es-datacentereuskadi DATA CENTER EUSKADI SOCIEDAD LIMITADA has to prove whether marketed data-centre capacity can survive power and carrier constraints

Summary

• The strongest operating evidence for es-datacentereuskadi DATA CENTER EUSKADI SOCIEDAD LIMITADA is not a slogan. The RIPE RDAP record for AS202765 identifies DATA CENTER EUSKADI SOCIEDAD LIMITADA as the registrant, and RIPE's current AS overview marks the ASN as announced on 12 July 2026.
• Current routing is visible but compact. RIPE announced-prefix data shows 185.163.164.0/23 plus the two component /24s; RIPE routing status counts 512 IPv4 addresses, no IPv6 space and one observed neighbour; CAIDA AS Rank reaches the same broad conclusion of one provider relationship and three prefixes.
• The facility evidence is stronger than the public routing evidence. Uptime Institute's client story says the Bizkaia data centre is being commissioned in two 1MW phases, that it uses telecom services from different companies, and that the completed Phase 1 received Tier IV Certification of Constructed Facility in May 2025. The Uptime Spain awards list separately lists Data Center Euskadi Bizkaia in Abanto-Zierbena with Tier IV design and constructed-facility certifications.
• The expansion claim remains a construction and commercialisation question. KREAN's February 2025 announcement places the planned Gipuzkoa building at Garaia Technology Park in Arrasate-Mondragon, with more than 5,000 square metres and up to 3MW of IT equipment in two rooms, equipped in phases as services are sold.
• The investment thesis is credible but not self-proving. Public and private shareholders, a certified Bizkaia phase and a visible ASN all support a real infrastructure effort; however, the public record still needs clearer evidence on power-feed independence, generator runtime, cooling failover, fibre entry diversity, actual customer failover and whether the marketed 2MW Bizkaia plus 3MW Gipuzkoa story has become usable capacity rather than staged potential.

The company is no longer just a plan

The first judgment should be plain: Data Center Euskadi is not an empty directory label. The entity has a Spanish corporate record, a public-private ownership story, a physical project in Bizkaia, a construction path in Gipuzkoa and a current autonomous-system number. A weak-footprint downgrade would be unfair if it meant “nothing can be verified.” Enough can be verified to place the company inside Spain's data-centre market and inside the Basque Country's policy effort around data locality, digital services and regional infrastructure.

The more useful judgment is different. The public record proves several layers of activity, but those layers do not all have the same operating value. A capital increase does not power a rack. A first-stone ceremony does not prove live customer failover. A Tier IV constructed-facility certification for a completed phase is meaningful evidence for the tested facility scope, but it is not a blanket certificate for every future room, every marketed megawatt, every carrier path and every customer service.

A current ASN proves public internet routing for 512 IPv4 addresses, but it does not prove that all cloud, platform or colocation services have independent external reachability under a carrier failure.

That distinction is the centre of this profile. The company can be real and still require stronger proof before public institutions, manufacturers, health organisations or financial firms treat the marketed capacity as a resilient regional dependency. The buyer's question is not whether Data Center Euskadi exists. It is which parts of the advertised capacity are installed, which are commissioned, which are sold, which are recoverable, and which still depend on future construction, power allocation or carrier turn-up.

Spanish registry evidence gives the legal base. The BOE text for BORME-A-2022-192-48 records DATA CENTER EUSKADI SOCIEDAD LIMITADA as incorporated with operations beginning on 13 September 2022, a Bilbao address at Plaza Pio Baroja 3, an initial capital of 600,000 euros, and an entity centred on centralised processing and management of data services and information-system infrastructure. That entity fits the data-centre thesis. It does not by itself identify racks, power contracts or customers, but it makes the company purpose explicit.

