Every ETO knows the moment: the vessel clears the harbour wall, the shore link hands over to satellite, and the owner's video call drops mid-sentence. Most superyacht connectivity complaints trace back to that handover, not to any single link being bad. Getting Peplink right on a yacht is less about buying hardware and more about engineering the transitions between shore Wi-Fi, cellular and satellite so the network never presents them to the guest.
Why single-link systems fail at sea
I have boarded plenty of vessels where the brief was "we have Starlink and it still drops." The assumption underneath is that high bandwidth equals high reliability, and at sea it simply does not. A single provider, however fast, is one atmospheric event, one blocked view of the sky, or one congested coastal mast away from taking the whole vessel dark.
Traditional failover makes this worse, not better, because it is reactive: it waits for the primary link to fail completely before switching, breaking every session in the process. For VoIP, video calls and anything the owner actually notices, that break-before-make behaviour is the problem we are hired to remove. The physical environment compounds it; saltwater, mast shadowing and hull steel interfere with RF in ways terrestrial engineers never deal with, and I find poor antenna placement behind a surprising number of "faulty router" complaints.
One tunnel, every link
The fix is architectural. SpeedFusion bonds Starlink, 5G and 4G into a single logical tunnel with one stable IP, so the question of which physical link is active becomes invisible to everything running on board. Corporate VPNs stay up, monitoring sessions persist, and the handover as you leave port happens inside the tunnel rather than in front of the user.
Bonding is not load balancing, and on a yacht the difference matters. Load balancing spreads sessions across links; when a link dies, its sessions die with it. Bonding splits traffic at packet level across every link at once, so losing one path costs capacity, not connections.
Starlink slots into this as the high-bandwidth pipe, not the whole answer. We typically run it as primary capacity with cellular as the low-latency anchor, then set traffic rules on top: guest streaming rides the satellite capacity, bridge and business-critical systems pin to a bonded tunnel that survives a blocked dish. The result is a hierarchy that matches how the vessel actually operates.
Hardware selection for the marine environment
On a 60-metre vessel with a dozen guests, I specify multi-modem routers in the HD4 class as the baseline: at least four cellular modems with SIMs from different providers active simultaneously, which is what keeps you connected crossing from French to Italian waters without anyone touching anything.
Two spec-sheet traps to avoid. First, judge routers on WAN inputs and encrypted throughput, not headline download speed; every packet in a SpeedFusion tunnel is encrypted and encapsulated, which costs CPU, and an undersized processor bottlenecks the network no matter how much bandwidth the links deliver. Under-speccing the core router is one of the most common causes of "instability" I diagnose. Second, respect the physical layer: long RF runs from mast to rack bleed signal, badly so at 5G frequencies. Dome systems that put the modems next to the antennas solve this elegantly; where conventional antennas are required, low-loss cabling such as LMR-400 is non-negotiable. We treat antennas and cabling as part of the network design, not an installation afterthought, along with the unglamorous details that decide reliability at sea: power redundancy, thermal management in the AV rack, and physically secure SIM slots.
Keeping the ETO in control
A network you cannot see mid-Atlantic is a liability. InControl2 gives the ETO live link performance, signal strength and data usage per SIM, remote firmware and configuration control, and GPS-correlated history that reveals dead zones along recurring cruising routes. For crews who do not want raw dashboards, we build simplified status portals that translate the telemetry into plain English, so anyone on watch can see at a glance whether the system is healthy and which link is misbehaving.
For vessels that want it fully off their plate, our managed service watches the SD-WAN fabric from shore; a degrading satellite link or congested tower usually shows in the telemetry before the bridge notices a slowdown, and we adjust before it becomes a guest-facing problem.
How we scope a deployment
No two vessels share an RF profile, which is why I hold that the design phase matters more than the hardware purchase. Our scoping starts with an RF survey of the superstructure to map signal shadows, an audit of existing cable runs for attenuation, and a realistic peak-load assessment of guest and crew demand. Then the configuration is matched to the cruising grounds: a Mediterranean season and a Pacific crossing need different modem bands, different SIM strategies and different SpeedFusion settings, and I oversee that configuration personally.
The short version
Bond every link into one tunnel, size the router for encrypted throughput rather than headline speed, keep the RF path short, and give the ETO real visibility. Do that and the harbour-wall handover stops being an event. If you have a refit or new build coming up, get in touch for a brief scoping conversation about the vessel's connectivity design.
Frequently asked questions
Can Starlink replace cellular and VSAT entirely?
Not on its own. It delivers impressive bandwidth but remains a single provider with a single failure profile: mast shadowing, weather and congestion in busy anchorages all show up in practice. Bonded with cellular it is excellent; alone it is a single point of failure with good marketing.
What does an HD4-class router give me that a single-modem unit does not?
Four simultaneous cellular modems across different carriers, which is what maintains service across borders and around provider-specific outages. It also carries the processing headroom to run SpeedFusion encryption at full tunnel speed.
Does the crew need SD-WAN expertise to run this?
No. The ETO manages day to day through InControl2 or a simplified status portal we build for the vessel, and a managed service can take the monitoring off the crew entirely.
Why does my fast connection still drop calls when we leave port?
Almost always the shore-to-sea handover. Standard failover breaks sessions when the active link changes; a bonded tunnel keeps one persistent connection across the transition, so calls survive the switch.
How disruptive is a retrofit?
Less than most owners expect if the design work is done first. The RF survey and cable audit tell us what can be reused; dome-based modem systems often reduce, rather than add, cabling through the vessel.