The average PC user probably thinks of ChatGPT when they hear of generative AI. According to a report examining the financials of its developer, OpenAI, despite its popularity, the huge costs of training AI models and running the servers that host them are so high that the company is on track to make a $5 billion operating loss.

This is according to The Information (via Window's Central), a forecast based on unpublished internal financial statements and numbers from various industry figures. founded in 2015, OpenAI has been a major player in Microsoft's multi-year, multi-billion dollar investment in AI companies. It has grown in size and scope through multiple investments, including the pouring of billions of dollars into the company.

Most recently, there were rumors of a $10 billion cash infusion in early 2023, a 75% share of OpenAI's profits and a 49% equity stake, as well as the integration of ChatGPT into Bing and other Microsoft systems. In return, in addition to the investment, OpenAI would get access to Azure cloud servers and significant fee discounts.

However, according to a report by The Information, OpenAI is not profitable enough and could post a $5 billion operating loss by the end of the fiscal year.

OpenAI claims to be spending roughly $7 billion on training and inference for LLMs (large-scale language models) and $1.5 billion on staffing. Other analysts estimate that running ChatGPT costs around $700,000 per day because of the cost of Nvidia's AI servers, but determining the exact cost is practically impossible. However, it is practically impossible to determine the exact cost. Even taking all this into account, why is OpenAI losing so much money when it should be making a profit?

The problem seems to be multifaceted. OpenAI has invested heavily in being the first company to make AGI (artificial intelligence) a reality, and that is undoubtedly eating into some of its profits. It is not the only company developing generative AI systems and faces competition from Anthropic, Amazon, Google, Nvidia, Meta, xAI, and many others.

The Information claims that OpenAI could go bankrupt, but this is not the first time such a thing has been said. Other analysts have noted that the relative costs of AI training, inference, and chip manufacturing will decrease over time as more companies enter the market.

Given the size of Microsoft's total investment to date, OpenAI is yet to attract managers. Earlier this year, OpenAI was estimated to be worth $80 billion, which is hardly chump change, but Microsoft could afford to buy it outright if things really seemed to be headed down the drain.

Then again, such an acquisition would undoubtedly be investigated by antitrust commissions around the world; OpenAI could raise ChatGPT's query costs to improve revenue, but this would likely result in fewer daily users.

Perhaps OpenAI's main hope for the future is the availability of cheaper AI servers to reduce operational costs.