In August, Google hired the executives and engineers of the artificial intelligence (AI) start-up Character AI. In June, Amazon did the same with recruits from Adept AI. And, in March, Microsoft had already made an impression by absorbing part of the Inflection AI team, including its co-founder, Mustafa Suleyman, a former DeepMind employee. However, this company had been created just two years earlier to compete with the chatbot ChatGPT and was valued at $4 billion (€3.6 billion), after having raised $1.3 billion in capital. These poachings were noted throughout the sector and seen as a sign of the economic pressure to which, after a phase of euphoria, young AI shoots are now subjected, in the United States or in Europe.
Read also | Article reserved for our subscribers After the euphoria, fears of a bubble around generative artificial intelligence
Add to your selections
“AI startups often spend a lot and can survive one to two years at most without raising funds.”explained Clément Delangue, co-founder of Hugging Face, in an interview with the World. “I think that[en 2024] we will see more and more focus on the profitability of AI start-ups”added the general director of this platform for publishing AI tools and models, recounting that he was contacted every week by “ten start-ups” wishing “to get bought out”.
Read also | Article reserved for our subscribers “Artificial intelligence is a bubble: there is a gap between the costs, which are very high, and the potential revenues”
Add to your selections
The sky has darkened a little since some have been deploring the existence of a speculative bubble in generative AI, which makes it possible to generate and process texts, images or sounds. “Generative AI: Too much expense for too little benefit?” asked the investment bank Goldman Sachs in June. The cost of computing to train and operate AIs, as well as the persistence of errors in their responses, raises questions about the pace of their deployment in companies.
Heavy investments
In addition, start-ups, often manufacturers of AI models or services, must also deal with digital giants, both partners and competitors: through their cloud business services subsidiaries (cloud computing), they provide start-ups with computing and distribute their models. In addition, Google, Microsoft or Meta can deploy AI assistants on their own smartphones, search engines, social networks or office software.
Even the company that created ChatGPT, a leading figure in the sector, is raising questions. “How can OpenAI survive?”, asked, in July, in a provocative post, the communicator and AI-critic essayist Ed Zitron, doubting the chances of “to make profitable” the company. According to the media The Information, OpenAI’s revenues have doubled in one year and exceed 3 billion dollars at an annual rate. But its deficit reaches 5 billion, due in particular to the 4 billion of computing costs paid to its partner Microsoft. The latter invested 13 billion dollars in OpenAI, but the start-up is reportedly looking to raise more than 5 billion again, which would value it at 150 billion dollars, according to the American press.
You have 48.6% of this article left to read. The rest is reserved for subscribers.
Related News :