What it is and is not
SMTP hosting is not the same as email hosting. Email hosting provides mailboxes people log into and read, whereas SMTP hosting is a relay for systems that send, with no mailbox behind it.
It is also distinct from running your own mail server. The transport is the easy part of that job, and the hard parts are reputation, blocklist monitoring, bounce handling and staying compliant with mailbox provider requirements that change.
The product is really the reputation and the reporting. Any host can open an SMTP connection, and what you are paying for is a sending identity receivers already trust plus a record of what happened to each message.
What providers offer
Providers cluster into three shapes.
Developer-focused transactional services sell an API alongside SMTP credentials, with per-message delivery events and bounce classification. They price on volume and target application mail.
Marketing platforms with SMTP access add a composer, list management and campaign scheduling on top. They suit teams sending campaigns as well as system mail.
Raw infrastructure sells sending capacity cheaply with minimal tooling, of which Amazon SES is the clearest example. The per-message cost is the lowest available and everything above the transport is your responsibility.
Testing services occupy a fourth position and are not sending services at all. They capture messages without delivering them, which suits development and misleads anyone who expects real delivery.
How to evaluate one
Four things matter once price is set aside.
Deliverability infrastructure means authentication managed as configuration, suppression enforced automatically, and reputation handled per sending domain rather than left to each customer.
Reporting depth is the feature teams underrate at purchase and want most during an incident. Per-message status, bounce reason, and history retained long enough to investigate a complaint that arrives a week later.
Stream separation determines whether the provider understands that transactional, marketing and cold outreach are different systems with different complaint profiles. One sending identity across all three means the worst degrades the best. Cold infrastructure is not marketing email. It runs on different domains, different mailboxes, different volumes and different deliverability rules, which is why our outbound-leaning users run it on separate tooling entirely.
Sending limits and ramp policy decide what happens on day one. New accounts are rate-limited everywhere, and a provider that explains its ramp is easier to plan around than one that throttles silently.
The cost question
The honest arithmetic favours raw infrastructure at scale. A team with engineering capacity can run its own SES or equivalent with a light automation layer for substantially less than a managed service, because the per-message rate is a fraction of the managed price.
I have run that arithmetic on real lists. At 100,000 contacts sending twice a month, Mailgun or SES costs $0.80 to $1.00 per thousand emails, about $160 a month in sending costs, against $400 to $600 for the same volume on Mailchimp or Klaviyo. For anyone with decent development resources, bringing your own Mailgun or SES account plus a lightweight automation layer cuts the bill by 70 to 80% at scale.
What that arithmetic omits is the layer being replaced: suppression, bounce classification, authentication management, reputation monitoring and the reporting to investigate a problem. Those have to be built and maintained.
The decision is therefore not price per message. It is whether you want to own that layer, and teams that decide they do are frequently right.
We built Nitrosend to integrate with the major SMTP relays, SendGrid, Postmark and SES among them, so you keep ownership of your sending infrastructure and its reputation while we run the layer above it.