OpenAI is exploring the possibility of creating its own chips for artificial intelligence tasks

Dado Ruvic/Reuters

OpenAI is considering making its own chips for artificial intelligence and has already started evaluating a potential acquisition target.

Here's What We Know

According to people familiar with the startup's plans, the management has not yet made a final decision. However, since last year, the company has been discussing various options to address the shortage of expensive AI chips.

Among the options OpenAI is considering are building its own processor, working more closely with third-party manufacturers, and diversifying suppliers outside of Nvidia.

Last year, CEO Sam Altman made acquiring more AI chips a top priority. He openly complained about a shortage of GPUs best suited to running AI applications.

The effort to get more chips stems from two major problems Altman identified: a shortage of modern processors and the "exorbitant" cost of operating the hardware needed to run the company and its products.

Flashback

Since 2020, OpenAI has been developing its generative artificial intelligence technologies on a huge supercomputer built by Microsoft. The installation uses 10,000 Nvidia GPUs.

Running ChatGPT is costing the company a lot of money. According to analysts, the cost of each request is about 4 cents.

If ChatGPT's query volume grows to a tenth of Google's search volume, it will require GPUs costing about $48.1bn initially and about $16bn a year to keep up.

Source: Reuters