On 12 July 2019, the municipalities of Amsterdam and Haarlemmermeer announced a moratorium halting new data center construction within their borders, pending new policy. The 2020 policy that followed only allows new projects in designated zones, under strict sustainability rules such as a PUE of 1.2 for new builds (Dutch Data Center Association). The practical effect for anyone running infrastructure in the region isn't that demand has slowed: DDA's own research found more than a third of the roughly 211,000 m² of new data-floor projects now in the pipeline are planned outside the Amsterdam metro (MRA) region entirely. Growth has been pushed outward, into secondary sites, edge PoPs, and regional facilities that now have to stay connected to the core as if they were still in the same building.
Why this is an infrastructure problem, not a real-estate one
When compute and storage capacity can no longer simply expand inside one campus, three things happen to the network layer that supports it:
- Existing racks have to do more work. Operators upgrade access and spine layers from 1G/10G to 25G/100G to extract more throughput per rack unit before they can add a single new hall.
- Metro and regional links become critical path. Edge sites and secondary facilities need low-latency, high-capacity fiber back to the core: which is exactly what dark fiber and long-range optics are built for.
- Multi-vendor environments become the norm. Sites built at different times, by different teams, under different procurement cycles rarely run a single switch vendor end to end. Optics and cabling have to bridge that gap without forcing a rip-and-replace.
Amsterdam is the sharpest current example of this shift in our own market.
The result is that the network layer, not the building, becomes the bottleneck you can actually do something about this quarter. Getting planning permission for a new hall is a multi-year process outside any single team's control. Re-specifying your switching layer and your inter-site fiber links is not: it's procurement and installation work that can start now, on hardware that does not depend on a new building permit, against a footprint you already have or can lease.
What this actually looks like on a shopping list
Higher port density switching. If you're consolidating workloads into a smaller number of racks, the practical upgrade path is moving your access and top-of-rack layer to 25G/100G. Our Cisco Nexus 9000 range covers this directly: for example the Nexus N9K-C93180YC-EX, which puts 48x 10/25G SFP28 ports and 6x 100G QSFP28 uplinks in a single rack unit. That's the kind of density swap that lets you serve more compute out of the same footprint instead of waiting on a new hall.
The right optic for the distance. This is where most buying mistakes happen. A 100GBASE-SR4 optic is multimode and rated for short intra-rack or intra-row runs: it will not reach across a campus, let alone to an edge site. For that you need single-mode, long-reach optics such as 100GBASE-LR4 QSFP28, rated to 10 km over single-mode fiber: the class of optic that actually makes a metro or dark-fiber link back to a secondary site workable. Our 100G QSFP28 compatible modules category lists both reach classes side by side, along with Cisco-, HPE-, Huawei-, Juniper- and MikroTik-compatible variants, so a mixed-vendor edge site isn't a procurement bottleneck.
Practical advice before you spec anything
- Match the optic to the actual link distance, not the port type. SR4 for the rack, LR4 (or longer) for anything leaving the room.
- Check compatibility before committing to a single-vendor optics contract. In a moratorium-driven, multi-site environment, you will end up running switches from more than one vendor: third-party compatible optics avoid locking your edge expansion to one supplier's lead times.
- Refurbished and pre-owned enterprise gear is a legitimate way to add 25G/100G capacity fast. It's how a lot of the density upgrades above actually get deployed on a realistic timeline.
The bottom line
A construction moratorium doesn't stop growth: it moves it. If new floor space in Amsterdam isn't the answer, the answer is denser switching in the racks you already have and reliable long-reach connectivity to the sites that pick up the overflow. Browse our Nexus 9000 datacenter switches and 100G QSFP28 optics, or get in touch and we'll help you spec the right mix for your site.