colozh NTS Colocation AG has to prove whether marketed data-centre capacity can survive power and carrier constraints

Summary

• colozh NTS Colocation AG should be treated as a verifiable Zurich infrastructure operator only to the extent that public evidence ties it to colozueri.ch, a Badenerstrasse 569 facility in Zurich, NTS Workspace AG operations, SwissIX access, PeeringDB facility records and AS15576 routing.
• The strongest evidence supports physical presence, carrier-neutral positioning, peering relevance and a marketed rack base of more than 400 racks; the weakest evidence concerns usable power per rack, generator endurance, fuel logistics, cooling reserve, maintenance outcomes and customer failover behavior.
• Public buyers should ask for current one-line electrical diagrams, recent load and temperature data, maintenance records, carrier meet-me diagrams, utility-feed evidence and incident history before treating the marketed capacity as proven production capacity.

Why this Zurich facility deserves scrutiny

colozh NTS Colocation AG sits in an awkward but important evidence band. It is not a name floating without geography. PeeringDB lists the facility as colozueri.ch Zurich, names NTS Colocation AG as the organization, gives the address as Badenerstrasse 569 in Zurich, and shows coordinates in the city. SwissIX separately lists colozueri.ch, Zurich at the same street address as an access location. NTS's own colocation pages describe Zurich and Bern data-centre sites and state that colozueri.ch was expanded in three phases to more than 400 racks on the NTS locations page. Those facts make the entity visible enough for infrastructure coverage.

The difficulty is that visible is not the same as fully verified. A data-centre article should not turn a rack count into available power, a carrier list into path diversity, or a sustainability claim into resilience. NTS says its owner-managed service company has carrier-neutral high-performance data centres in Bern and Zurich on its colocation overview. It also describes redundant UPS clusters, emergency diesel generators, redundant dedicated transformer stations, fire and smoke detection, Inergen fire suppression, redundant air-conditioning, biometric access control and video surveillance for colozueri.ch on the locations page. That is an unusually useful public equipment list for a small or mid-sized operator. Still, it leaves the core commercial question unanswered: how much of the advertised Zurich estate can be powered, cooled and connected during a compound stress event?

The practical answer matters because colocation contracts are sold as an alternative to operating customer infrastructure in offices, basements, branch rooms or single-provider hosting arrangements. If a customer leases a quarter rack, half rack, full rack, cage or private suite, the promise is not just space. It is a bundle of controlled environment, protected power, cross-connect access, remote hands and transit options. NTS makes that bundle explicit on its offer page, where rack size, passive pre-cabling, base-load power charging and renewable hydropower energy charging are part of the product. But that same public offer also shows why the asset must be tested through constraints. Power is charged against a base load per kilowatt that must cover peak current. Energy price depends on the supplier. Additional copper or fibre connections are available on request. Those words are reasonable for a colocation operator, yet they also imply that the usable shape of the facility is contractual and electrical, not just architectural.

The site therefore deserves a deliberately cautious reading. NTS appears to operate a genuine Zurich colocation facility with meaningful network relevance. It does not, in the public record, publish enough current electrical, thermal and operational detail to prove that every marketed unit of space can be sold and run at the density modern buyers may assume. The difference between "more than 400 racks" and "more than 400 racks at the power density, redundancy level and network diversity needed by each customer" is the central risk.

Physical location and operating boundary

The public location evidence is stronger than the public capacity evidence. PeeringDB, SwissIX, Data Center Map and Datacenters.com all place the Zurich site at Badenerstrasse 569, 8048 Zurich. The Data Center Map listing describes colozueri.ch as rackspace and colocation in Zurich, while Datacenters.com gives the same address and an external estimate of total and raised-floor area. Those third-party listings are useful signals, but they should not outrank the operator's own pages or live network records. They help confirm that the market recognizes a facility at that address; they do not prove current occupancy, available capacity or electrical margin.

The ownership and operating boundary is also layered. PeeringDB's facility record names NTS Colocation AG for the colozueri.ch facility, while NTS Workspace AG presents the customer-facing service portfolio. NTS Workspace AG's company page says the owner-managed company operates a fibre-optic backbone across Switzerland, is subject to a Federal Office of Communications notification requirement, and maintains carrier-neutral data centres in Bern and Zurich. Its imprint lists NTS Colocation AG as an investment. That does not by itself describe every legal agreement between the holding, service and facility entities, but it does show that the Zurich site is part of a wider NTS operating story rather than an isolated shell.