The ownership story also has public roots. The Parke announcement from September 2022 said Euskaltel, Teknei Group, Dominion and the Basque Government created Atlantic Data Infrastructure, with the first installation planned for the Abanto campus and later extensions elsewhere in the autonomous community. It described an initial ownership split in which Euskaltel, Dominion, Teknei and Sistemas y Tratamientos Automaticos each held 21.25%, with the Basque Government holding 15%. Later public material expands that shareholder circle. Uptime Institute says about a quarter of funding comes from public institutions, including Bizkaia, Gipuzkoa and the Basque Government, while private stakeholders include Dominion, Euskaltel, Gertek, Iberdrola, Kutxabank, Mondragon Corporacion Cooperativa and Teknei.

Capital records show the project did not remain at a symbolic 2022 scale. The BOE BORME notice for 3 January 2025 records a 4.8 million euro capital increase for Data Center Euskadi, bringing subscribed capital to 9.6 million euros. A later BOE BORME notice for 16 June 2026 records board changes and says the number of board members was expanded to nine. Corporate filings do not prove technical readiness, yet they do show an entity receiving additional capital and governance attention while the facility project moved toward certification and expansion.

The company therefore has a stronger base than many small hosting names that appear only in number-resource records. It has official legal identity, regional policy backing, recognised shareholders and physical construction evidence. The risk is not fictitiousness. The risk is the familiar data-centre gap between institutional ambition and independently verifiable service resilience.

Bizkaia is the verified anchor, but Phase 1 is the important boundary

The Bizkaia site is the anchor asset. Public sources place it at the Abanto campus of the Bizkaia Science and Technology Park, in the Ezkerraldea-Meatzaldea area outside Bilbao. The Data Center Map listing gives a site address at Biscay Science and Technology Park, Lot SZAE-15-B, and describes the facility as a two-module design with 1MW per module. The same listing says its own data still marked the site under construction, which is useful caution: market directories can lag real commissioning, but they also remind readers not to turn a planned maximum into a fully occupied live load.

Uptime's evidence carries more weight because it is tied to certification. Its client story says the Bizkaia data centre is being commissioned in two 1MW phases and designed for energy efficiency and Tier IV fault tolerance. More importantly, it says the completed Phase 1 received Tier IV Certification of Constructed Facility in May 2025. The Spain awards list independently records Data Center Euskadi Bizkaia in Abanto-Zierbena with both Tier IV Certification of Design Documents and Tier IV Certification of Constructed Facility.

That is real progress. Tier IV constructed-facility certification is not a marketing badge loosely attached to a future campus. It indicates that Uptime evaluated the constructed facility against its standard within the certified scope. For buyers, that makes the Bizkaia facility more credible than a project that has only design renderings, shareholder names or press releases.

But the boundary matters. Uptime's own description is phase-specific: completed Phase 1. The public story does not say that every planned room, future module, Gipuzkoa room or later expansion has the same constructed-facility certification. It does not publish the full test dossier, the exact IT load certified, the generator fuel runtime, the maintenance bypass results or the customer restoration performance. That is normal for public summaries, but it limits what an outside reader can conclude.

The public record also mixes several capacity units. Parke's 2022 announcement said each installation would consist of two 1MW modules commissioned progressively. DCD's 2023 report on the groundbreaking described a 2MW site and an initial investment of 25 million euros, potentially rising to 100 million euros with future deployments. Uptime describes two 1MW phases at Bizkaia. KREAN describes up to 3MW of IT equipment at the later Gipuzkoa building. These numbers are consistent as an expansion story, but they are not the same as measured usable capacity.

Installed capacity, commissioned capacity and sellable capacity diverge in any data centre. A building can be designed for 2MW and have only part of that load commissioned. It can have commissioned electrical and cooling plant but retain unsold halls. It can have sold customer cabinets but hold back headroom for redundancy, maintenance and growth. It can have physical power but lack enough carrier handoffs for the customers it hopes to attract. The public record points to a credible 2MW Bizkaia plan and a certified Phase 1, but it does not publish a current live-load figure or a customer-occupied megawatt figure.

