To reduce its dependence on Nvidia, OpenAI is actively working on designing its own chips dedicated to artificial intelligence. This ambitious project, which will require significant investment – both in time and money – could not only transform OpenAI's business model, but also the user experience of its various solutions, starting with ChatGPT.
Today, the proper functioning of OpenAI relies largely on Nvidia chips, which help to operate its various artificial intelligence models. This leads to a certain dependency, which OpenAI is reportedly seeking to reduce by developing its own semiconductors.
According to various US media outlets, the American startup hopes to finalize the design of its first chip this year, and then outsource its manufacture to the Taiwan Semiconductor Manufacturing Company (TSMC) in 2026. However, the manufacturing process for a brand-new chip is extremely complex, long and costly.
By developing its own chips, OpenAI hopes to reduce its dependence on Nvidia and gain autonomy in controlling its costs in the future. But above all, producing its own chips will enable the firm to adapt them to its own needs, in order to optimize the performance of its AI models.
For ChatGPT users, chips specially adapted to the proper functioning of these models could improve the platform's performance, if only by speeding up the processing of requests and responses, which are becoming ever faster and more accurate. In the long term, a reduction in costs could also have an impact on the prices of the services offered.
This initiative is reminiscent of those of Apple or Google, which now design their own chips in-house. For OpenAI, however, this deployment is likely to be carried out gradually, as there is currently no question of severing ties with Nvidia. – AFP Relaxnews
