Best SMTP for Cold Email: SMTP Servers and Providers That Allow Cold Outreach

Not within their rules. SendGrid requires affirmative consent for every non-transactional email, and states that consent cannot come from a purchased list, a lead generator or an affiliate: you have to be the party that obtained it. A cold prospect never gave it. People do try, and accounts do get shut down, usually once complaint volume shows up rather than at signup.

The same reasoning disqualifies Mailgun, Postmark, SMTP2GO and Elastic Email. The technical capability is there in every case. The permission is not. If you want the detail on how a relay differs from a mailbox, we cover it in SMTP relay vs SMTP server for cold email.

If you are keeping SendGrid for the opt-in mail it is built for, the current cost is $19.95 a month for 50,000 emails and $34.95 for 100,000 on Essentials, with a dedicated IP only from Pro at $89.95. The full band ladder is on our SendGrid pricing breakdown.

SES is excellent infrastructure and very cheap at volume, which is exactly why it comes up. The problem is the AWS Acceptable Use Policy, which prohibits sending "unsolicited mass email". A cold sequence to a prospect list is, on a plain reading, both unsolicited and mass. AWS also reviews sending reputation and can pause an account with little warning.

Treating SES as a loophole because the per-email price is low tends to end with a suspended AWS account, which for most companies is far more expensive than the sending was ever going to be.

On price alone SES is unbeatable, at $0.10 per 1,000 emails a la carte or $0.16 on the Essentials plan that new accounts now default to. Our full breakdown of Amazon SES pricing has the whole rate card, including the bounce and complaint thresholds AWS enforces.

No. Resend's Acceptable Use Policy, updated on August 5, 2026, lists prohibited sending purposes and the first one reads: "Spamming: You are prohibited from sending unsolicited messages of any kind, including cold outreach, purchased lists, or scraped contact data." Cold outreach is named directly, not implied by a general spam clause.

The thresholds attached to it are tighter than anything else on this page. Resend requires a complaint rate below 0.08% and a bounce rate below 4%, and states that accounts above those may be shut down without warning. Google's bulk sender guidance asks for under 0.3% complaints, so Resend's ceiling is roughly a quarter of that. It also states that no refunds are issued for policy breaches, including unused quota.

This is the same answer as SendGrid and Amazon SES, and for the same structural reason: a shared relay only works while every sender on it keeps complaints near zero. Resend is a genuinely good transactional email API, and its pricing is competitive, which we cover on our Resend pricing page. It is simply not sold for prospecting, and using it that way risks the account rather than just the campaign.

No, and Brevo's rules are the strictest of any platform on this page. Appendix 7 of Brevo's Terms of Service prohibits contacts "scraped on the internet, acquired or purchased from a third-party" and states the ban applies "whether You send transactional or marketing electronic communications." Most relays stop at marketing. Brevo closes the transactional door too.

The consent standard goes further than opt-in as most people use the phrase: "the Contact has to check a checkbox to subscribe to your newsletter AND the registration checkbox CAN'T be pre-checked", and "You must be able at any time to provide a proof of opt-in for each of your Contacts." A prospect who has never heard of you cannot have ticked a box, so cold outreach fails that test by definition.

Breaches are handled at the account level: Brevo warns that non compliance "may result in the immediate suspension or restriction of your entire account." If your product's password resets run through the same account, a prospecting experiment can take them down with it. The full plan ladder and cost per 1,000 is on our Brevo pricing page.

Mailbox providers take the opposite position. They sell you an inbox for ordinary business correspondence, and cold outreach that a human could plausibly have typed sits inside that. What they impose instead is a hard volume cap per mailbox: Google allows 2,000 messages a day and 3,000 external recipients, Microsoft allows 10,000 recipients a day at 30 messages a minute.

Microsoft is explicit that Exchange Online "isn't suited to accommodate bulk-mailing scenarios", which is the right way to read both caps. They are ceilings that describe an unusual day, not targets to plan against. This is the whole reason serious outbound runs on cold email infrastructure made of many modest mailboxes.

Work backwards from the send target at about 30 cold emails per mailbox per day, three mailboxes per domain, across 22 business days a month. Ten thousand emails a month is 455 a day, which needs roughly 16 mailboxes spread over 6 domains.

At $7.00 a seat on an annual commitment plus about $11 a year per domain, that lands near $117 a month in infrastructure. Fifty thousand a month needs about 76 mailboxes and 26 domains, near $556 a month, and at that scale infrastructure is roughly 85% of the total bill rather than the software. The calculator at the top of this page runs the arithmetic for any volume.

A relay quotes fractions of a cent per email and a mailbox costs $7 a month, so on a spreadsheet the relay wins by an order of magnitude. That comparison leaves out the two costs that actually bite: the account you can lose, and the domain reputation you cannot buy back.

Sending reputation is the only durable asset in outbound. It accrues to a domain over months of ordinary behavior, and it does not transfer. When a shared pool degrades, or an account is suspended, you restart that clock from zero. The $7 mailbox is not a sending cost, it is what a reputation you control costs to rent.

Once you accept that the answer is real mailboxes, the shortlist is short: Google Workspace and Microsoft 365 cover almost every case, and the choice usually follows whatever the rest of the company already uses. Beyond that, the questions worth asking are about your sending tool, not your mail host.

Does it connect to mailboxes you own over standard SMTP rather than locking you into its own pool? Does it rotate across the pool automatically? Does it cap per mailbox rather than per account? A tool that answers yes to all three lets you keep the reputation you build. Our SMTP email sender connects to any mailbox you already own.

