TSMC tipped to make OpenAI's AI chips instead of Samsung Foundry

Development of AI-powered tools seems to be a lucrative business. More and more investors are interested in funding such projects. However, developing AI products is quite expensive if you want to compete against the best in the industry. OpenAI, ChatGPT’s parent company, is well aware of this, which is why it has been looking for ways to optimize operating costs.

According to reports, one of the measures that OpenAI will take to save on costs in the long term is developing proprietary hardware to run AI services on. Currently, the company prefers Nvidia, the main supplier of AI hardware in the world. However, Nvidia's position in the market allows it to set prices that some, such as OpenAI, consider high.

OpenAI reportedly chooses TSMC to manufacture its AI chips, ditching Samsung

Samsung Foundry emerged as one of the main candidates to produce OpenAI's AI hardware following a meeting between Samsung Electronics Chairman Jay Y. Lee and OpenAI CEO Sam Altman last week (Feb 3 to Feb 9). Some sources suggested that production of OpenAI's AI chips on a Samsung 3nm process was one of the topics discussed.

However, a new report from Reuters claims that OpenAI has actually chosen TSMC's 3nm process instead.

The AI-focused company would not be the first big name to “get away” from Samsung after a few years of controversy and poor performance of its wafers. Companies like Qualcomm or Nvidia have also moved to TSMC.

OpenAI reportedly wants to start mass production of its AI chips in 2026. In the coming months, TSMC could receive the company’s designs to begin production tests. Reuters claims that the hardware design is in its final stages.

OpenAI would invest around $500 million in the design of its own AI chip. However, this would bring them huge savings in operating costs in the long run. Apple made a similar move by ditching Intel in favor of its own ARM chips for Macs.