Sources claim that OpenAI, the parent company of ChatGPT, is exploring developing its own AI processors and looking into possible acquisition targets.

Reuters claims that internal discussions show the company has decided against moving further. Sources claim that OpenAI has been considering solutions to their expensive AI processor shortage since last year.
These plans include developing its AI processor, working more closely with chipmakers like Nvidia, and expanding its supply base. OpenAI denied the statement.

Sam Altman, the CEO, gives buying additional AI chips priority. He has openly lamented the lack of graphics processing units, as Nvidia has over 80% of the global market for CPUs that are designed for AI. Altman lists an absence of potent processors to drive OpenAI’s software and the “eye-watering” costs of maintaining its hardware to support its programs and products as justifications for acquiring more chips.

Microsoft has created a massive supercomputer employing 10,000 Nvidia GPUs to develop its generative artificial intelligence capabilities starting in 2020. Microsoft is one of OpenAI’s major backers.

The company has to pay a lot to run ChatGPT. According to Bernstein analyst Stacy Rasgon, each inquiry should cost 4 cents. ChatGPT would require $16 billion in CPUs and $48.1 billion in GPUs per year to run at a tenth of Google’s search volume.

Customized chips

OpenAI would join a narrow number of significant tech firms that have tried to construct their CPUs, including Alphabet’s Google (GOOGL.O) and Amazon.com (AMZN.O). If OpenAI will create a proprietary chip, it is unknown. Veteran businesspeople believe it would be a significant strategic investment that might cost hundreds of millions annually. Resources would not guarantee OpenAI’s success.

As it did for Amazon.com and Annapurna Labs in 2015, a semiconductor company acquisition may hasten OpenAI’s chip development. According to one source, OpenAI investigated potential purchase candidates using due diligence. The buying company that OpenAI looked at was unidentified.

Let’s say OpenAI makes acquisition-related efforts to obtain a proprietary processor. In that situation, it may take years and become reliant on AMD.O and Nvidia. Large IT companies have produced CPUs for years but with mixed results. According to Reuters, Meta (META.O) has abandoned several AI processors due to problems with its bespoke chip project. The owner of Facebook is creating a new processor for all AI operations.

The Information claims that Microsoft (MSFT.O), a key proponent of OpenAI, is developing and testing a proprietary AI processor. Plans might point to a growing separation between the businesses.

Since ChatGPT started last year, specialized AI chips have gained in popularity. AI accelerator chips are necessary for the newest generative AI technology to learn and function. Nvidia is the market leader and one of the few semiconductor manufacturers that produce powerful AI processors.