That is why the title's phrase “marketed data-centre capacity” is precise. Marketed capacity is the number customers hear when a project is announced. Usable capacity is what remains after construction, utility service, switchgear, cooling, cross-connects, security, staffed operations, network handoffs, maintenance reserve and customer acceptance are all working together. Data Center Euskadi has moved a meaningful part of that path from announcement into certification. It still has to show, for each phase, how much has reached the last category.

The public internet edge is alive, but it is not diverse enough on its face

AS202765 is the best direct network evidence. The RDAP record identifies the name as es-datacentereuskadi, status active, with DATA CENTER EUSKADI SOCIEDAD LIMITADA as the registrant and a Bilbao address. It records registration on 17 December 2025 and a last change on 18 December 2025. The associated RDAP IP record for 185.163.164.0 identifies the 185.163.164.0 to 185.163.165.255 range as ES-DATACENTER-20160815, allocated PA, country Spain, also tied to the same organisation.

RIPE sees the ASN as active. The RIPE AS overview marks AS202765 announced. RIPE announced prefixes show three entries during the late June to 12 July 2026 window: 185.163.164.0/23, 185.163.164.0/24 and 185.163.165.0/24. RIPE prefix overview says the /23 is announced by AS202765 and has two related more-specific /24s. RIPE RPKI validation marks the /23 valid for origin AS202765.

The scale is modest. RIPE routing status reports three IPv4 prefixes, 512 IPv4 addresses and no IPv6 /48 equivalents. That is enough for an operator's public edge, management services, customer services, portals or a small allocation for hosted workloads. It is not enough to prove broad cloud address capacity or a large colocation customer base. A data centre can host customer equipment that uses customers' own IP space or carriers' IP space, so the small allocation does not cap the facility. It does, however, cap what can be inferred from Data Center Euskadi's own visible origin routes.

The more important finding is adjacency. RIPE's ASN-neighbours data shows one unique neighbour at the latest available snapshot: AS12338. RIPE routing consistency records three registered import and export peers: AS12479, AS3262 and AS12338. It marks AS12338 in BGP and the other two not in BGP for the snapshot. That means the registered policy is broader than the observed current topology.

The names are not obscure. AS12338 is Euskaltel. AS12479 is Orange Spain. AS3262 is SAREnet. The presence of Orange Spain and SAREnet in registered policy is useful because it shows a declared design or intended carrier menu. The absence of those two from BGP in the RIPE snapshot is just as important. A buyer should not treat all three as current active upstream diversity unless route collectors, contracts or technical evidence show they are live.

CAIDA AS Rank agrees with the narrow reading. It marks AS202765 as seen, assigns it a prefix cone of three and an address cone of 512, and records one provider relationship. BGP.Tools similarly describes the ASN as a small BGP network with one upstream carrier and three valid RPKI prefixes. Hurricane Electric's BGP Toolkit lists three originated IPv4 prefixes, no originated IPv6 prefixes and all three originated prefixes as RPKI valid. These independent views do not replace RIPE, but they make it harder to argue that public measurement is simply missing a large multi-carrier edge.

This does not mean the facility has only one physical carrier in the building. Uptime says the Bizkaia site uses telecom services from different companies. Data Center Map lists nearby carrier and data-centre assets around Bilbao, including Telxius, Sarenet, IREN and EXA facilities in the wider region. ADI may have carrier rooms, customer cross-connects or private links that do not appear as AS202765 origin adjacency. The public internet edge is only one layer of a data-centre connectivity stack.

Still, for a buyer whose service depends on Data Center Euskadi's routed edge, the public evidence supports a downgrade from “proven carrier diverse” to “active but not publicly proven diverse.” The settlement evidence would be straightforward: current looking-glass output or route data showing multiple active upstreams, physical route-diversity documentation, distinct meet-me rooms or entry points, maintenance records for carrier failover, and customer tests showing that route withdrawal or one-carrier loss does not break hosted services.

