San Francisco-based OpenAI doubled its losses last year to $540 million, according to a report by The Information, citing anonymous sources at the company.
The steep losses come from a combination of hiring high cost talent and training its machine-learning models key to its most popular product, ChatGPT.
Training AI models is incredibly energy and compute-intensive. The exorbitant costs are likely what brought about OpenAI's multibillion dollar partnership with Microsoft, where a key part of the deal pertained to Microsoft supplying the AI startup with its massive need for computing power.
The Information also reported that OpenAI CEO Sam Altman has privately suggested that the company must raise as much as $100 billion in the next few years to meet the costs of training the AI models it hopes will one day achieve the status of "artificial general intelligence," a term denoting an AI so advanced it can accomplish any intellectual task better than a human and has the ability to improve itself.
The company expects to earn $200 million in revenue this year, up from $28 million last year. The massive popularity of ChatGPT has opened up new revenue streams for the company as the public can pay for a premium version for $20 a month and app developers pay fees to access the AI software.
OpenAI is currently valued at $29 billion having raised at least $4 billion up to now and recently signed an additional $10 billion deal with Microsoft for funding over the next few years.
