OpenAI, the company known for developing ChatGPT, is reportedly exploring the possibility of building its own custom artificial intelligence (AI) chips. The company has also been considering acquiring a chipmaker to address the shortage of expensive AI chips it relies on. While OpenAI has not made a final decision on these plans, it has been discussing various options to solve the shortage issue, including working closely with chipmaker Nvidia and diversifying its suppliers.
The scarcity of graphics processing units (GPUs), the chips best suited for running AI applications, has been a concern for OpenAI. With more than 80% market control, Nvidia dominates the global market for these chips. OpenAI’s CEO, Sam Altman, has made acquiring more AI chips a top priority for the company. The shortage of advanced processors and the high costs associated with running the necessary hardware have been major challenges that Altman aims to address.
To tackle these challenges, OpenAI has been developing its generative AI technologies on a large Microsoft supercomputer equipped with 10,000 Nvidia GPUs since 2020. However, running its flagship product, ChatGPT, comes with significant expenses. According to an analysis by Bernstein analyst Stacy Rasgon, each ChatGPT query costs roughly 4 cents. If ChatGPT queries were to reach a scale similar to Google search, OpenAI would need billions of dollars worth of chips to maintain its operations.
By exploring the development of its own AI chips, OpenAI would join the ranks of other tech giants like Google and Amazon that have pursued chip design to enhance their businesses. However, building custom chips is a major strategic initiative that requires substantial investments, likely amounting to hundreds of millions of dollars annually. Success is not guaranteed, and even with the resources dedicated to this endeavor, OpenAI would still rely on commercial providers like Nvidia and Advanced Micro Devices in the interim.
OpenAI has even considered the possibility of acquiring a chip company to expedite the development of its own chips, following in the footsteps of Amazon’s acquisition of Annapurna Labs in 2015. While the potential acquisition target remains undisclosed, it indicates OpenAI’s seriousness in pursuing alternative solutions.
However, some tech companies that have ventured into building their own processors have faced challenges. Meta, previously known as Facebook, has encountered issues with its custom chip efforts and had to abandon some AI chips. The company is now working on a new chip designed for various AI workloads. This demonstrates the complexities associated with creating custom chips.
The demand for specialized AI chips has surged since the launch of ChatGPT, as these chips are crucial for training and running advanced generative AI technology. Nvidia currently dominates the market by producing AI chips that are widely used. OpenAI’s main backer, Microsoft, is also reportedly developing a custom AI chip for testing purposes, potentially signaling further divergence between the two companies.
In conclusion, OpenAI is exploring the possibility of building its own custom AI chips to address the shortage it faces. This strategic move would provide the company with more control over its hardware requirements. However, the decision-making process is still ongoing, and other options such as collaborating with chipmakers and diversifying suppliers are being considered. The efforts to secure more AI chips highlight the challenges of running AI-driven technologies like ChatGPT and the need for innovative solutions in the market.