CUDA isn't Nvidia's moat anymore. Its balance sheet is.
This video discusses Nvidia's competitive advantages in the AI data center market, arguing that its balance sheet and ability to finance customers and power projects are now more significant than its CUDA platform. Analysts expect Nvidia's AI data center market share to decrease from 90% to 70% over two years, with Google's TPU, Amazon's Trainium, and AMD's GPUs taking share, but this is not seen as a major concern due to overall market growth.
Ahead of its earnings report, analysts discussed Nvidia's position as a bellwether for AI and the data center buildout. Patrick Moorhead, founder, CEO, and chief analyst at Moor Insights & Strategy, contended that while Nvidia's CUDA platform remains a moat, its substantial balance sheet has become an even greater competitive advantage.
He explained that Nvidia's financial strength, including $250 billion in float and up to $500 billion in total financing capabilities, allows it to extend credit to customers and even finance power generation projects, citing an "Ohio deal" as an example. Moorhead noted that most competitors cannot match this financial capability, although Google and Broadcom have engaged in similar deals.
Regarding market share, Moorhead projected Nvidia's AI data center market share to fall from approximately 90% to 70% within two years. He identified Google's TPU as the primary taker of unit share, followed by Amazon with its Trainium, and AMD's GPUs. Despite the projected decline in percentage share, he emphasized that the overall market is more than doubling annually, mitigating concerns for Nvidia investors. The discussion also touched upon Nvidia's investment portfolio, including a significant stake in SpaceX, and its efforts to diversify its customer base beyond hyperscalers by breaking out sales to other data center entities (ACIE).