Nvidia announced its third-quarter earnings today
, reporting record revenue of $2.6 billion and net profit of $838 million. However, if you’re into handicapping the market for artificial intelligence processors, the number that probably matters most is the $501 million in data center revenue that Nvidia earned last quarter. On the one hand, that’s more than double the $24o million that the data center business brought in a year ago.
On the other hand, however, is Intel, which brought in $16.1 billion during the third quarter and $8.9 billion from its data center division (although that’s only a 7 percent increase). This is the same Intel that just last month announced its upcoming line of Neural Network Processors
targeting data center workloads. Today, news broke that Intel has hired Raja Koduri away from AMD as part of an effort to build its own line of GPUs. According to Ryan Smith at AnandTech:
As part of today’s revelation, Intel has announced that they are instituting a new top-to-bottom GPU strategy. At the bottom, the company wants to extend their existing iGPU market into new classes of edge devices, and while Intel doesn’t go into much more detail than this, the fact that they use the term ‘edge’ strongly implies that we’re talking about IoT-class devices, where edge goes hand-in-hand with neural network inference.
I know I might be beating a dead horse at this point, but I think it’s too early to crown Nvidia the king of AI hardware. Its early dominance and fast growth have been very good from an investment perspective, but Intel and maybe even AMD could prove tough competition when it comes to actually shipping AI processors in the years to come. Nvidia dominated the first 5 years of what one might call the deep learning era, but it might have to look outside its comfort zone to keep that up as AI techniques and workloads, and the devices they run inside, continue to evolve.