Gipuzkoa adds capacity ambition before it adds operating proof

The Gipuzkoa site is the second major capacity story. KREAN's project announcement says the new data-centre building will be located in the Garaia Technology Park in Arrasate-Mondragon, Gipuzkoa, with more than 5,000 square metres of floor space and up to 3MW of IT equipment in two rooms. It says those rooms will be equipped in phases depending on commercialisation. That last phrase is crucial: it frames 3MW as an upper build-out path rather than proof that 3MW is already carrying customer workloads.

Garaia Technology Park's December 2024 announcement said construction at the Garaia Technology Park began the project's second phase, while the first infrastructure in Abanto had 2MW capacity and was expected to begin operation in the first quarter of 2025. DCD's Spanish report on the KREAN award also described the new Garaia project as a 3MW data centre. Baxtel's ADI profile classifies ADI as having two facilities across two markets, with Abanto operating and Arrasate building.

This expansion logic is commercially sensible. A distributed Basque data-centre network could lower latency for regional institutions, create a local alternative to Madrid or foreign hosting, and give customers a more sovereign placement option. A two-site model could eventually support geographic redundancy, split workloads and local recovery, if the sites have independent power, network entry, operations teams and replication paths.

But public construction evidence does not yet establish that outcome. A future Gipuzkoa room cannot be used as a recovery site for a Bizkaia customer until it is built, powered, cooled, connected, secured, monitored, staffed, commissioned, contracted and tested. The public record does not yet show a Uptime constructed-facility certification for Gipuzkoa, live routes originated from that site, customer failover exercises between the two campuses, or published service-level terms describing inter-site recovery. It is reasonable to treat Gipuzkoa as expansion potential, not as current redundancy.

The market should also watch for phasing risk. KREAN says rooms will be equipped depending on commercialisation. That is normal and prudent: data halls are expensive, and overbuilding ahead of demand ties up capital. Yet phasing changes the risk profile. Customers cannot rely on a second room or second site merely because the building shell has capacity for it. They need to know what is equipped now, what is reserved, what has utility service, what has committed generators and cooling plant, and what is merely a designed path for later.

This matters because regional data-centre projects often sell two promises at once: local trust and scale. Local trust can be present early, through ownership, proximity and legal jurisdiction. Scale arrives later, through commissioned megawatts, carriers, staff and occupied halls. Data Center Euskadi's Gipuzkoa project strengthens the scale story, but the evidence should stay staged until commissioned facts catch up.

Public-private backing reduces some risks and introduces others

Data Center Euskadi's shareholder mix is a strength. Public participation, telecom-sector shareholders, an energy company, financial institutions and industrial groups create a broader base than a thinly capitalised single-owner host. They also create a market reason for the project: the Basque economy has manufacturers, finance, health, public administration, energy and technology firms that may want local hosting, data locality and reliable proximity services.

The Basque public-sector data-governance strategy PDF places Data Center Euskadi inside a wider public data strategy. It notes that in July 2022 the Basque Government authorised participation in Data Center Euskadi, S.L., whose entity was centralised processing and management of data services and information-system infrastructure for public or private entities, and describes the project as responding to rising demand for data hosting among administrations and the industrial and business fabric, especially small and medium enterprises. The BOPV decree page is the legal hook for that participation.

That context explains why the project matters. It is not merely another colocation room in northern Spain. It is part of a regional attempt to keep critical data processing closer to the institutions and companies that generate it, to improve sovereignty and to support local digital transformation. Ayesa's site-visit account echoes this policy framing, describing the Abanto facility as a large and advanced northern Spain data centre serving public and private sectors, with claimed benefits for proximity, security and regional investment.

But shareholder quality does not remove operational proof needs. A public-private vehicle can still face utility interconnection delays, construction inflation, staffing shortages, supplier delays, carrier concentration or a mismatch between demand and equipped capacity. Public institutions can bring patient capital and anchor demand, but they can also create governance complexity. Telecom shareholders can bring connectivity, but they can also leave the facility visibly dependent on one affiliated or regional carrier unless independent carriers are actively turned up and tested.

