ONTEL - The Platform for the AI epoch?

It does not take a rocket scientist to observe that the next great platform in computing is in play. It also does not take a rocket scientist to observe that OpenAI and NVIDIA have first mover advantages. A perhaps less talked about issue, but still well known, is the goodput stalls that occur between AI host operating CPUs and the GPUs doing the massively parallel and scalable AI heavy lifting.

Therefore, it takes no significant leap of faith to assert that two first movers, and an incumbent, have the opportunity to put together a platform for the next epoch of computing, and in the process not only lock-in a customer base for the next few decades of computing, but also work on some of the performance gaps in AI pipelines that adds another leg to the capacity stool - better utilization of whatever capacity is available.

In addition, as AI workloads become more varied there may emerge the opportunity for optimizations through complex instructions, and what better partner to have on that journey than Intel - whose incumbency is largely built on that foundation.

The game is not over, and nothing is set in stone. However, recent moves on the AI chessboard reinforce we are in the platform times. The game is not JUST about the best transformer, the best chip, the largest training center, or the first to AGI, it is about the platform that delivers the best outcomes. And in that game, ONTEL - OpenAI, NVIDIA, and Intel are definitely a player.