OpenAI wants to venture into chip development for better performance

OpenAI, the creator of ChatGPT, is currently exploring the development of its own artificial intelligence hardware. It has even conducted assessments of a possible acquisition target.

Some recent reports have suggested that the company has not reached a definitive decision regarding this course of action. Regardless, inside sources have revealed that OpenAI has been exploring different options since last year to address the lack of expensive AI chips upon which it heavily depends.

One of the options explored by OpenAI is to create its own unique AI chips. They have also been strengthening ties with chip manufacturing giants such as Nvidia. There were internal discussions held about expanding the supplier network beyond Nvidia.

OpenAI’s CEO, Sam Altman, has emphasized the acquisition of additional AI chips as one of the main objectives for the company. He has openly stated his concerns over the lack of graphics processing units (GPUs). Nvidia is a global leader

in chip manufacturing, commanding over 80% of the global market share.

Sam Altman has pointed out two reasons behind this venture to procure chips. The first one is a deficit in the advanced processors essential for driving OpenAI's software, and the second is the high cost associated with operating the hardware required to support the organization's products.

Since 2020, OpenAI has been making giant strides in its generative artificial intelligence technologies on an extensive supercomputer. This progress is backed by Microsoft, which has also pledged to provide 10,000 GPUs from Nvidia.

The workings of ChatGPT create a huge financial burden on the company. An assessment conducted by Bernstein analyst Stacy Rasgon revealed that each query on ChatGPT carries an approximate cost of 4 cents. If the volume of ChatGPT queries were to reach one-tenth the magnitude of Google search, it would require an initial investment of approximately $48.1 billion in GPUs and an annual cost of roughly $16 billion on chips to maintain its operation.

If OpenAI succeeds in the development of its own AI chips, it would put the company on the list of a select few major tech giants, including Google within Alphabet and Amazon.com, which have undertaken initiatives to take control over the design of the chips that form the core of their operations.

A potential merger with a chip company will accelerate the development of OpenAI's proprietary chip, much like how Amazon.com expedited its chip-related efforts through the acquisition of Annapurna Labs in 2015.

Even in the event that OpenAI proceeds to develop its own chip, including the possibility of an acquisition, it is expected that this endeavor will be long-term. Until this endeavor becomes a reality, the company is likely to rely on commercial providers such as Nvidia and Advanced Micro Devices.

Several giant tech companies have started manufacturing their own processors over the years, but with somewhat constrained outcomes. Meta's endeavor to create custom chips has encountered challenges, resulting in the abandonment of certain AI chips. The parent company of Facebook is now in the process of developing a new chip designed to accommodate a wide spectrum of AI tasks.

Similarly, Microsoft, which is OpenAI’s primary investor, is actively working on its own custom AI chip, which OpenAI is currently evaluating. These

developments could indicate a growing separation between the two companies.

The demand for specialized AI chips has increased since the introduction of ChatGPT last year. Dedicated chips, often referred to as AI accelerators, are essential for the training and operation of the latest generative AI technologies. In this scenario, Nvidia stands out as one of the few chip manufacturers that produces valuable AI chips and holds a dominant position in the market.