For customers, that boundary matters in four ways. First, the counterparty for physical colocation may differ from the network service provider that sells internet access, BGP support, transport or managed services. Second, the Zurich site is one location in a Bern-and-Zurich portfolio, so redundancy claims need to be separated by site. Third, the NTS Workspace AS15576 network can create real operational advantages for customers colocated in Zurich, but it can also create dependency if a buyer leans on a single local provider for space, transit and support.

Fourth, the best public certification signal appears to be tied to Bern rather than Zurich.

That last point is important. NTS's locations page states that Colobern North was built in 2018 and 2019 and is certified by the Uptime Institute Tier IV, and the Uptime Institute lists an NTS Colocation AG award entry. The public page that identifies the Zurich site does not clearly say colozueri.ch itself holds the same certification. NTS and its energy-storage partner describe the portfolio as Tier IV-certified in a broad way on the 2026 NTS news item, but a buyer should not let a Bern certification migrate silently to Zurich. The correct reading is narrower: NTS has at least one public Tier IV certification signal in the group, while the Zurich facility needs its own site-specific certification, design or operational evidence if a customer needs that assurance.

The physical facility also sits inside Zurich's urban infrastructure fabric. That brings advantages and constraints. Urban sites can be close to fibre paths, exchanges, enterprise customers and service staff. They can also face tighter expansion boundaries, utility limits, cooling constraints, neighbourhood sensitivity and permitting complexity than greenfield campuses. NTS says the Zurich site has already gone through three expansion phases.

That history is a positive sign of demand and execution, but it also raises the natural next question: after three expansions, is the next bottleneck room, power, cooling, cross-connect space, utility capacity or customer appetite?

Installed capacity is not the same as usable capacity

The easiest number in the public record is more than 400 racks. The harder number is how much dependable load those racks can carry. NTS's locations page says colozueri.ch was extended during three expansion phases to more than 400 racks. The offer page says full racks have 46 or 47 rack units, half racks have 20 or 21, and quarter racks have 11. It also states that power supply and air-conditioning are charged as a base load per kilowatt that must cover peak current.

That combination is enough to frame the risk: the commercial inventory is expressed in rack units, but the real capacity ceiling is expressed in kilowatts, cooling capacity and redundancy design.

This is not a criticism unique to NTS. It is a general colocation problem. A room can hold racks long before the electrical plant can support high-density customer use across every rack. In older or urban facilities, the spread between physical rack count and usable IT power can be especially important. A low-density customer with a few switches, storage appliances and lightly loaded servers may fit comfortably. A dense customer running modern compute hardware may exhaust a circuit, cooling zone or contract limit quickly.

When an operator publishes the rack count but not the site megawatts, available kilowatts per rack, average load, peak load, PUE trend, cooling reserve and utility service capacity, the prudent conclusion is not that capacity is absent. It is that capacity is not fully proven from public data.

NTS's information page adds a useful operational clue. In its discussion of power redundancy, NTS tells customers with redundant rack power supplies that the relevant power circuit cannot bear a load of more than half of the fused level, normally 16 amps. The technical point is straightforward: a redundant A/B rack design only works if either side can carry the surviving load when the other side is unavailable. If customers load both sides too heavily, redundancy becomes cosmetic. That warning is the kind of detail that serious operators publish because customer behavior can defeat facility design. It also shows why a buyer should ask for load rules, circuit monitoring, alarm thresholds and enforcement practice, not just for a generic statement that A and B power are available.

Cooling has the same translation problem. NTS says colozueri.ch has high-performance air-conditioning, redundant air-conditioning and a cold-corridor principle across its colocation sites. The NTS company page also highlights efficient data-centre operation through cold-corridor design and modern turbocor compressors. Those are credible design elements, but customers need current evidence of thermal reserve by room or pod. A cold corridor helps separate supply and return air; it does not make a hot spot impossible.

Redundant cooling helps if the redundancy is sized for the current load and maintained under realistic conditions; it does not prove that an overfilled high-density row will stay inside temperature limits during a chiller fault, pump issue or control failure.

The energy-storage announcement adds another layer. In February 2026, NTS said it had partnered with Inlyte Energy to deploy long-duration iron-sodium storage, starting with a 600 kWh installation targeted for the end of 2026 and a roadmap of up to 2 MW by 2028, beginning at the Bern data centre. That item is valuable because it shows NTS is thinking about power resilience beyond UPS and diesel. It is not proof that the Zurich site already has battery-backed long-duration resilience. The article says the first phase is in Bern and that broader deployment would depend on successful validation.

