OpenAI, the company behind ChatGPT, is reportedly exploring the possibility of developing its own artificial intelligence (AI) chips and has even considered acquiring an AI chip company, according to insiders familiar with the matter. While the company has not yet made a final decision, it has been discussing potential solutions to address the shortage of expensive AI chips it relies on. Options being considered include building its own AI chips, collaborating more closely with chipmakers like Nvidia, and diversifying its suppliers beyond Nvidia.
OpenAI’s CEO, Sam Altman, has identified acquiring more AI chips as a top priority for the company. He has expressed frustration with the scarcity and high costs associated with graphics processing units (GPUs), which are essential for running AI applications. Nvidia currently dominates the market for these chips, controlling over 80% of the global market share.
The scarcity of GPUs and the cost of running them are two major concerns for OpenAI. The company has been using a massive supercomputer built by Microsoft, one of its major backers, which relies on 10,000 Nvidia GPUs. Running its ChatGPT system is expensive, with each query costing around 4 cents. If ChatGPT queries were to reach a scale even a tenth of that of Google search, it would require billions of dollars’ worth of GPUs to keep operational.
If OpenAI decides to venture into developing its own AI chips, it would join the ranks of other tech giants like Google and Amazon, which have sought to design their own chips to support their businesses. However, building custom chips would require a significant investment and could cost hundreds of millions of dollars annually. Even if OpenAI were to commit to developing its own chips, success would not be guaranteed.
Alternatively, OpenAI has also considered acquiring an existing chip company, a move that could expedite the process of building its own chip, as was the case with Amazon’s acquisition of Annapurna Labs in 2015. Although the identity of the potential acquisition target remains unknown, OpenAI has reportedly conducted due diligence on the company.
Even if OpenAI proceeds with plans for a custom chip, including an acquisition, it would likely take several years before the chips are ready. In the meantime, the company would continue to rely on commercial providers like Nvidia and Advanced Micro Devices (AMD).
Developing custom AI chips has been a challenge for some big tech companies. Meta (formerly Facebook), for example, has faced difficulties with its chip efforts, leading to the scrapping of some AI chips. Microsoft, OpenAI’s main backer, is also working on a custom AI chip that OpenAI is currently testing. This development could signal a further separation between the two companies.
The demand for specialized AI chips has significantly increased since the launch of ChatGPT. These chips are crucial for training and running the latest generative AI technology. Nvidia currently dominates the market as one of the few chipmakers producing AI chips.
Overall, OpenAI’s exploration of developing its own AI chips reflects its efforts to address the shortage and high costs associated with GPU chips. While the company has not made a final decision, the potential move could position OpenAI alongside other tech giants that have sought greater control over chip design and supply.