You cannot buy a warmed up reputation, because the reputation that matters is attached to your sending domain, and your domain has no history until you send from it. That is the short answer to the whole category of pre-warmed SMTP servers, rented bulk mail servers and "buy SMTP for bulk email" offers.

Here is what is actually being sold under those names, and what each one really gets you.

Mailbox providers judge you on several signals at once: the sending domain, its authentication records, the IP, and above all how recipients react to your mail. The domain and the recipient behavior are the parts you cannot buy, and they are the parts that carry the most weight. An IP with a clean past will not rescue a brand new domain sending to a cold list, and a shared pool means someone else's complaint rate becomes your problem.

The strongest evidence that a dedicated IP is not a shortcut is that the reputable providers refuse to sell one to low volume senders. Every serious vendor gates it behind a volume floor, because an IP that sends inconsistently never establishes a stable reputation in the first place.

If you are sending a few thousand cold emails a month, that table is telling you something useful: you are below the volume where a dedicated IP helps, and a well run shared pool will almost certainly deliver better for you. Buy the dedicated IP when your volume justifies it, not before.

What actually works instead is unglamorous. Register your sending domains well ahead of time, set up SPF, DKIM and DMARC correctly on each, and ramp volume gradually from each new mailbox over several weeks while watching replies and bounces. Google and Microsoft published bulk sender requirements in February 2024 that made authentication and complaint rates non negotiable, and those rules reward exactly this kind of patient setup. Our guides to email warm up and choosing cold email domains cover the sequence in detail, and inbox rotation explains how to spread volume once you have more than one mailbox running.

You cannot send 300,000 cold emails a day as compliant cold outreach, and no SMTP provider will let you try for long. It is worth walking through the arithmetic, because it shows why the answer is structural rather than a question of picking a better provider.

Cold email deliverability caps a mailbox at roughly 30 sends a day. To reach 300,000 a day you would need about 10,000 mailboxes spread across roughly 3,300 sending domains. At $7.00 per Google Workspace seat that is $70,000 a month in seats alone, plus about $3,078 a month in domain registration. Around $73,000 a month before you have written a single line of copy.

Routing it through a relay instead makes the invoice smaller and the problem worse. Six and a half million messages a month costs only a few hundred dollars at Amazon SES rates, but SES opens a review at a 0.1 percent complaint rate and pauses sending at 0.5 percent. At that volume 0.1 percent is 6,600 complaints, which you would cross within days. Every reputable provider enforces a similar ceiling, and most name unsolicited bulk mail as grounds for termination outright.

The honest read is that a 300,000 a day target is a list problem rather than an infrastructure problem. Volumes like that come from bought or scraped lists, which is exactly what the acceptable use policies quoted further up this page prohibit. If you genuinely have that many qualified prospects, the workable route is a smaller daily send to a tightly targeted segment, which almost always produces more meetings than a blast does. A realistic ceiling for well-run cold outreach is in the low tens of thousands a month, and our page on cold email infrastructure covers how to build to that number safely.

Every shared SMTP provider polices consent, because on shared infrastructure your sending affects everyone else's delivery. What differs is how each one words the rule and where it enforces it. This is the comparison to run before you pick a cold email server, because a provider that is cheaper on paper is worthless if its policy shuts the account in week two.

Read the enforcement column rather than the policy column. Import time enforcement, which is what Mailchimp and ActiveCampaign use, tells you immediately: the upload fails and you know where you stand. Send time and statistical enforcement, which is what SendGrid and Amazon SES use, tells you late. That lag is the expensive part.

The SendGrid behaviour is the one that catches the most people. While an account sits under review, SendGrid keeps accepting and queueing your mail rather than rejecting it, so your API returns success and your dashboard shows sends. Nothing lands. A campaign can run a full week before anybody spots it. If you are weighing whether to stay on that stack at all, we priced nine SendGrid alternatives against SendGrid's own current ladder.

The compliance floor that applies to everyone, whatever provider you use

Separate from vendor policy, Gmail and Yahoo's bulk sender requirements have applied since February 2024 and no SMTP provider exempts you from them. SPF and DKIM on the sending domain, a DMARC record even at p=none, one click unsubscribe on bulk mail, valid forward and reverse DNS, and a spam complaint rate held under 0.3 percent. Miss any of those and delivery degrades no matter whose relay you rent.

This is also why a rented bulk mail server with a good price and no authentication story is a bad trade. You are not buying delivery, you are buying an IP whose reputation you did not build and cannot see.

Time to first send is the factor most SMTP comparisons leave out, and it is the one that decides whether you launch this week or next month. Two providers on any shortlist will gate you behind a manual review, and the gate has nothing to do with which plan you bought.

The SES row is the one that surprises people. AWS places every new account in a sandbox, and the sandbox status is unique per Region, so approval in us-east-1 does nothing for eu-west-1. While you are in it you can only send to addresses you have verified, which rules out cold outreach entirely by definition. Verifying your sending domain before you submit the request is AWS's documented way to get approved faster.

For prospecting specifically, the last row is usually the right answer regardless of what the relays charge. A mailbox you already own has sending history, an existing reputation with the receiving provider, and no third party deciding whether your use case is acceptable. The relay approach puts a policy team between you and your recipients, and cold outreach is the category those policies are written to catch. If you are working through the relay options anyway, our comparison of Amazon SES alternatives prices eight of them against SES's own current ladder, and Amazon SES vs Mailgun pricing works through the two cheapest in detail.