For colozueri.ch, the sensible reading is that storage may become a portfolio advantage, but it should not be counted as present Zurich resilience until site-specific deployment is documented.

The same distinction applies to renewable energy. NTS says its energy prices vary depending on the current supplier and are always 100% renewable or hydropower. The company page says NTS operates photovoltaic systems and participates in the Energy Agency of the Swiss Private Sector, whose public NTS entry is visible through EnAW. Renewable sourcing matters for emissions and customer reporting. It does not remove the need for grid capacity, transformer capacity, switchgear maintenance, generator testing, fuel logistics or cooling power. In a stress event, the greenest kilowatt is still unavailable if the electrical chain cannot deliver it to the rack.

Carrier density: a real strength with some documentation gaps

The network picture is the strongest part of the public case for colozueri.ch. PeeringDB lists the facility with 60 networks, six exchanges and one carrier in its facility record. SwissIX lists colozueri.ch among its access locations. NTS's locations page describes the Zurich site as one of the five main SwissIX sites. NTS's own carrier list for Zurich includes AEW Litecom, Arelion, Axpo Systems, BICS or Proximus Global, Cogent, Colt, ewz Telekom, Gas&Com, iWay, Lumen, NTS Workspace AG, Salt, Sunrise, Swisscom, TON, WWZ and Zayo. That is a serious roster for a Swiss urban colocation site.

The important nuance is that these are not all the same type of proof. An operator carrier list tells buyers who can be reached or ordered through the site, but it may include carriers available by fibre, through meet-me arrangements, through nearby access, through resale, or through a cross-connect order that depends on current feasibility. PeeringDB's carrier count of one is narrower and seems to reflect the formal PeeringDB carrier table, while its network and exchange counts show a much broader interconnection environment. A buyer should not collapse those records into a single number.

The right question is not whether there are "many" carriers. It is which carriers have physically diverse entrances, separate ducts, separate meet-me paths, independent power in the meet-me area and tested handoff procedures for the customer's actual service.

NTS Workspace AG's own network presence strengthens the facility's role. The NTS network page says NTS operates AS15576, is connected to the internet through multiple independent paths and has peerings through SwissIX to most Swiss providers. PeeringDB's AS15576 record classifies NTS Workspace AG as a regional network service provider, shows IPv4 and IPv6 support, estimates traffic at 20 to 50 Gbps, lists six internet exchanges and four facilities, and notes an RPKI route-origin validation policy. RIPE's aut-num record for AS15576 adds peering points at SwissIX, CIXP, AMS-IX, VIX, BIX and RomandIX. RIPEstat's routing-status view shows AS15576 visible across RIS peers and carrying observed neighbouring networks, while its announced-prefixes view shows many IPv4 and IPv6 announcements.

For colozueri.ch customers, that routing evidence means the NTS group is not just a landlord with a website. It operates a live network with public routing history, peering policy and Swiss internet-exchange participation. That makes the site more useful for customers who need BGP help, managed routers, transit, private network services or cross-connect access. The NTS internet access page explicitly says NTS offers internet connectivity from points of presence and data centres, bandwidths up to 100 Gbps, redundant connections via different paths and devices, BGP support, management of customer addresses and router rental.

But the same strength can become a dependency. If a customer buys space, transit, BGP management and operational support from the same provider, the provider can deliver a clean bundle. It can also become a concentration point. The buyer's resilience then depends on whether the customer has independent carrier contracts, independent routing policy, out-of-band management, equipment with two power feeds, cross-connects to more than one physical path and a plan for moving traffic during a provider incident. The NTS public record supports the idea that such designs are possible at colozueri.ch.

It does not prove that every customer has implemented them.

The meet-me question is especially important in a Zurich facility that appears in SwissIX and PeeringDB. Internet-exchange presence can reduce latency and transit cost, but it adds a specific failure surface. A route server issue, optical distribution fault, patching error, shared switch problem or maintenance event can affect many entities at once. NTS publishes a Peering Policy PDF and tells users technical information is available through its customer portal. Public buyers should ask for enough detail to separate exchange fabric risk from carrier risk and NTS backbone risk. Peering is not a substitute for private transit diversity; it is one part of a layered network design.

