ChatGPT developer reportedly considering manufacturing its own chips to deal with shortage

To avoid depending on chip manufacturers, the creator of ChatGPT OpenAI would consider manufacturing its AI chips itself. Enough to also gain independence and possibly optimize your expenses.

The figure is staggering: with the worldwide interest in artificial intelligence, Nvidia has multiplied its profits eightfold in one year. Since to run AI programs, you need servers, powered by specialized chips. Today, the firm is valued at more than $1,000 billion on the stock market. What interests OpenAI: the company, mainly known for ChatGPT, would consider manufacturing its own AI chips, revealed Reuters.

OpenAI is not really limited by its financial means, but by the chip market

The news agency was able to consult internal company discussions. We learn that the company would seek to deal with the shortage of AI chips, which makes them all the more expensive. OpenAI is reportedly considering several options to deal with it, but for now, no decision has been made.

The company could start manufacturing these components itself, for example in collaboration with other manufacturers. The name Nvidia appears, but the company would like to diversify its partners and not limit itself to just one. Besides, Reuters recalls that OpenAI CEO Sam Altman had already complained about the shortage of chips, particularly with regard to Nvidia, which controls 80% of the global market for models dedicated to AI. Regarding these revelations, no comments have been made by OpenAI.

The other advantage of OpenAI in manufacturing its chips

The shortage problem alone would not be the only one facing OpenAI. Despite the billions of dollars invested by Microsoft, the company faces staggering costs to acquire the hardware necessary to operate its servers and ultimately, its services. Moreover, Microsoft is developing an AI chip that OpenAI is testing right now, reports The Information. This could be another avenue to take.

Especially since according to Stacy Rasgon, analyst at Bernstein, if ChatGPT was one-tenth of Google, OpenAI would need around $48.1 billion in GPUs and $16 billion in chips each year to keep the chatbot running. .

Also, OpenAI would be in discussion with Jony Ive, historic designer at Apple, in the design of a hardware product based on artificial intelligence. The project would still be in its early stages, but this could encourage the company to invest in component manufacturing.

Becoming a chip builder is hard

However, certain obstacles could prevent the creator of ChatGPT from manufacturing its own chips. Other large technology companies have already tried the adventure: we are talking about Google, Meta and Amazon, all three of which are among the Gamam. A very closed circle. For these three giants too, chips and GPUs are fundamental to their activities.

One can easily imagine that OpenAI would need hundreds of millions of dollars per year in investment to get started in this business. And even if the company had the required funds, that doesn’t mean it would succeed. Especially since it would take years. Another challenge to take up: that of the potential artificial intelligence bubble. Other technological innovations have not met with great success, at least not yet: cryptocurrency, virtual reality, the metaverse. All have had their period of glory, but have not (yet) generated a revolution.

The solution to this would be to acquire a chip manufacturer. OpenAI is reportedly considering this route and has carried out checks on a potential target. However, we do not know which actor it would be.