How we compare the two operators
This is not a league table. Digital Realty and NorthC are both good operators with an Amsterdam footprint, and the right answer is nearly always driven by your workload and your timeline rather than by the logo on the building.
Digital Realty bought Interxion in March 2020, so datacenters that were Interxion’s are now Digital Realty’s.
We place racks with both, and we name them as examples, never as an exclusive list. We are an advisor, not a reseller, so no margin of ours rides on which one you pick. What follows is the set of questions we ask on a site call, and what the answers tend to mean.
Power ceilings: ask about the hall, not the building
Marketing talks about a campus. Engineering happens in a hall. Inside one building you can find legacy space designed for 3 to 5 kW a rack and newer space that will take far more, and the difference is not always visible on a tour.
- What is the design density of this specific hall, per rack, not averaged across the floor.
- Is that a hard cap or a cooling-limited soft cap, and what happens if I exceed it for an hour.
- Can I go above the design density in a contained aisle, and at what price.
- How is power delivered: A+B feeds or a single feed with UPS, three-phase or single-phase.
- If I step from 5 kW to 15 kW in month 18, does that mean a different hall and a different contract.
Carrier mix and interconnect
Amsterdam’s advantage is that most of the large colocation buildings in the metro have AMS-IX or NL-ix in-house, one cross-connect away. AMS-IX, NL-ix and Nikhef are the reference points. What varies is how many carriers are physically lit in your building and what a connect out of the building costs you.
- How many carriers are lit here today, and how many of those are ones I want.
- Is my IX port in-building or over a metro wave, and what is the monthly for each.
- What does a cross-connect cost, and can I have a bundle instead of a unit price.
- Who owns the meet-me room, and is there a second one.
- Lead time for a new carrier to enter the building: weeks or months.
Liquid cooling
Every operator in this market now has a liquid cooling answer, and the answers differ in maturity. Rear-door heat exchangers are the common ground. Direct-to-chip is increasingly available in newer halls. Immersion is still the exception and usually a project rather than a product.
- Which technique is available in the hall I would take, today, with a customer already running in it.
- Who owns the CDU and the loop, you or me, and who maintains it.
- What water temperature and approach temperature, and does my vendor’s kit accept it.
- What is the uplift per kW versus air, and is it a one-off or monthly.
- If the loop fails, what is the fallback and how long do I have.
Availability and lead time
Amsterdam has been supply-constrained for years and that has not fully unwound. Available capacity moves week to week, and the honest answer from any operator is a date, not a yes.
Ask for the contracted-ready date and the ready-for-service date separately. Ask what is already built and what is still a phase on a drawing. Ask whether the kW you want is held under someone else’s option that expires in six weeks. That shape is common, and it matters whether you are first or second in the queue.
From our own bookings: if you need racks live inside eight weeks the shortlist shrinks fast and the price improves less. If you can wait a quarter you get both more choice and more room.
Contract style
Broadly, larger global operators run a more standardised paper trail: their master agreement, their schedules, their indexation clause, with room to move on commercial terms rather than on the form. Regional operators are more often flexible on the shape of a deal (ramps, sub-metering, mixed density) and correspondingly more likely to want a longer term in exchange.
Neither is better. It matters because it decides what you can change. If your blocker is a clause, ask early whether that clause moves. If your blocker is the price, a longer term and a denser footprint move it in both houses.
Which campus for which workload
Both operators are present across the Amsterdam metro, so the campus question is usually more useful than the brand question.
- Science Park
- Research and academic gravity, dense carrier and exchange presence. The default for interconnect-heavy workloads and anything that wants to sit near AMS-IX and Nikhef.
- Schiphol-Rijk
- The airport-side cluster: strong international connectivity and easy road access. Often the answer for larger footprints and for teams flying in to rack their own kit.
- Amsterdam-Zuidoost
- An established multi-operator area with newer capacity. Frequently where the room is when Science Park is tight, and simple to reach by metro for hands-on work.
How we would choose
Site-level detail lives on the Digital Realty Amsterdam page and the NorthC Amsterdam page. Tell us the workload and we will shortlist two or three halls, brief them properly and come back with quotes you can compare line by line.
- GPU, or anything above roughly 15 kW a rack: cooling availability decides first, lead time second.
- Hosting, CDN, anything egress-heavy: carrier mix and cross-connect economics decide.
- Disaster recovery: distance from your primary site and contract flexibility decide, not price.
- First footprint outside your home country: ask who does the hands, because you will not be there.
