However, there are potential risks for Nvidia in the longer term. Bloomberg Intelligence analyst Mandeep Singh suggests that some hyperscalers may eventually reduce orders, diluting Nvidia's market share. Furthermore, Nvidia's dominance in training chips is not matched by inference chips, which run generative AI models after they have been trained. As more businesses utilize AI, inference will become a more meaningful business, posing a risk to Nvidia as it faces increased competition from rivals like AMD and companies with their own custom silicon like Tesla.
Key takeaways:
- Nvidia, the world's most valuable company, is heavily reliant on a few key customers that contribute significantly to its revenue, with three customers each contributing over 10% of the company's global consolidated turnover.
- Nvidia is supply constrained and cannot simply produce more chips as it outsources the fabrication of its AI microchips to Taiwan's TSMC and does not have its own production facilities.
- The identities of Nvidia's major customers, referred to as Customers A, B, and C, are kept secret for competitive reasons and can change from one fiscal period to the next.
- Long-term risks for Nvidia include the shift from training to inference chips, with the latter expected to become a more meaningful business as more companies utilize AI, and the potential for hyperscalers to reduce orders, diluting Nvidia's market share.