I keep hearing people compare Nvidia to Cisco Systems circa 1999. Bulls say Nvidia is different. Bears say it's the dot-com bubble all over again. I own 190 shares and here's my honest assessment.
First, the bull case. Nvidia's real moat isn't just GPUs. It's CUDA. 4.5M developers vs 1.8M in 2020. That's an ecosystem Cisco never had.
Cisco laid 'dark fiber', infrastructure nobody used. Nvidia's GPUs are powering real products generating real revenue. There are already 10 AI products over $1B ARR, 50 over $100M.
Enterprise AI spending is $37B in 2025, up from $11.5B in 2024. That's 3x growth. The demand is real, not hypothetical as it was for Cisco.
But the bear case is legitimate. At 50x P/E, Nvidia is priced for perfection. If hyperscaler capex peaks or competition emerges, the stock has a long way to fall.
Google's TPU, Amazon's Trainium, Microsoft's Maia - all custom chips designed to reduce Nvidia dependence. If any succeed at scale, Nvidia's moat gets tested (although these are arguably a long way off).
The wildcard is export controls. As a lawyer and former compliance officer at a big bank, I see something most overlook - the US govt plans to take 25% of Nvidia's China chip sales. They've literally crowned Nvidia as the winner.
This creates a regulatory moat. The government has skin in the game. But it's also a risk because China has to play along (which they soon likely will) and policy could change overnight.
But here's the valuation reality - Nvidia needs \~$198B earnings by 2028 to justify current prices. In my newsletter, I break down various scenarios: Bull case (25%): doubles from here. Base case (50%): down \~15%. Bear case (25%): down \~65%.