OpenAI is set to burn all of its cash at current expenditures in 12 months' time, according to a report by The Information. The San Francisco based AI startup leader is on track to lose $5 billion, mainly due to high capital expenditure on cloud infrastructure required for training and running AI programs, including that for ChatGPT.
OpenAI has long relied on Microsoft for providing computing infrastructure for running its AI programs. Microsoft was also one of the earliest investors, with a $1 billion investment, three years before the launch of ChatGPT.
The $80 billion market valued company OpenAI has around 350,000 Nvidia A100 chips for running its programs (for inference alone), with 290,000 of those specifically for running ChatGPT, a source told The Information. These servers containing A100s are rented to OpenAI at $1.3 per hour by Microsoft, resulting in $4 Billion of estimated expenditure on these servers in 2024 alone.
For training alone, the company plans to spend $3 billion in 2024 and another $1.5 billion on salaries for its 1500 strong employees.
OpenAI generates a revenue of about $2 billion through ChatGPT and about $1 billion through providing access to its large language model. According to its recent monthly revenue report, OpenAI is projected to earn between $3 billion to $4.5 billion in 2024, leaving behind a deficit of $5 billion in losses. The company needs to raise fresh funds to keep the pace of its development with OpenAI setting its eyes on developing Artificial General Intelligence (AGI).