI chip market.

According to a close source, the chip was developed in partnership with US semiconductor giant Broadcom. Broadcom CEO Hock Tan recently said that the company had received a US$10 billion (around RM46 billion) order commitment from a new customer, later confirmed to be OpenAI.

Moreover, the new chip, dubbed the “XPU,” will be used internally by OpenAI to train and operate AI models, including GPT-5, which increasingly require massive computing power.

Previously, other technology companies such as Google, Amazon, and Meta have also developed their own specialised chips to address the surge in AI demand.

Additionally, Broadcom previously worked with Google to produce TPU chips. Now, the deal with OpenAI is expected to boost the growth of Broadcom's custom chip business, which is expected to grow faster than Nvidia by 2026.

While Nvidia remains the leader in GPUs for AI, its sales growth is starting to slow compared to the initial boom of the AI investment wave.

Meanwhile, OpenAI is seen increasingly aggressively increasing capacity, with plans to double its computing centres in the next five months. Unfortunately, there are no further details on the news. So, we should wait for more announcements from OpenAI.

Did this news catch your attention? Stay tuned for more news like this at TechNave!