Low-end Disruption of AWS

Sep 1, 2022

As part of moving this blog over to AWS, I needed to get "production access" to Amazon SES to send emails. The free tier for SES is substantial: 62,000 messages per month for free. The equivalent service on other platforms: Sendgrid $35/mo, Mailchimp $90/mo.

All AWS accounts start in the SES Sandbox – they can only send messages to verified emails. This is to protect AWS's IP email reputation (spam lists are archaically tied to IP addresses). Spammers and other bad actors could ruin the deliverability of real AWS customers (the top ones most likely have dedicated spam IPs).

Getting out of the "sandbox" to "production access" requires filing a help ticket and filing out some information. It was straightforward for me to do: (1) I've done it many times before, (2) I worked at a competitor (Google Cloud), and (3) I have access to a developer support plan.

But for many customers, it might be equivalent to shouting in the void. Like a job or investment rejection, you're given a generic "no", with little recourse.

Email is an interesting case of where AWS structurally can't serve both large and small customers at the same time. In terms of Christensen's disruptive innovation, this is low-end disruption. AWS will continue to ignore downmarket in favor of the higher margin upmarket customers.

My enthusiasm for AWS's monopolistic advantage can't be understated (read my AWS is Not a Dumb Pipe article). AWS has been successful at fending off other low-end disrupters:

  • Low-cost/bare metal – Digital Ocean, Hetzner
  • Developer experience/low friction – It doesn't come naturally to a large unit like AWS, but it isn't antithetical to serving both types of customers. With enough firepower, AWS could replicate (acquire) better developer experience. Not to mention the DNA of building one of the most-used consumer sites in the world.

I'm sure there will be other scenarios where the needs of large customers are directly opposed to that of small ones. At scale, you might run into the same issues as AWS (e.g., deliverability), but one of these openings might provide the opportunity to gain enough momentum to compete with the hyperscalers.