OpenAI’s original artificial intelligence chatbot, ChatGPT, is proving to be a much more expensive endeavor to operate than initially thought, according to a new report by SemiAnalysis. With a whooping cost of approximately $694,444 a day, ChatGPT requires the use of 3,617 HGX A100 servers and 28,936 GPUs to keep functional.The project is a joint venture between OpenAI and Microsoft, with the latter being a substantial stakeholder in the firm.
One of the primary reasons behind such a financially draining system lies in the expensive specialized chips that are used to manage the AI Bot. To combat this, Microsoft is reported to be in the process of creating their own AI chip – called Athena – that would theoretically reduce running costs drastically.
Nonetheless, ChatGPT has so far shown itself to be a very capable system, capable of writing stories, emails, and even generating code from scratch. A team of four computer scientists, Raphaël Khoury, Anderson Avila, Jacob Brunelle, and Baba Mamadou Camara, have tested the system and found it to perform below the minimal security standards that are applicable in most contexts. Though it is still able to generate certain secure codes after the right prompting, it is certainly not at the level where it can be used to safely program apps and other data-sensitive projects.
All in all, the costs associated with ChatGPT’s operations may be too expensive for OpenAI to continue to use indefinitely. However, if their joint venture with Microsoft proves successful, then the AI chatbot could soon become a viable, self-sustaining platform with lower running costs. Dylan Patel, Chief Analyst at SemiAnalysis, stated that this would be the case “if GPT-4 is indeed even more expensive to operate than GPT-3”.
OpenAI is a company dedicated to the advancement of artificial intelligence and its detection and understanding of the world around it. With its technology, the company aims to create machines that possess the ability of human intelligence. The firm was founded back in 2018 by members of some of the world’s leading AI research institutions, including Google Brain, DeepMind, and OpenAI. Prominent people in the company’s boardroom include Elon Musk and research scientist Greg Brockman.
Dylan Patel serves as the Chief Analyst at SemiAnalysis, a research firm that specializes in AI technology. Dylan has a long list of accomplishments in the field ranging from recently launching a successful AI-driven fund to being published in some of the best-known AI and data-related journals. Dylan’s passion for his work is evident in his analytical insights and attention to detail.
