--- title: "They Keep Saying Self-Hosted Email Is Dying, Mine Delivers Fine" description: "A new RIPE Labs study shows email consolidating around two providers, and the takes write themselves. I have run my own mail server for two years, and the numbers describe a trend, while my inbox describes a choice." author: "SelfHostedApp" pubDate: 2026-08-19T00:00:00.000Z tags: ["opinion", "email", "privacy", "self-hosting"] canonical_url: "https://selfhostedapp.com/blog/self-hosted-email-is-not-dying/" source_url: "https://selfhostedapp.com/blog/self-hosted-email-is-not-dying.md" --- # They Keep Saying Self-Hosted Email Is Dying, Mine Delivers Fine import { Picture } from 'astro:assets'; import EmailHero from '../../assets/images/heroes/self-hosted-email-is-not-dying-hero.jpg'; A RIPE Labs study landed in my feed last month with a title that the comment sections immediately turned into an obituary, ten years of DNS measurements across the million most-visited domains showing email consolidating around two providers, a stubborn plateau in DMARC enforcement, and a long tail of infrastructure the researchers struggled to even classify, and the discussion underneath wrote the conclusion for me in the first five comments, nobody can run their own mail server anymore, stop trying, pay Google. I run my own mail server, and I wrote the guide on this very blog, so you can calibrate my bias from that alone, and I want to walk into the data honestly before I walk out of it stubborn, because the study is good work and the trend it measures is real. The million most popular domains on the internet increasingly hand their MX records to Google or Microsoft, that consolidation has been steady for a decade, and if you measure email by what big websites do, the story is over and the duopoly won. But the interesting line in the study, the one the obituaries skipped, is the long tail, a stubborn mass of domains running their own infrastructure that refuses to fit the consolidation curve, and the measurement paper had to build special classification just to describe us, which tells you we are real, persistent, and apparently annoying enough to count. Here is the thing about the decline of self-hosted email that nobody in the discussion seems to say out loud, the number of people running mail servers was never the point, because email is not a hobby scale activity, it is a choice about custody, and custody is binary in a way that market share is not. When I moved my mail to my own domain on my own server, the question was never whether Google and Microsoft would keep growing, they will, the question was whether my correspondence would live in a system where an algorithm decides what is spam and a policy team decides when your account dies, or in a Postfix container I can read the logs of, and that question has the same answer at any market share. The deliverability fear is the real objection and I treat it seriously, because it is the one with teeth, and my two years of running Mailcow say it is a solved engineering problem with a maintenance tail rather than an impossible wall. SPF and DKIM and DMARC configured properly, a VPS provider whose IP ranges are not a landfill, a few weeks of warming a fresh IP by sending mail you would send anyway, and a warm-up period where you expect the occasional spam folder trip to a cold correspondent, and after that the system just works, my mail lands in inboxes at Gmail and Microsoft and everyone else at a rate indistinguishable from the years I paid for the privilege, and the two or three deliverability hiccups a year are debugged with headers and log lines instead of a support chat that cannot help you anyway. The honest costs, since this blog keeps a running list of those, are real, port 25 is blocked on half the cheap VPS providers so you choose the other half, some recipient networks treat small senders with suspicion that takes reputation to overcome, and you inherit a service that must be patched and monitored like everything else you run, with Healthchecks watching the queues and the certificates. What you get in exchange is the part that keeps me paying those costs, mail that cannot be scanned for ad targeting, an address that survives any company's product sunset, archives in a format older than most companies, and the specific satisfaction of watching a newsletter unsubscribe link point at a domain I own. The duopoly consolidation the study measures is, to me, the argument for the long tail rather than against it, because two companies holding the world's correspondence is exactly the concentration that makes an outage or a policy change or a business decision a global event, and resilience on the internet has always come from the boring people running their own small things badly and then less badly. Every self-hosted mail server that survives another year is a proof that the protocol still works without permission, that the system remains open at the edges even while the middle concentrates, and proofs like that only exist while someone keeps making them. So the decline is real and I am not in it, one stubborn domain in a very long tail, delivering mail on infrastructure nobody can reprice. The obituary was written for a body that keeps answering its own email, and I plan to keep being the typo in that story for a long time. If you are still brave enough to run a mail server, subscribe to the newsletter.