Generative AI is expensive. Even very expensive. Last year, OpenAI, creator of the ChatGPT chatbot and image generator DALL-E, suffered a massive net loss of $540m (€490m), according to The numbers obtained by the site the information. This deficit, which doubled compared to the previous year, is mainly explained by the invoice sent from Azure, Microsoft’s cloud platform on which the American startup trains and then runs its artificial intelligence models. Despite the expected jump in revenue, these losses are expected to widen further in 2023, the first full year of ChatGPT availability, whose massive success translates into significant additional expenses.
high costs – The costs of running a synthetic AI-powered chatbot are already very high. At the beginning of February, Semi Analysis Put them at $700,000 a day, or $250 million a year. Since then, the chatbot’s audience has tripled to 1.76 billion visits in April, SameWeb estimates. Based on mid-term analysis calculations, this would represent $750 million in expenses on an annual basis — without taking into account any bill-reduction improvements. At the same time, OpenAI continues to train new generations of its AI models, which have an increasing number of parameters. Hence it requires more computing power.
paid offer – The expected increase in expenses should not be fully offset by the increase in turnover, generated by the huge enthusiasm around generative AI. according to the information, OpenAI’s revenue now reaches “hundreds of millions of dollars” at an annual rate, compared to just 28 million over the whole of 2022. The startup has several sources of income. It charges for access to its templates to developers who want to integrate it into their web or mobile applications. It also markets a subscription to ChatGPT at $20 per month. And it will launch a new paid offer dedicated to companies, which in particular will offer them more Data control.
$100 billion – Faced with these exorbitant expenses, to which are added the high salaries of AI researchers, its president Sam Altman believes that OpenAI will become “The most capital-intensive startup in Silicon Valley history”. According to him, it will have to raise $100 billion over the next few years to achieve its goal of developing general artificial intelligence capable of learning and carrying out any task. In the past year, the company has already collected “several billion dollars” from Microsoft. The exact amount has not been confirmed, but many press reports indicate the figure is ten billion. OpenAI should not receive all that much: part will be paid in kind, allowing it to use Azure supercomputers.
– In the face of OpenAI, Google combines its own artificial intelligence research laboratories
Elon Musk launched an AI startup to compete with OpenAI
“Hardcore beer fanatic. Falls down a lot. Professional coffee fan. Music ninja.”