CPS Internetworking BV and the margin risk below cloud scale

Summary

• CPS Internetworking BV is best understood from public evidence as a small Dutch RIPE NCC member and number-resource holder, not as a proven scaled access network, cloud platform or national connectivity challenger.
• The strongest economic signal is resource monetisation: a RIPE-allocated /22 IPv4 block and /29 IPv6 allocation, recent sub-allocation of both IPv4 /23 halves to an IPXO-linked counterparty, and live routing through larger network operators. That can create revenue, but it also leaves CPS exposed to buyer power, lease-rate pressure, compliance overhead and weak differentiation.

Management's relevance problem below cloud scale

Management's incentive is clear even though management's financial disclosure is not. A small Internet infrastructure company that holds address space has to remain relevant in a market where customers can buy connectivity from incumbent telecom operators, cloud capacity from hyperscalers, hosting from large data-centre platforms and IPv4 access from specialised brokers. The practical question is not whether CPS Internetworking BV possesses something scarce. It does.

The question is whether the company can attach that scarcity to a customer problem for which buyers will pay enough to cover fixed overhead, abuse handling, registry costs, counterparty risk and the opportunity cost of selling or leasing the same resource elsewhere.

The public evidence makes this a resource-economics story before it is a service-growth story. RIPE NCC lists CPS Internetworking BV at Amberlint 30, 1906KK Limmen, Netherlands, with a RIPE contact email at cps-i.nl and an area serviced of NL. RIPE's database identifies the organisation as ORG-CIB23-RIPE, a Local Internet Registry with Dutch country code, Chamber of Commerce registration number 37073047 and a maintainer created in September 2019. The GLEIF LEI record independently confirms the legal name CPS Internetworking B.V., the same Limmen address, active status, Dutch jurisdiction, registration authority and registration number.

That is enough to establish the legal and registry boundary.

It is not enough to establish pricing power. No accessible company material found in this pass sets out a product catalogue, customer base, audited revenue, margin, network footprint, colocation estate, managed-services book or contractual term profile. Creditsafe classifies the business in computer consultancy and computer facilities management activities and says it was incorporated in 1995. Drimble also identifies the company at the Limmen address and shows the cps-i.nl website, but neither source supplies customer economics. The company's own web domain timed out during collection rather than supplying usable public service detail.

The absence matters because an infrastructure company below cloud scale can look asset-rich while still having little control over the economics of demand.

The managerial problem is therefore one of conversion. CPS can be relevant if its resources support a contracted use case: SME service continuity, static addressing, delegated address management, specialist hosting, network migration, routed business connectivity or a reliable address-leasing relationship with enforceable terms and clean-abuse processes. CPS is less valuable if it is simply a passive holder whose IPv4 space is routed or marketed by stronger counterparties. In the latter case the company owns a useful input, but the customer relationship, compliance process, route visibility and pricing discovery sit elsewhere.

That distinction shapes the whole assessment. Revenue growth would be easy to describe if the public record showed a growing number of customers paying CPS for service. It does not. Value creation would be easier to believe if the record showed differentiated SLA commitments, owned interconnection, a local access network, a sector-specific customer niche or evidence that CPS reduces switching costs for Dutch businesses. It does not.

What the evidence does show is narrower: a small Dutch legal entity, RIPE membership, a scarce IPv4 allocation, a broad IPv6 allocation, current routing observations and third-party signals consistent with address monetisation through larger network or leasing structures.

The verified boundary is a Dutch LIR, not a full-stack carrier

CPS's verified operating boundary starts with the RIPE member record. The public RIPE member page lists CPS Internetworking BV with a Limmen address, phone number, cps-i.nl RIPE email and Netherlands service area. The RIPE database goes further: ORG-CIB23-RIPE is recorded as an LIR, with country NL, registration number 37073047, admin and technical contact SB27035-RIPE, abuse contact AR55286-RIPE and CPS's own maintainer. In RIPE terminology, that makes CPS a registry-facing resource holder. It does not automatically make the company a retail Internet access provider, transit operator or cloud provider.

That boundary is important because public telecom language often blurs three activities that have different economics. A company can hold resources without selling broadband. It can route resources through another autonomous system without controlling a physical access network. It can lease or sub-allocate address space without operating the end-user service that ultimately consumes the addresses. Each activity can be legitimate. Each has different margin, risk and bargaining power.

The legal record is also modest. The GLEIF record shows an active Dutch private limited company, legal form 54M6, registered with the Netherlands Chamber of Commerce as 37073047, with headquarters and legal address at Amberlint 30 in Limmen. Bloomberg's LEI page repeats the same LEI, address, jurisdiction, legal form and active status, with an LEI registration that was last updated in July 2025 and due for renewal in July 2026. Those are identity facts, not growth facts. They confirm the company exists and is active, not that it has large operating scale.

The long history signal is mixed. Creditsafe says CPS Internetworking B.V. was incorporated in 1995 and operates within computer consultancy and computer facilities management activities. Drimble gives an August 1995 founding date and characterises the legal form as a private company. A company can survive for decades with a lean consulting or infrastructure book; longevity can indicate resilience. But in telecom economics, longevity alone does not prove a moat. A business with one or two durable contracts can persist for years and still have limited negotiating leverage if customers, upstream networks or address brokers can replace it.

This is why the article treats CPS as a company, not as the address block itself. The address block is evidence of a resource position. The company is the legal and managerial actor that decides whether to operate, lease, sell, delegate or reserve those resources. The economic thesis has to stay with the company. If CPS has a managed-network service book around Dutch SMEs, a local hosting cluster or a set of recurring consulting contracts that require its address space, the public evidence has not surfaced it. If CPS is primarily monetising IPv4 scarcity, the public record is more consistent with that interpretation.

