I'd compare it to Windows Phone instead: third player coming in very late with a good product. We know how well that worked out.

One thing people hate to do the most is learn something new. You have a bunch of engineers who already know ins and outs of AWS. Your stuff _works_. Would you consider moving it elsewhere? I didn't think so. Now consider that there are literally thousands of ISVs offering their stuff there, for whom porting to GCP means hiring more people, increasing complexity of the product (so that it could run everywhere), etc. Why would they pick GCP instead of e.g. Azure, which has _insane_ level of sales access to enterprise accounts?

If I'm spending a million dollars a year and GCE can cut my costs by 30%, I'd consider switching because the cost savings of dedicating a could FTE's for a few months can pay for itself in under a year.

I'm a big fan of AWS, but I purposely avoid using too many AWS services that would lead to me being locked in and able to move if I wanted to. Porting deployment scripts to another API isn't too hard, but having to build and support my own replacement to SQS or DynamoDB (and then port my services to the replacement system) adds a lot more to my costs

Once enough people like you consider moving over, Amazon will just rejigger the pricing to not make it happen. It's not like Google has some kind of insurmountable capex or opex advantage. They just don't worry about the margins as much. Yet.

That's what I'm counting on -- Amazon and Google (and maybe Azure) will keep each other in check - if either one gets too much more expensive than the other, then customers will shift and the more expensive service will have to drop prices. It's a free market that works.

Everyone keeps underestimating Azure, and yet they'll be the only ones making money when price wars erupt and margins evaporate. They're the only ones who have unique offerings (Office, Exchange, MS SQL, Sharepoint, Dynamics, AD/Enterprise management, etc) that enterprise customers already use outside the cloud, and they're the only ones with a massive sales channel into small to medium businesses. Much like it took a decade and billions of dollars for Amazon to build the distribution network, building the stack Microsoft has also takes decades and billions, and there's also an entrenched competitor now: Microsoft, so no one is even going to try.