OpenAI eyes chipmaking amid AI boom
CALIFORNIA, UNITED STATES — With its popular artificial intelligence (AI) chatbot ChatGPT driving rapid growth, OpenAI is exploring manufacturing its own chips to reduce reliance on vendors like Nvidia, according to Reuters. But the move into chipmaking would be costly and time-consuming.
Even with annual revenues nearing $1 billion, OpenAI struggles with supply constraints and inflated costs for the specialized graphics chips needed to train AI models like GPT-4.
Now valued at over $1 trillion, Nvidia controls over 80% of the global market for AI accelerators. Developing its own chips could give OpenAI more control and help manage expenses exceeding $700,000 per day.
Major tech firms like Google, Amazon and Microsoft have also developed custom AI chips to reduce reliance on vendors.
But building its own chips would be a costly, years-long endeavor for OpenAI, which currently relies on Nvidia’s GPUs to develop models like GPT-4 and DALL-E 3.
Recent struggles at AI chip startups also highlight the risks and uncertainties in the market. With Microsoft among its backers, OpenAI would need strong investor appetite for the capital-intensive move into chipmaking.
Recent internal discussions indicate that OpenAI has not made a final decision to enter chip manufacturing.
The decision, if confirmed, raises questions about OpenAI’s future relationship with Nvidia, as well as Microsoft, which is also developing its own AI chip.
