Satellite Failover Is Not the Same as Network Independence

What does redundancy actually mean for a remote site?

On 8 September 2026, AT&T Business announced an expanded agreement with Amazon Leo to bring low Earth orbit satellite connectivity into its existing business fibre and 5G network architecture.

AT&T describes the approach as a single managed architecture spanning fibre, 5G and satellite, rather than treating satellite as a standalone service. Customers will have a single AT&T account and bill, with AT&T providing the equipment, installation and 24/7 technical support.

The target market is familiar territory for anyone managing connectivity at remote or distributed sites: construction, mining, oil and gas, agriculture, logistics, emergency response and other operations where terrestrial connectivity may be unavailable, unreliable or difficult to install.

AT&T says fibre and 5G remain the backbone of its network, serving 98% of its traffic through its converged wireless and fibre offerings, while satellite adds another layer of resilience and geographic reach.

There is an important qualification, however. This is currently an announced service architecture rather than a fully launched, generally available product with published pricing and configurations. AT&T says availability, pricing and specific enterprise product configurations will be announced in the coming months, while the companies continue a phased deployment.

So what does this mean for businesses considering satellite as a backup? The answer depends on what you mean by redundancy.

Two connections don’t automatically mean two independent networks

At its simplest, network resilience means having another way to get online when the primary connection fails.

A fibre connection backed up by a satellite connection is clearly more resilient than fibre on its own. The two access technologies have very different physical characteristics, and a failure affecting a terrestrial connection does not necessarily affect a satellite connection.

However there are several different layers to consider. A site can have:

• two physical access technologies;
• two different network paths;
• two different network providers;
• two independent pieces of customer-premises equipment;
• or two completely independent connectivity services.

These are not necessarily the same thing. For example, a second connection may protect against a fibre cut without protecting against a problem with the router, the configuration, the site’s power supply or the system managing the connection.

That doesn’t make the second connection useless. It simply means that the resilience you get depends on which failure you are trying to protect against. This is why the word redundancy deserves a little more scrutiny than it sometimes gets.

Physical diversity is only one part of resilience

AT&T’s architecture brings together three different access technologies: fibre, 5G and LEO satellite. That provides genuine diversity at the access layer. But when evaluating a resilient connection, it is useful to ask what happens at each layer.

• Access – are there genuinely different ways for the site to reach the internet?
• Physical path – could the same physical incident affect both connections?
• Network – are the connections dependent on the same underlying network?
• Provider – are both services ultimately controlled by the same provider?
• Equipment – is the same router or configuration a single point of failure?
• Addressing – does the site’s public IP remain usable after failover?
• Failover – is the switch automatic, and how quickly does it happen?
• Applications – do cameras, NVRs, PLCs, SCADA and remote-access systems continue to work?

For an office where the main requirement is simply to keep staff online, these questions may not all matter. For a solar farm, construction site, remote industrial facility or other unmanned location, they can matter considerably more.

The public IP question is easy to overlook

One of the less obvious issues with backup connectivity is what happens to the site’s public IP address when the connection changes. For a normal user browsing the internet, changing IP addresses during an outage may make little practical difference. For a remote site running CCTV, NVRs, access control, industrial equipment or remote-management systems, it can be very different.

If systems or users need to reach the site using a known public IP address, simply having another internet connection available doesn’t necessarily mean that everything continues to work in exactly the same way after failover. AT&T’s announcement does not currently specify how public IP addressing will operate across failover in this new satellite service. That is not evidence that it cannot provide a suitable solution; it is simply a detail that has not yet been published.

For a business evaluating the service, it is therefore a sensible question to ask what happens to my site’s public IP address when the primary connection fails?

Where does this leave satellite?

Satellite is becoming an increasingly practical part of the connectivity mix for remote sites. That is the important point behind AT&T’s announcement. LEO satellite can provide connectivity where fibre is difficult or uneconomic to deploy, and it can provide a physically different access path when terrestrial connectivity is disrupted.

For many remote businesses, that is a significant improvement in resilience; however satellite does not need to be the entire resilience strategy. A site can use fibre, 4G, 5G and satellite as different connectivity bearers and then use a network layer to decide which connection should be active, when to fail over and how the site’s traffic should be handled.

That is where the distinction between having multiple connections and managing multiple connections as one resilient network becomes important.

Where Netcelero fits

Netcelero does not provide the underlying fibre, 4G, 5G or satellite service; it sits above those connections. With i-ctrl, a site can combine different connectivity bearers, including fixed line, 4G, 5G and satellite, and automatically fail between them. The key point is that the bearers can come from different providers.

That means a business can choose the connectivity that makes sense for the location rather than having to source every connection from a single carrier. For sites that require inbound connectivity, i-ctrl also provides a static public IP that persists across failover, so the site can remain reachable even when the underlying bearer changes.

It is only fair to apply the same questions to our own approach. On site, a single i-ctrl is still one device, and like any router it should be considered as part of your resilience planning. For sites where that matters, two i-ctrl units can be deployed in a high availability (HA) configuration, with the second unit on standby, ready to take over if the first fails. Both units share the same static public IP, so the site remains reachable at the same address even if the hardware changes. Behind the devices, Netcelero’s network runs across two data centres, in Dublin and Amsterdam, with failover between them, so the service does not rely on a single location.

That can be particularly relevant to remote CCTV, NVR, industrial and operational technology environments where changing the connectivity path should not mean changing how the site is accessed.

Netcelero is not a replacement for AT&T, Starlink, Amazon Leo or any other connectivity provider. It is the network layer that can sit above those services and manage multiple bearers as a single resilient connection.

Not sure how your site would answer the questions above? Contact Netcelero and we can review your current connectivity with you and show where the gaps are, from single points of failure to what happens to your public IP after failover.