A bulk email provider can truthfully say:
We did not write that message.
That does not mean the provider has no role in what happens next.
Email delivery services supply the machinery customers need to send at scale: SMTP or API access, domain authentication, sending IPs, reputation monitoring, bounce processing, complaint handling, analytics, and enough throughput to reach enormous lists quickly.
That infrastructure is useful to legitimate businesses.
It is also useful to abusive senders.
Amazon SES makes the tension visible in its current enforcement documentation. AWS tracks bounce and complaint rates, places accounts under review when complaint rates become too high, and can pause sending if problems continue. Its documentation says high complaint rates are commonly associated with mail recipients did not expect or did not want. See Amazon SES sending review process FAQs.
AWS currently recommends keeping complaint rates below 0.1%; accounts at or above that level can be reviewed, and substantially higher rates can lead to sending being paused.
The exact threshold is less important than the principle:
The infrastructure provider is watching what its customers do because customer abuse damages the infrastructure itself.
Reputation is shared farther than the customer thinks
A bad sender does not only hurt recipients.
It can damage IP reputation, domain reputation, feedback relationships, abuse desks, and the deliverability of other customers using nearby infrastructure.
That gives bulk vendors a practical reason to enforce acceptable-use rules.
The vendor may require opt-in practices, investigate complaints, suspend credentials, rate-limit suspicious activity, or demand remediation before restoring service.
These controls are not proof that every complaint means spam. People sometimes report legitimate mail. Lists age. Addresses are mistyped. Transactional messages can annoy recipients too.
Patterns matter.
Large customers create awkward incentives
The uncomfortable part is economic.
A high-volume customer can also be a high-revenue customer.
Removing one may protect infrastructure while costing the provider money. Keeping one may preserve revenue while increasing complaints and reputational risk.
The provider therefore sits in the middle of the Spam Empire supply chain: not necessarily the author, not necessarily the beneficiary of the advertised product, but the party operating the pipes.
Industrial email abuse survives when everybody in that chain can point downstream and say the bad behavior belonged to somebody else.
The pipes still carried it.