The main failure paths are ordinary, not exotic

The failure path that matters most for colozueri.ch is not an exotic catastrophe. It is a sequence of ordinary constraints arriving together. A utility feed trips or the local grid imposes a limitation. UPS absorbs the first interval. Generators start, but fuel, load balance, exhaust, acoustic constraints and maintenance status determine how long the facility can remain stable. At the same time, cooling plant runs under changed electrical conditions, and customer load may already be near peak.

If a carrier maintenance event, fibre cut or exchange issue lands in the same window, the site can remain physically powered while some customers still lose reachability.

NTS's public equipment list directly addresses pieces of that path: redundant UPS clusters, emergency diesel generators, dedicated transformer stations with a medium-voltage ring connection, fire and smoke detection, Inergen suppression, redundant cooling and access controls. Those claims are meaningful.

The missing public details are the ones that turn equipment into resilience: generator runtime at current site load, refuelling contracts, test frequency, last full-load generator run, battery autonomy, transformer ownership and utility-feed separation, switchgear maintenance history, thermal reserve, and how customer circuits are monitored during maintenance. Without those, the equipment list is a design hypothesis rather than complete operating proof.

Cooling failure is the second path. Data-centre outages are often less dramatic than a fire. A failed compressor, control sensor, chilled-water fault, blocked air path or underestimated high-density rack can push a room outside acceptable thermal bands. NTS's cold-corridor design and redundant air-conditioning language are relevant, but customers should request recent temperature and humidity reports by zone, alarm history, maintenance windows and the maximum supported load per rack under N, N+1 or 2N assumptions.

This is particularly important because colozueri.ch's public capacity number is expressed as racks after expansion, while the market increasingly thinks in power density.

Carrier-meet interruption is the third path. NTS's Zurich carrier list is broad. PeeringDB and SwissIX reinforce interconnection relevance. But a carrier list is not the same as route diversity. Multiple carriers can share ducts, building entrances, risers, meet-me rooms or maintenance contractors. A facility can advertise many reachable networks while a customer's actual service rides one physical cable path.

The practical resilience evidence is a site diagram showing separate entries, diverse paths to customer racks, clear demarcation, documented maintenance coordination and a test record showing customers can fail over between providers.

Construction delay is the fourth path. NTS says colozueri.ch expanded over three phases and says the Bern region has a new data centre under development around 2028 in the energy-storage news item. Expansion history is useful, but any future capacity plan remains exposed to power equipment lead times, utility approvals, civil works, cooling procurement, fire-safety approvals and commissioning. A customer considering a larger deployment should ask whether capacity is already live, reserved, under construction, dependent on a utility upgrade or dependent on another site in the NTS portfolio.

The word "capacity" should be split into space, power, cooling, cross-connect and operational support.

Fire and flood exposure round out the physical risk picture. NTS publishes fire and smoke detection plus automatic Inergen suppression for Zurich. Inergen is a standard gaseous suppression approach, useful where water damage to IT equipment must be avoided. It still requires detection, enclosure integrity, maintenance and trained response. The public record does not publish Zurich-specific flood, water-ingress or nearby-civil-works exposure. A facility on an urban street can face water ingress from municipal works, building systems, neighbouring construction or severe rain even if it is not on a dramatic floodplain.

Buyers should ask how low-level rooms, cable entries, generators, fuel systems and meet-me areas are protected.

Who is affected if the site fails

The affected population is broader than the legal entity's footprint. colozueri.ch is not a hyperscale region, but smaller carrier-neutral sites can concentrate local economic dependency. A quarter-rack customer may host a firewall, storage array, VoIP platform, monitoring server, e-commerce stack or regional application that supports a business far larger than the rack suggests. A network operator may use the site for SwissIX, transit or last-mile aggregation. A managed-services customer may rely on NTS staff for remote hands and router support. A local fibre provider may treat the site as a handoff point.

PeeringDB's facility record showing 60 networks and six exchanges is therefore not just a vanity statistic. It suggests that many network operators recognize colozueri.ch as a useful interconnection point. SwissIX access at the same address makes the facility part of the Swiss peering fabric. NTS's own AS15576 presence adds another dependency layer because customers may use NTS for both colocation and connectivity. If the site suffers a prolonged power, cooling or meet-me incident, the impact would be uneven. Customers with genuine dual-site, dual-carrier and dual-power designs may reroute or degrade gracefully.

Customers using a single rack, single carrier or single NTS-managed edge could see a much harder outage.