Energy-sector participation can help with knowledge and procurement, but it is not the same as published dual-feed proof.

The operating proof should be treated as its own layer. A customer should ask: who operates the site day to day; which plant is owned by Data Center Euskadi and which is under supplier contract; which carriers have installed equipment; whether customer cross-connects can avoid AS202765 entirely; how maintenance windows are communicated; and how governance decisions are made when public and private priorities diverge. Those are not hostile questions. They are the normal questions asked when a data centre becomes a dependency for hospitals, factories, banks, municipalities or software providers.

The ownership story therefore supports a medium evidence grade, not a strong one. It is credible and strategically coherent. It is not a substitute for power diagrams, carrier-diversity test records, incident postmortems, service inventories or customer recovery evidence.

The failure path starts with electricity and cooling, not with the ASN

For a data-centre operator, internet routing is only the public edge of a larger physical system. The first failure path is power. A regional utility outage can be brief and still be damaging if transfer systems, battery strings, generator start sequences or fuel contracts fail. Uptime's client story says ADI addressed the coordination of power, cooling and fire suppression under failover conditions and that resiliency was tested in a regional outage. That is an encouraging statement.

The public page, however, does not disclose outage date, affected utility area, duration, IT load, generator runtime, customer impact or whether any maintenance mode was active.

The second failure path is cooling. A 1MW room concentrates heat. Even if electrical supply remains available, cooling-system failure can force load shedding, emergency shutdown or hardware stress. Tier IV design and constructed-facility certification implies a high standard within scope, but buyers still need facility-specific details: cooling topology, concurrent-maintenance approach, water or refrigerant dependencies, alarm response, spare-parts inventory, seasonal test results and what happens during a heat wave or local water restriction. The public record does not provide those details.

The third failure path is fire detection and suppression. Uptime says ADI coordinated power, cooling and fire suppression under failover conditions, but the public story does not identify the specific suppression technology, detection zones, compartmentalisation or recovery process after an alarm. That omission does not indicate weakness. It simply means the outside reader cannot translate “certified facility” into a granular claim about how fast a customer recovers after a smoke event, false positive, isolation action or post-event inspection.

The fourth failure path is carrier-meet interruption. AS202765's current public view concentrates through AS12338. If a hosted service depends on the company's own routed address space and that handoff fails, public reachability is at risk unless another observed path is active. If customers bring their own carriers or private cross-connects, the risk shifts to the meet-me room, cross-connect management, building entry diversity and customer design. Public route data cannot show all of that, which is why the carrier question must be asked at the facility level as well as the ASN level.

The fifth failure path is construction delay. Bizkaia Phase 1 has moved past pure construction risk, but the second Bizkaia phase and Gipuzkoa are still capacity-expansion questions in the public record. Customers planning multi-year migrations need to know whether future halls are contractual commitments or target dates. A delay in Gipuzkoa could leave a customer with no local second site. A delay in a second Bizkaia module could constrain available cabinets, high-density deployments or growth. The public evidence makes this a live question because several sources describe progressive commissioning or phased equipment.

The sixth failure path is evidence drift. Public statements from 2022, 2023 and 2024 promised start dates and capacity targets. Some of those have moved forward into 2025 certification evidence; others remain plans. That is normal for infrastructure. It means old launch dates should not be recycled as current operating proof. The current public assessment should privilege dated 2025 certification and 2026 routing data over early project announcements.

Who is affected when this system fails

The affected audience is not limited to generic cloud customers. Uptime says ADI serves public institutions and private enterprises across finance, healthcare, government and manufacturing. Parke and the Basque Government data-governance strategy frame the project as regional infrastructure for administrations, industry and enterprises. Ayesa describes the facility as supporting advanced infrastructure, platform and application services for sectors including energy, manufacturing, healthcare and finance. These are high-dependency sectors.

