Clever Cloud’s Paris region now runs its own prefixes through AS213394 with 4 upstream providers spread across 3 datacenters. That sounds like a networking footnote right up until you remember the network path is the product. If someone else controls your announcements, your failover, latency tuning, and incident response are operating on borrowed authority.
Why Cloud Operators Reclaim Routing Control
The interesting part of Clever Cloud’s move is not the symbolism. It is the operating leverage. After years of relying on partners to announce its space, the company now controls routing policy directly in Paris, which means it can decide how traffic enters and exits, how quickly links fail over, and how much time gets wasted waiting for another provider’s NOC to bless a change.
That is the real milestone. Owning servers was phase 1. Owning racks and colo relationships was phase 2. Eventually, mature providers figure out that outsourcing BGP policy is a polished way to keep the steering wheel decorative.
Prefix Ownership Is Not Prefix Governance
Plenty of providers have address space on paper and very little control in practice. If a partner controls announcements, route acceptance, or mitigation actions, then the provider has prefixes the way a tenant has a say in the wallpaper. Technically involved. Not actually in charge.
That is why route objects, ROAs, filtering, and policy hygiene matter as much as possession. A clean BGP security posture and disciplined IPAM are not governance theater. They are what turn address resources into something an operator can defend, steer, and restore under pressure.
Paris Shows Why Third-Party Announcements Age Poorly
France is not a sleepy, single-homed market where basic transit is enough. Arcep put inbound interconnection traffic to France’s top 4 ISPs at 50.8 Tb/s at the end of 2024, with 54.2% carried through transit, 44.4% through private peering, and just 1.4% through public peering. In that kind of environment, routing policy is not back-office plumbing. It is service quality, cost control, and fault isolation dressed up in route maps.
That also explains why the sovereignty conversation keeps moving down the stack. Buyers may start with data residency, but operators eventually end up talking about who can change policy, publish ROAs, reroute around trouble, and keep services reachable without filing a polite request into someone else’s queue. Direct network connectivity control travels a lot better in a board meeting than “our upstream is looking into it.”
IPv4 Scarcity Still Makes Control Worth More
IPv6 adoption is improving, but unevenly, and dual-stack reality is not going anywhere on a convenient schedule. That leaves IPv4 as an operational asset, not a legacy souvenir. When RIPE-region pricing for some smaller blocks still sits around €27-31 per IP, mature providers have every reason to treat prefix policy as a business capability instead of an outsourced convenience.
The broader lesson from Clever Cloud is simple: the next sign that a hosting company has become a serious cloud operator may not be a flashy region launch. It may be the moment it stops letting somebody else decide how its prefixes behave on the internet.
FAQ
Why does owning an ASN matter for cloud operators?
It gives the operator direct control over routing policy, failover behavior, upstream selection, and incident response instead of depending on a partner to make BGP changes.
What is the difference between owning IP prefixes and owning prefix policy?
Owning prefixes means you hold the resources. Owning prefix policy means you control how they are announced, filtered, secured, and adjusted during outages or performance events.
Why is Clever Cloud’s Paris deployment significant?
It is a live operator example showing that reclaiming announcements produces practical control over latency, resilience, and escalation paths, not just cleaner architecture diagrams.
Does IPv6 make ASN and prefix control less important?
No. Dual-stack operations increase the need for disciplined routing governance because operators must manage security, policy, and reachability across both IPv4 and IPv6 at the same time.
