OpenAI, the creator of ChatGPT, is exploring options to start manufacturing its own AI chips. This initiative aims to combat the ongoing shortage of AI processing power.
According to sources familiar with the matter, the company has even considered potential acquisition targets related to this endeavor. OpenAI weighs options, including in-house chip development and closer ties with Nvidia, to reduce reliance on costly AI chips.
OpenAI CEO Prioritizes Acquiring More AI Chips
According to recent reports, the top priority for OpenAI’s CEO, Sam Altman, is acquiring more AI chips for the company. He has publicly voiced concerns about the scarcity of graphics processing units (GPUs).
Notably, the GPU market is predominantly controlled by Nvidia, which holds over 80% of the global market share for AI-capable chips.
More so, Sam Altman’s drive for more chips is due to the scarcity of crucial advanced processors and the desire to reduce external dependence. This also includes the exorbitant costs associated with maintaining the necessary AI infrastructure.
Moreover, running ChatGPT is a significant expense for OpenAI, with each query costing approximately 4 cents, according to analyst Stacy Rasgon. If ChatGPT’s query volume reached a scale close to Google’s search volume, it would necessitate an initial investment of about $48.1 billion in GPUs.
An annual expenditure of roughly $16 billion on chips would also be required. If OpenAI proceeds with developing its custom AI chips, it would place the company among a select group of tech giants. This group includes Google and Amazon, which have sought greater control over designing chips crucial to their operations.
Meanwhile, OpenAI’s decision to pursue custom chips would constitute a major strategic undertaking, entailing substantial financial commitments. The cost could reach hundreds of millions of dollars annually, but success remains to be determined even with dedicated resources.
The firm has contemplated acquiring a chip company to expedite its custom chip development efforts. This is similar to Amazon’s acquisition of Annapurna Labs in 2015. While the identity of the acquisition target remains undisclosed, OpenAI has progressed to the point of performing due diligence on potential candidates.
However, even if OpenAI opts to pursue custom chip development, the process is anticipated to extend over several years. In the interim, this would result in the company’s continued reliance on commercial chip providers such as Nvidia and Advanced Micro Devices (AMD).
Challenges in Custom Chip Development
Several major tech companies have ventured into custom chip development with varying degrees of success. For instance, Meta faced issues with its custom chip initiative, leading to the abandonment of some AI chip projects. The social media giant is now working on a newer chip to support a broader range of AI tasks.
In addition, Microsoft, a major backer of OpenAI, is reportedly working on its custom AI chip, which OpenAI is testing. This development could indicate a growing divergence between the two companies. The launch of ChatGPT in the previous year has fueled a surge in demand for specialized AI chips.
This particular AI accelerator is required for training and running the latest generative AI technologies. Presently, Nvidia stands out as one of the few manufacturers capable of producing effective AI chips and currently dominates this burgeoning market.
Read the full article here