OpenAI is considering developing its own AI chips to address chip shortages

TL;DR:

  • OpenAI is considering developing its own AI chips.
  • Evaluation of potential acquisition targets has been undertaken.
  • CEO Sam Altman prioritizes acquiring more AI chips.
  • OpenAI faces challenges due to the scarcity of graphics processing units (GPUs).
  • Running ChatGPT is costly, with each query at around 4 cents.
  • Building custom AI chips would align OpenAI with tech giants like Google and Amazon.
  • Developing a custom chip is a major strategic initiative with uncertainties.
  • Acquisition of a chip company could expedite OpenAI’s custom chip plans.
  • Microsoft is also developing a custom AI chip.
  • Increased demand for AI chips since the launch of ChatGPT.

Main AI News:

In a strategic move to address the ongoing challenges posed by a shortage of AI chips, OpenAI, the organization behind ChatGPT, is considering the development of its own artificial intelligence chips. According to insiders familiar with the company’s plans, OpenAI has even gone as far as evaluating potential acquisition targets in this pursuit.

While a final decision is yet to be made, recent internal discussions have revolved around tackling the scarcity of expensive AI chips that OpenAI relies on. These discussions have explored various options, including the possibility of manufacturing proprietary AI chips, forging closer partnerships with established chipmakers such as Nvidia, and diversifying its supplier base beyond its current dependence on Nvidia.

OpenAI’s CEO, Sam Altman, has unequivocally declared the acquisition of more AI chips a top priority for the organization. He has openly expressed concerns about the inadequacy of graphics processing units (GPUs), a market dominantly controlled by Nvidia, which commands over 80% of the global market for chips optimized for AI applications.

The urgency to secure a stable supply of chips is rooted in two major challenges that Altman has identified. Firstly, there’s a shortage of advanced processors essential to power OpenAI’s software. Secondly, there are the staggering costs associated with maintaining the hardware infrastructure required to drive OpenAI’s ambitious initiatives and product offerings.

Since 2020, OpenAI has been leveraging generative artificial intelligence technologies on a colossal supercomputer, which was constructed in collaboration with Microsoft, one of its primary supporters. This supercomputer utilizes a staggering 10,000 of Nvidia’s GPUs.

Running ChatGPT, however, comes at a considerable cost for OpenAI. As per an analysis by Bernstein analyst Stacy Rasgon, each ChatGPT query costs approximately 4 cents. Extrapolating to a scale even a fraction of Google’s search volume it would necessitate an initial investment of around $48.1 billion in GPUs and an ongoing expenditure of roughly $16 billion annually to maintain operations.

Should OpenAI proceed with its own AI chip development initiative, it would place the organization among a select few tech giants, including Alphabet’s Google and Amazon.com, who have ventured into designing chips tailored to their specific needs.

Nevertheless, the decision to embark on such an endeavor is not without its challenges. Developing a custom chip is a significant strategic undertaking and could entail investments amounting to hundreds of millions of dollars annually, as industry experts suggest. Furthermore, success is not guaranteed, even with ample resources allocated to the task.

Drawing inspiration from Amazon.com’s experience, OpenAI has considered the possibility of accelerating the process through acquisition. This approach aligns with Amazon.com’s acquisition of Annapurna Labs in 2015, which expedited its foray into designing custom chips.

The identity of the potential acquisition target remains undisclosed, and OpenAI’s course of action remains uncertain. Even if OpenAI proceeds with plans for a custom chip, including an acquisition, it is likely to be a multi-year effort, leaving the organization reliant on commercial providers such as Nvidia and Advanced Micro Devices in the interim.

Notably, some major tech players have attempted to develop their own processors with mixed results. Meta, for instance, faced challenges in its custom chip efforts, leading to the abandonment of certain AI chip projects. The company is now refocusing on a new chip designed to encompass a broader range of AI applications.

Intriguingly, Microsoft, OpenAI’s primary supporter, is also in the midst of developing a custom AI chip, as reported by The Information. These developments may signify a shift in the dynamics between the two organizations.

The demand for specialized AI chips has surged since the launch of ChatGPT. These specialized chips, also known as AI accelerators, are essential for training and operating the latest generative AI technologies. Nvidia currently stands as one of the few chip manufacturers with the capacity to produce highly effective AI chips, solidifying its dominance in this critical market segment.

Conclusion:

OpenAI’s exploration of in-house AI chip production signifies a proactive response to chip shortages and escalating costs. If successful, this move could reduce dependency on commercial chip providers, potentially impacting the competitive dynamics of the AI chip market. The involvement of major players like Microsoft suggests a broader trend toward custom AI chip development in the tech industry.

Source