The Cost of Building AI Language Models and OpenAI’s Ambitious Future Plans
- OpenAI’s losses doubled to $540 million in 2021 during ChatGPT development and the acquisition of key employees from Google.
- Despite revenue reaching an annual pace of hundreds of millions of dollars, costs are expected to rise with growing customer usage and further development.
- OpenAI CEO Sam Altman has privately suggested the company may try to raise up to $100 billion in the coming years to achieve its goal of creating artificial general intelligence.
OpenAI’s losses soared to approximately $540 million in 2021, doubling as the startup developed ChatGPT and hired key personnel from Google. This previously unreported figure reveals the substantial costs associated with training machine-learning models in the period before OpenAI began selling access to ChatGPT.
Although revenue has surged—reaching an annual pace of hundreds of millions of dollars just weeks after launching the paid version of ChatGPT in February—costs are expected to continue rising. Increased customer usage of OpenAI’s AI technology and further software development will contribute to these growing expenses. In light of this financial strain, CEO Sam Altman has privately suggested that OpenAI may attempt to raise up to $100 billion in the coming years to develop artificial general intelligence capable of improving its own capabilities.
The substantial cost of building large AI language models like ChatGPT has been unclear until now. OpenAI’s case sheds light on the significant financial investment required to create such cutting-edge technology. As expenses continue to grow, it remains uncertain whether other AI companies can keep up with industry trendsetters like OpenAI.
One of OpenAI’s key objectives is to acquire new data sets for its software, some of which are not currently available on the internet. However, this has raised concerns among companies like Samsung, who are wary of their proprietary data inadvertently entering the ChatGPT data pool. Samsung’s recent ban on using generative AI with company materials illustrates these concerns.
In 2020, OpenAI was prepared to spend up to $75 million on Google Cloud services before Microsoft’s Azure platform became a more appealing option for future projects. The Information’s report also notes that OpenAI’s communications and policy chief, Steve Dowling, is planning to leave the company soon. As some of OpenAI’s expenses have resulted from hiring professionals from renowned companies like Apple and Google, it remains to be seen who will replace Dowling.