nvidia hidden gem ai gpus market potential

Nvidia Seems Like a Hidden Gem With Underestimated Market Potential
A few months back, I dug into Micron’s financials and realized how crucial High Bandwidth Memory (HBM) is for their earnings. Some folks even say this memory isn't a typical cyclical product, leading me to wonder, "Who's buying this memory?"
One name that jumps out is Nvidia, and the data suggests it's getting less attention than it should. Micron forecasts a $100B market for HBM, with about 60% going into GPUs and accelerators. That makes sense with Nvidia's $40B revenue for Q3 2026.
This means $60B of HBM is used just for GPUs. If about 10% of a GPU’s cost is HBM, then the demand for $60B of HBM suggests a potential $600B market for those GPUs. Compared to Morgan Stanley's $400B GPU forecast from their $1.1T AI spending estimate, that brings the total market to over $1.3T.
Nvidia currently takes about 35% to 40% of AI spending, so maintaining that pace could mean $465B in data center revenue, already surpassing Wall Street's $390B consensus. Adding $15B–$30B for edge and retail GPUs, the total could hit $495B.
Right now, Nvidia converts about 66% of its revenue into EBITDA, with some estimates up to 70% by 2027. Even on the lower end, that's around $325B EBITDA, with an enterprise value of $4.5T, giving a 14x forward EV/EBITDA multiple or $192 per share. That's half of today’s multiple and about three times less than AMD’s 46x.
With 35% of hyperscaler spending, and today’s multiple, Nvidia appears to be the cheapest large-cap AI stock. Even if the multiple drops to 14x–30x and they trade at 25x compared to peers, it’s still around $336 per share. It's like the market has almost forgotten about it. AMD and Intel are trading at multiples much higher than Nvidia, and even Micron at about 30x.
Thoughts? Just my take while doing some research. Not investment advice.