Nvidia's Next Five Years: From AI Training Monopoly to Broader Compute Platform
Nvidia is the leading company in artificial intelligence (AI) today, but its business model may undergo significant changes over the next five years. Currently, the majority of Nvidia's revenue comes from selling powerful GPUs to a few giant cloud companies building AI models.
However, as the market for running AI models, known as inference, is expected to dwarf the market for training them, Nvidia will likely broaden its customer base and move beyond its current dominance in the cloud computing space. The company is also quietly building a software business on top of its hardware, which will provide recurring revenue with higher margins than one-time chip sales.
Nvidia's leadership has been talking about 'physical AI,' referring to robots, factories, and self-driving cars that need its chips to see and think. This expansion into new markets could make Nvidia less dependent on a single customer group, reducing its risk.
On the other hand, competition from rivals who are designing their own custom AI chips will likely pressure Nvidia's profit margins in the long run. The company may become larger but grow more slowly as the market becomes increasingly cyclical.