Sure thing 
I have to admit that I havenât really thought about retail profitability beyond the following:
i) profitability can be compromised as long as the cash register rings, and
ii) it rings more over time, because by compromising on profitability, market share can be maintained/grown.
In other words, in my view, Amazon could even do retail at zero margin, because the purpose of this business is to bring liquidity to the company and grow the absolute number of users in Amazonâs ecosystem (buyers/sellers, users/IT solution providers, etc.) and maximize the network effects between them.
The most important driver in retail is the global growth of e-commerce. Amazonâs value depends on how this growth continues and how the cash can be utilized in creating network effects.
AWS is a damn good example of the companyâs ability to generate cash: AWS was initially the engine room of the online store, which became a âplatform within a platform,â and subsequently began to live a life of its own. Amazon hardly ever needs to consider how large investments they have to make in IT infrastructure for any business, because these investments are likely to be repaid by AWS users in the future.
The ad business is another good example of the same capability. Itâs profitable, of course, when there was already a ready infrastructure (online store + AWS), as well as the parties involved (buyers and sellers, on the AWS side, sellersâ IT departments and their IT suppliers)
This business is currently most strongly linked to retail.
I donât quite understand yet how Rivian fits into the overall picture, but itâs most likely an extension of both AWS and the ad business â and why not the online store too, if it can handle Rivianâs aftersales from maintenance to tuning?
The more ways they find to create new network effects, the more valuable it will become â and at the same time, harder to value solely on the cash flow from retail, if and when AWS grows larger than retail. Edit: I agree that retail alone would not justify such multipliers.