Former $NVIDIA (NVDA.US)$ employee shares his insight into t...

Former $NVIDIA(NVDA.US)$ employee shares his insight into the industry and $NVIDIA(NVDA.US)$ specific:

- $NVDA's moat is CUDA. It's the reason why everybody says; okay we'll just use $NVDA. The math behind LLMs that is being used right now has been around for many years, but making that work required hardware of speed that has only existed recently.

- ChatGPT started the gold rush. Now, everyone needs AI accelerators. He thinks that you will have many companies offering inference chips, but $NVDA doesn't necessarily want to be in that market segment.

- He explains in simple terms what CoWoS is and how it emerged over the years because of the need for more complex chips.

- He sees the whole AI wave at two speeds. The hardware side moves slowly every year, while the software side makes fast advancements every week with open source.

- He doesn't believe there is a moat for many AI startups because the big companies can make your product irrelevant.