OpenAI

OpenAI is looking forward to developing its own AI Chips

According to persons familiar with the business's ambitions, OpenAI, the startup behind ChatGPT, is considering developing its own artificial intelligence chips and has gone as far as analysing a prospective acquisition target.

According to recent internal negotiations disclosed to Reuters, the corporation has not yet decided whether to proceed. According to those familiar with the situation, it has been discussing different solutions to the lack of pricey AI processors on which OpenAI relies since at least last year.

These alternatives have included developing its own artificial intelligence processor, collaborating more closely with other chipmakers, notably Nvidia, and diversifying its suppliers outside Nvidia (NVDA.O).

OpenAI did not respond to requests for comment.

CEO Sam Altman has made the purchase of more AI processors a primary priority for the firm. He has openly expressed his dissatisfaction with the paucity of graphics processing units, a sector controlled by Nvidia, which controls more than 80% of the worldwide market for the processors best suited to running AI applications.

Altman attributes the endeavour to two key concerns: a lack of the powerful processors that power OpenAI's software and the "eye-watering" price involved with running the gear required to power its activities and goods.

Since 2020, OpenAI has been developing generative artificial intelligence technologies on a gigantic supercomputer built by Microsoft, one of its most significant investors, that employs 10,000 Nvidia graphics processing units (GPUs).

The cost of running ChatGPT is too high for the firm. According to Bernstein analyst Stacy Rasgon's study, each inquiry costs about 4 cents. If ChatGPT inquiries grew to a tenth the size of Google searches, it would require around $48.1 billion in GPUs initially and approximately $16 billion in chips every year to remain functioning.

CUSTOM CHIPS ERA

An endeavour to build its own AI chips would place OpenAI in a tiny club of huge tech organisations, like Alphabet's Google (GOOGL.O) and Amazon.com (AMZN.O), that have sought control over the design of the chips that are critical to their operations.

It is unclear whether OpenAI will proceed with their goal to develop a bespoke microprocessor. According to industry professionals, doing so would be a big strategic undertaking and a significant investment that may cost hundreds of millions of dollars every year. Even if OpenAI devoted resources to the mission, success is not certain.

The identity of the firm OpenAI investigated purchasing could not be discovered.

Even if OpenAI proceeds with its ambitions for a bespoke processor, which may include an acquisition, the endeavour is expected to take many years, leaving the business reliant on commercial vendors such as Nvidia and Advanced Micro Devices (AMD.O) in the interim.

For years, some large technology corporations have been developing their own CPUs, with mixed success. According to Reuters, Meta's (META.O) bespoke chip endeavour has been plagued by problems, prompting the company to abandon several of its AI processors. The Facebook owner is currently developing a newer processor that will be capable of performing all forms of AI operations.

According to The Information, OpenAI's major supporter, Microsoft (MSFT.O), is also constructing a unique AI processor that OpenAI is testing. The proposals might indicate a deeper schism between the two organisations.

Since the release of ChatGPT last year, there has been a surge in demand for specialised AI processors. To train and execute the most recent generative AI technology, certain processors, or AI accelerators, are required. Nvidia, which dominates the industry, is one of the few chipmakers that makes usable AI processors.