This is where the customer evidence gap becomes important. Public pages can describe redundant power and multiple paths, but they rarely show whether customers actually test failover. The most useful proof would be anonymized incident reports, maintenance statistics, successful generator transfer reports, cross-connect repair times, remote-hands response metrics and customer failover examples. NTS may have those records privately. They are not visible enough in the public record to support a broad claim that marketed capacity has been proven under stress.

NTS's support structure is a positive signal. Its website emphasizes long-term employees, monitoring, 24-hour support, customer portal information and remote hands. The team page shows named operational roles, including network and electrical expertise. That matters for a smaller operator because people and institutional memory often decide whether an incident is contained quickly. Yet people are not a replacement for published resilience data. A buyer can value NTS's local team and still ask for service records.

There is also a public-interest dimension. Data centres increasingly face questions about grid use, diesel backup, water, heat reuse and transparency. The International Energy Agency says on its data-centre energy page that data centres and data transmission networks are an important source of energy demand, while its electricity-sector analysis projects strong growth in data-centre electricity use. NTS's renewable-energy language, PV systems, waste-heat use and energy-storage pilot are therefore relevant. They should be examined as resilience and sustainability measures, not accepted as a blanket answer to power constraints.

European transparency rules are also changing the background against which Swiss operators are judged. Switzerland is not an EU member, but European customers and cross-border reporting expectations can still shape procurement. The European Commission's data-centre energy performance page describes monitoring and reporting of data-centre energy performance and water footprint for significant energy users in the EU. That matters because buyers are becoming accustomed to asking for PUE, water use, energy reuse, renewable energy factors and backup-fuel disclosure. NTS already publishes some sustainability claims; the next step is site-level data that lets customers compare Zurich against alternatives.

How unofficial listings should be used

Third-party data-centre listings are useful but limited. Data Center Map, Datacenters.com, Cloudscene, Inflect and similar directories help confirm market recognition, addresses, service categories and sometimes floor-area or connectivity estimates. They are not operating certificates. They can be stale, scraped, contributed by providers, simplified for procurement audiences, or inconsistent with the operator's current contracts. For colozueri.ch, these listings support the conclusion that the Zurich facility is externally recognized. They do not settle the power question.

The most useful unofficial listing for this case is PeeringDB because it is maintained for network interconnection and is regularly used by operators. Even there, interpretation has to be careful. PeeringDB can show that networks say they are present, that an exchange is associated with a facility, or that a carrier record exists. It cannot prove that a specific customer will receive a diverse, tested path. Its timestamps and self-reported elements matter. In this case, PeeringDB's facility record was updated in 2025, while its network and exchange relationships include older and newer updates.

That makes the facility alive in the interconnection record, but not fully audited.

Data Center Map says colozueri.ch offers rackspace and standard full racks, and Datacenters.com gives floor-area estimates. Cloudscene describes the facility in a way that suggests high technical equivalence. Those statements are useful as market signals, but they should be downgraded unless they align with current NTS documentation or direct customer due diligence.

The settlement evidence would be a current facility pack from NTS: site address, legal entity, service contract terms, floor plan, usable power, cooling design, carrier meet-me details, maintenance policy, sustainability metrics, incident reporting practice and proof of the customer's requested redundancy level.

The same rule should govern public announcements. The 2026 Inlyte partnership is significant because it speaks to a real problem: data-centre growth is putting pressure on power strategy, and NTS is exploring longer-duration energy storage. But announcements are not operations. The NTS article says the first Bern installation is targeted for the end of 2026 and the larger roadmap runs to 2028. Until installed and tested at a named site, it is a roadmap. The article is useful evidence that NTS leadership understands the backup-power gap between UPS, diesel and longer-duration storage. It is not evidence that colozueri.ch has solved that gap.

The evidence standard should therefore be staged. Public evidence proves existence, location, service scope, network relevance and a credible operator. Public evidence supports, but does not prove, the full usable capacity of the Zurich estate. Public evidence does not prove customer failover, generator endurance, fuel continuity, utility-feed diversity or cooling reserve at current load. That is not a negative verdict. It is the proper treatment of a thin but real public footprint.

What NTS needs to prove for the marketed capacity to be bankable

The first proof point is electrical. NTS should be able to show a current electrical single-line diagram to serious buyers, with utility feeds, transformers, switchgear, UPS blocks, generator connections, maintenance bypasses and rack distribution. The buyer should know whether the Zurich site has genuinely diverse utility feeds, redundant transformer capacity, ring connection details, generator N+1 or 2N status, battery autonomy, generator runtime at actual load and fuel replenishment commitments.