That audience changes the standard of proof. A software test environment can tolerate a maintenance outage. A hospital appointment system, a manufacturing control support platform, a municipal service, an identity store, a finance application or a local backup repository may not. For these workloads, “Tier IV” is an opening question, not the end of diligence. The customer must still map its own application architecture onto the facility's failure domains.

If a manufacturer uses the facility for local backup, the issue is recovery time. Can data be restored if the primary site fails and the Bizkaia carrier path is degraded? If a public institution uses it for sovereign data hosting, the issue is jurisdiction plus availability. Does the service remain reachable during a regional cable incident or maintenance window? If a health provider uses it for sensitive workloads, the issue is not only secure storage but support escalation and audited recovery after a facility incident.

If a finance or energy customer uses colocation, the issue is whether power, cooling and carriers can sustain the customer's own resilience design.

The company can answer many of these questions privately. Public reporting cannot see customer contracts, maintenance test records or actual incident tickets. But the public article should make clear that customers are not protected by the existence of a data hall alone. Protection comes from fit between workload design and facility evidence.

The regional consequences also deserve attention. One attraction of Data Center Euskadi is proximity: services closer to Basque institutions can reduce latency and make data-governance arguments easier. The downside of regional concentration is correlated failure. If customers move several critical systems into one local facility without independent off-site recovery, a local power, cooling, fire or carrier event may concentrate risk rather than disperse it. A future Gipuzkoa site could help only when it is commissioned and used as an independent recovery location.

What the evidence says about advertised capacity

The advertised capacity story has three tiers. The first tier is corporate and project capacity: Data Center Euskadi was created for data-processing and system-infrastructure services, backed by public and private stakeholders, capitalised and governed as a serious project. This tier is well supported by BOE records, Parke, Uptime and regional sources.

The second tier is facility capacity: Bizkaia has a credible two-phase 1MW-per-phase plan, and Phase 1 achieved constructed-facility certification in May 2025 according to Uptime. This tier is stronger than an ordinary market rumour. It still needs scope precision. The public evidence should be read as “certified completed Phase 1 at Bizkaia, with a broader 2MW Bizkaia plan,” not as “all advertised capacity across the Basque network is operational and sold.”

The third tier is network capacity: AS202765 currently originates a small IPv4 allocation with valid RPKI and one observed neighbour. This tier is active but narrow. It proves a public routed edge; it does not prove the multi-carrier, meet-me-room and customer failover depth that a carrier-neutral data centre would need to demonstrate in a formal buyer review.

The fourth tier is expansion capacity: Gipuzkoa is publicly planned and under project development, with up to 3MW of IT equipment in two rooms, equipped in phases. This tier is commercially relevant but remains future-facing unless later public evidence shows commissioning, certification, network turn-up and customer use.

Together, those tiers justify a cautious positive grade. Data Center Euskadi is stronger than a name-only network and weaker than a fully transparent multi-site operator with published live load, carrier matrix and customer recovery evidence. The most defensible label is medium evidence with conditions.

What would settle the remaining questions

The first missing item is a current facility scope statement. It should state which Bizkaia rooms are built, commissioned, certified, sold and available; what IT load each room supports; how much load is reserved for redundancy; and what remains future capacity. This does not require revealing customer names. It requires aligning public capacity language with engineering state.

The second missing item is power evidence. A buyer should see utility-feed design, generator count and runtime, UPS topology, fuel arrangements, load-bank testing, maintenance-mode procedure and evidence that the certified topology remains intact after customer fit-outs. Uptime certification is a strong starting point; the next level is the operator's own current operating pack.

The third missing item is cooling and fire evidence. Customers should know how cooling is divided, how high-density cabinets are handled, how cooling alarms escalate, and how suppression and recovery work after a zone event. This is particularly important if the site markets advanced computing, dense cloud or high-performance workloads, because heat density can change the risk profile faster than floor area changes.

