OpenAI Set to Manufacture its In-House Custom Chipset to Take on Nvidia: Check Details

After the launch of the DeepSeek, a company that reportedly used Nvidia’s H800 chips for training its AI models are lower-cost chipsets compared to its competitors, another report hints that now the leading artificial intelligence company OpenAI is set to unveil their in-house chipset in order to take on Nvidia. OpenAI is set to launch its in-house artificial intelligence silicon, as per Reuters.

Right now, OpenAI relies on Nvidia’s GPUs (Graphics Processing Units) to train and run their AI models. But in a new report, the tech giant is now finalizing the design of its upcoming chipset in the next few months. The company will send its in-house chipset to Taiwan’s Semiconductor Manufacturing Co. for fabrication. The Reuters report also states that the company will meet its mass production at TSMC in 2026.

The process of giving the first design of a chipset to the chip factory is called ‘Tapping Out,’ and it costs tens of millions of dollars. Moreover, it will also take more than six months to produce a finished chip. But to reduce this timeline, OpenAI has to pay more to accelerate production and manufacturing.

If the initial tap-out goes smoothly, then the company will go for the mass production of its first in-house chipset and launch the Nvidia alternate in the market. The launch of the first in-house chipset is viewed as a strategic move to bolster its bargaining with the other chip suppliers in the market.  The chipset is designed by Richard Ho and Broadcom in collaboration with their teams in OpenAI. The report states that Richard has doubled his team in the past few months, and hence, we can soon see the new chipset in the market. The upcoming chipset will be built on cutting-edge 3-nanometer technology.