The public statement about redundant dedicated transformer stations with medium-voltage ring connection is a strong starting point; it needs operating evidence.

The second proof point is power allocation. More than 400 racks is impressive only if the operator can explain how power is allocated across them. Buyers should ask for total contracted IT load, available uncommitted load, average and peak utilization, per-rack density limits, circuit monitoring, breaker sizing and how NTS prevents customers from overloading A/B designs. The public warning about half of the fused level for redundant circuits is a good sign of operational discipline. It should be paired with monitoring records.

The third proof point is cooling. NTS should document cooling redundancy, design temperatures, alarm bands, maintenance history, cold-corridor containment, high-density accommodations and thermal reserve under a failed-unit scenario. Redundant air-conditioning is meaningful only if it is sized against current and contracted load. If the site has zones with different age, density or expansion history, the customer should see the zone that will host its racks, not just a site-level statement.

The fourth proof point is interconnection. NTS should provide a carrier and exchange map that separates on-net carriers, carriers reachable through cross-connect, carriers reachable through a third party, exchange fabric, NTS transit and private transport. Buyers should see which paths have diverse entrances and which share physical dependencies. For a customer using SwissIX, the design should separate peering from paid transit and out-of-band access. A serious customer will not treat a long carrier list as proof until it sees the demarcation.

The fifth proof point is incident behavior. Operators often hesitate to publish outage history, but serious colocation decisions require it. NTS can offer anonymized maintenance and incident summaries: generator transfer tests, UPS events, cooling alarms, cross-connect repair intervals, power maintenance outcomes and customer communication timelines. If there have been no material incidents, that itself can be stated with enough detail to be useful. If incidents occurred, the response quality matters as much as the fact of the event.

The sixth proof point is portfolio clarity. NTS has Bern, Zurich and network assets. Some resilience claims may belong to the portfolio, some to Bern, and some to Zurich. Customers should not have to infer which is which. The Uptime Institute signal belongs publicly to NTS Colocation AG and Colobern North in the visible NTS site context. The energy-storage pilot begins at Bern. Zurich has its own published equipment list and carrier environment. Keeping those facts separate is the difference between credible marketing and overreach.

The seventh proof point is contract scope. A colocation customer should know where NTS responsibility ends and where the customer's responsibility begins. Space, power, cross-connects, transit, managed routers, remote hands, backup services and monitoring can be sold as one integrated account, but each component fails differently. A customer whose firewall is hosted in the rack, whose BGP session is managed by the same provider and whose support contact is the same service desk has a different risk profile from a customer that brings independent carriers, independent routing policy and its own out-of-band access.

The site evidence should therefore be translated into a responsibility matrix: who owns each cable, device, circuit, ticket queue, maintenance approval and recovery step. That matrix is not administrative paperwork. It is the document that decides whether a real incident is a facility event, a carrier event, a managed-service event or a customer configuration problem, and it determines how quickly the next action can happen.

Bottom line

colozh NTS Colocation AG is best viewed as a real but under-disclosed Zurich infrastructure asset. The public evidence supports a serious facility at Badenerstrasse 569, a marketed colozueri.ch service, a rack base said to exceed 400 racks, a meaningful carrier list, SwissIX access and NTS Workspace AG network operations through AS15576. That is enough to justify attention from customers who need Swiss colocation, peering and managed network services.

It is not enough to treat every advertised rack as bankable production capacity. The missing public evidence sits exactly where data-centre failures usually happen: power delivery, generator endurance, fuel logistics, cooling reserve, carrier path diversity, maintenance behavior and customer failover. NTS may be able to prove those points privately. The public record does not yet do so.

The fair conclusion is therefore neither dismissal nor uncritical acceptance. colozueri.ch appears to be a genuine Zurich colocation node with more public network evidence than many small facilities. Its risk is that the most important resilience claims remain broad. Before a buyer places critical systems there, the buyer should translate every headline into an operating test: more than 400 racks into kilowatts and thermal reserve; carrier neutrality into diverse physical paths; redundant power into tested A/B survival; sustainability into site-level metrics; and portfolio announcements into Zurich-specific evidence.

Only then can the marketed capacity be judged as capacity that will survive the constraints around power, cooling and carriers.