The fourth missing item is carrier evidence. Public BGP should eventually show more than one live observed neighbour if AS202765 itself is the customer-facing routed edge. If carrier-neutral colocation is the central service, the operator should instead provide a facility carrier list, route-diverse entry information and cross-connect procedures. The current public gap is not fatal; it is simply unresolved.

The fifth missing item is inter-site recovery evidence. When Gipuzkoa comes online, the key question will not be whether two buildings exist. It will be whether customers can actually use them as independent failure domains. That requires inter-site latency, replication options, operational separation, network diversity, coordinated maintenance windows and tested failover.

The sixth missing item is incident and outage transparency. Uptime says resiliency was tested during a regional outage. A public summary of what happened, what failed or did not fail, what load was present and what was improved would sharply increase confidence. It would also distinguish Data Center Euskadi from operators that rely on certification language but never show how real incidents are handled.

There is also a commercial proof point that sits between technical diagrams and customer names: the terms under which capacity is actually offered. A buyer does not need the operator to publish confidential contracts, but it does need clarity on reservation, activation and exit rights. If a cabinet, cage, cloud zone or platform service is sold from Bizkaia Phase 1, the service description should say which power class, cross-connect model, support window and recovery responsibility applies.

If a later Bizkaia module or the Gipuzkoa building is being sold ahead of completion, the contract should distinguish reserved future capacity from live service. That distinction prevents a regional hosting decision from turning into an unplanned construction bet.

This is especially important for customers that view the Basque location as a sovereignty or continuity advantage. Data locality is useful only when the service remains reachable and recoverable under stress. A local facility with weak external connectivity can become a beautiful single point of failure; a certified room without tested customer restoration can protect plant equipment while leaving applications unprepared; a second building can sound like geographic redundancy while lacking live replication, separate carriers or independent maintenance crews. None of these risks is unique to Data Center Euskadi.

They are the ordinary gaps that open whenever a serious infrastructure project moves from construction evidence to operational dependency.

The fair customer stance is therefore evidence-seeking rather than adversarial. Accept the public-private backing as a sign of strategic commitment. Accept the Uptime record as strong support for the certified Bizkaia scope. Accept AS202765 as proof that the company has a current public routing edge. Then ask for the next layer: current occupancy, live-load ranges, carrier handoff diagrams, failover test dates, maintenance-history summaries, backup and restore terms, support escalation names, and a clear statement of which claims apply to Bizkaia Phase 1, which apply to later Bizkaia capacity, and which remain Gipuzkoa plans.

If the operator can supply that material privately, the public evidence grade should rise. Until then, the public grade should remain disciplined.

The evidence grade

Data Center Euskadi earns a medium network-and-facility evidence grade. The reason is balanced: the company has a valid corporate record, a public-private shareholder base, a credible Bizkaia facility, Uptime Tier IV design and constructed-facility certifications for Data Center Euskadi Bizkaia, a live AS202765 route surface and a concrete Gipuzkoa expansion project. Those are not weak signals.

The grade is not strong because the public operating evidence still has gaps that matter directly to resilience. The live ASN surface is small, IPv4-only and visibly connected through one observed neighbour. Registered policy mentions more counterparties than BGP currently confirms. Public facility material does not publish current live load, dual-utility proof, generator runtime, exact certified scope beyond Phase 1, carrier entry diversity or customer failover results. Gipuzkoa is a future capacity story until it is commissioned and independently evidenced.

The practical conclusion is therefore clear. Data Center Euskadi should be treated as a real Basque data-centre operator with a credible certified anchor site, not as a speculative shell. It should also be treated as a provider whose marketed capacity still needs phase-by-phase proof before customers use it for critical regional workloads. The right due-diligence posture is neither dismissal nor trust-by-announcement. It is acceptance of the verified Bizkaia Phase 1 evidence, with hard questions on power, cooling, carrier diversity, live load and recovery before any broader 2MW-plus-3MW capacity story is treated as operational fact.