A new report has shown that to justify the massive cost associated with the artificial intelligence (AI) infrastructure build-out, AI companies need to start generating $600 billion in annual revenue – more than the entire individual GDPs of advanced economies like the United Arab Emirates (UAE), Singapore, Norway and Austria.
As Nvidia moved (briefly) into the position of the most valuable company in the world, questions have started to arise about how AI companies are going to turn the massive infrastructure investment into profit.
David Cahn, an analyst at Sequoia Capital (which has previously invested in Elon Musk’s xAI) shared a report last week suggesting that an issue he highlighted last September has grown from a $200 billion problem into a $600 billion one.
Cahn looked at Nvidia’s run-rate revenue forecast and estimated that the GPUs account for 50% of the cost, so doubled the figure to account for energy, buildings, staff, etc. That figure is then doubled again to allow for a 50% gross margin. This does not account for any revenue share for cloud vendors and middlemen.
Nvidia has reported that approximately half of its data center revenue comes from large cloud providers (Amazon, Microsoft, and Google, to name a few) with Microsoft alone accounting for 22% of its Q4 2024 revenue.
Despite warnings that return on the heavy investment will take time, Meta continues to invest heavily in increasing its AI capacity, with expenditure forecast to reach $40 this year.
Amazon has also announced its plans to continue to invest heavily into AI, with a planned $150 billion spent over the next 15 years on data centers.
Further fuelling the massive increases is Nvidia’s continued innovation in its chips. The B100 chip which it launched earlier this year promises 2.5 times the performance at only a 25% higher price and looks likely to drive further investment in infrastructure as tech giants scramble to keep up with each other.
At present, OpenAI is dominating the lion’s share of money in AI companies. The Information revealed that the company behind ChatGPT has increased its revenue to $3.4 billion, but its domination has not stopped continued investment by other companies.
Cahn’s conclusion is that Meta’s expectations of a long road ahead are closer to reality than anyone hoping to make a quick profit in AI ventures.
Featured image credit: generated by Ideogram
The post AI industry needs annual revenues of more than the UAE’s GDP to offset costs appeared first on ReadWrite.
