The goal is to come up with a best case scenario for Nvidia (an upper bound valuation). Any approach here is likely to be fallible due to all the unknown unknowns. My intuition is that Nvidia’s market cap already surpasses the upper bound valuation. If thats the case then selling would be appropriate.

Rough back of the napkin calculation for upper bound valuation:

• Data centers spend about $25 billion annually on CPUs
• GPUs cost about 7x more than CPUs
• Suppose the demand for GPUs were to equal that of CPUs, annually.
• Also suppose that the price of GPUs did not decline over time
• Also assume that Nvidia had 100% market share for GPUs
• All of the above implies that Nvidia could grow to about $175 billion in potential revenue
• Currently Nvidia has about 90% market share
• Suppose Nvidia scales to a quarterly revenue of $40 billion (175 / 4 * 0.9). Based on their latest 10Q filing, this would imply an EPS of $10 approximately, assuming costs remained the same, which is a generous assumption
• Assume a conservative P/E ratio were applied since growth will slow down. Lets say the PE ratio is 25. This implies a stock price of $25 * $10 = $250
• Current share price is $124. So Nvidia in the best case can grow to a ~ $6 trillion valuation

Although this “best” case scenario seems unlikely to play out, my conclusion is that despite Nvidia’s recent growth, its still not abundantly clear that it is overvalued. I don’t think Nvidia is the Cisco of the 2000’s bubble.