Jensen Huang Knows NVIDIA’s $5 Trillion Valuation Is Totally Not Based On A Bubble

When People Are Looking For Gold, Sell Them Shovels

The AI bubble, for that is what it is, hasn’t really made anyone a profit unless they are selling the hardware needed to run an LLM like NVIDIA has been.  It seems like every company with a bit of cash is jumping on the AI bandwagon and are desperately trying to find a way to make a profit.  Unfortunately even the big guys like Microsoft are being forced to shoehorn their AI features into every product they can in the search of a killer app.  Regardless if their users want an LLM in their text editor or not, AI being inserted into every app on their computer, phone or tablet. 

So far, it’s an interesting toy but when companies actually investigate the benefits of AI, they are finding that at best it is similar to the work produced by one of their employees.  Generally the output is worse and needs massaging to be acceptable for clients.  That hasn’t stopped mass layoffs ruining a large number of people’s livelihoods, including those working on implementing AI.

Meanwhile adding AI to your products has a definite cost and that is where NVIDIA shines.  They have both the best hardware and programming language for AI, and have convinced a generation of programmers to learn CUDA which runs best on their hardware. If you want to compete in the AI market, you have to buy NVIDIA hardware which has led to their $5 billion valuation.  That lead to a huge investment cost for businesses and a constant economic drain to power that equipment.  AI is not cheap to run, it eats electricity faster than Chrome devours free memory. 

The AI bubble is great for NVIDIA, and to a lesser extent AMD and Intel, but sooner or later companies are going to realize it isn’t actually making them any money.