What the address record says about operating leverage

The RIPE inverse lookup for ORG-CIB23-RIPE returns two primary resources tied to CPS: IPv4 range 45.149.168.0 through 45.149.171.255, corresponding to 45.149.168.0/22, and IPv6 allocation 2a0f:6600::/29. Both are under the netname NL-CPSINTERNETWORKING-20190905. The IPv4 allocation was created on 5 September 2019 and last modified in December 2024. The IPv6 allocation was created the same day in September 2019. That timing matters: a /22 allocation in the RIPE region immediately before the late-2019 exhaustion point is not a trivial resource for a small company.

A /22 contains 1,024 IPv4 addresses. In pre-exhaustion Internet economics, a block of that size might have been operationally ordinary. In 2026, it is a scarce asset, especially for operators that need routable IPv4 for hosting, VPNs, customer premises equipment, firewalls, legacy applications, mail reputation separation or migration away from older allocations. The scarcity is visible in RIPE policy and market evidence. RIPE NCC says its remaining IPv4 pool was exhausted in November 2019, and current waiting-list rules limit eligible LIRs that have not previously received IPv4 to one /24 allocation of 256 addresses.

CPS's /22 therefore represents four times the current single-allocation waiting-list size.

That scarcity creates operating leverage only if CPS controls the route to paying demand. The resource could support CPS's own services. It could be leased through a platform. It could be sub-allocated to another infrastructure company. It could be sold, subject to transfer policy and business objectives. It could remain partly reserved for future operations. The public record points away from a simple story of CPS directly using every address in its own network. RIPE records show both IPv4 halves as sub-allocated PA to a counterparty identified as ORG-IL687-RIPE, Internet Utilities Europe and Asia Limited, under the IPXO netname.

The first /23 covers 45.149.168.0 through 45.149.169.255 and the second covers 45.149.170.0 through 45.149.171.255. Both sub-allocation records were created on 10 December 2024 and modified on 23 April 2026.

That is the most material business-model clue in the file. It suggests that the monetisable IPv4 resource is not merely sitting idle. It also suggests that the address economics may be intermediated. Internet Utilities Europe and Asia Limited appears in RIPE as a UK LIR with an IPXO-linked role and maintainer. Third-party IP intelligence pages similarly label the relevant /23s with IPXO, Internet Utilities or data-centre/transit usage. Those signals do not prove the commercial contract between CPS and any counterparty.

They do, however, support a cautious thesis: CPS's near-term value may come more from being a resource holder plugged into a leasing/sub-allocation chain than from being an independently branded Dutch service provider.

That model can be attractive because it reduces sales and engineering burden. A small holder can convert addresses into recurring income without building a full customer-acquisition engine. But it also reduces differentiation. If a lessee or platform mainly cares about clean routable IPv4, it can compare CPS's block with many other blocks. The resource remains scarce, but the seller of a standardised resource faces price discovery. That is a different business from selling a managed service deeply embedded in a customer's operations.

IPv4 is the scarce asset, but scarcity is not a moat

The best case for CPS begins with the IPv4 market. RIPE's exhaustion notice and waiting-list rules make it difficult for new RIPE-region operators to obtain meaningful fresh IPv4 allocations. IPv4.Global auction data in July 2026 showed RIPE /22 transactions at about $16.25 per address on recent auction records, while broader broker and marketplace commentary shows meaningful price variation by block size, region, cleanliness and demand.

IPXO's market material describes an average IPv4 lease price in the range of roughly $0.34 to $0.40 per IP per month across recent 2025-2026 months, with a May 2026 reference around $0.36 per IP per month and campaign pricing lower than the standard rate.

Applied mechanically to 1,024 addresses, those external price references frame the opportunity without proving CPS's actual revenue. A sale at $16.25 per address would imply about $16,640 gross value before costs, taxes, negotiation, transfer eligibility and any block-specific discount or premium. A lease at $0.36 per address per month would imply about $369 per month if every address were leased at that average. Higher small-block rates could improve that figure; broker fees, dirty-address discounts, unused space, abuse handling, or weaker terms could reduce it. This is not a valuation of CPS.

It is a sanity check on the scale of the address opportunity relative to fixed costs.

RIPE's 2026 charging scheme sets an annual fee of EUR 1,800 per LIR account, plus charges for independent Internet number resource assignments and ASN assignments where applicable. The annual membership cost alone is not crippling, but it is not zero. For a small company, registry fees, domain operations, route management, accounting, legal compliance, abuse response and time spent managing counterparties all eat into the address yield. The margin question is therefore sharper than the headline scarcity story suggests.

If the IPv4 block produces only low hundreds of euros per month after platform costs and operational time, the asset is useful but not transformative. If it is tied to a broader managed-service relationship or a high-quality multi-year lease, it can support a stronger small-company economics case.

The sell-versus-hold decision is also non-obvious. Selling converts a scarce asset into immediate cash but removes future optionality. Leasing preserves ownership but creates ongoing operational and reputational exposure. Holding unused space protects future strategic flexibility but carries opportunity cost in a market where buyers and lessees still pay for IPv4. A management team below cloud scale has to decide whether it is a network operator, a service integrator, an address landlord or a seller of scarce resources. The public evidence does not reveal that decision directly.

The sub-allocation records imply that CPS has at least explored, or allowed, third-party operational use of the IPv4 space.

Scarcity also attracts substitutes. Customers needing IPv4 can lease from many platforms, buy on the transfer market, use carrier-grade NAT, shift parts of workloads to IPv6, place services behind cloud load balancers or buy bundled hosting from larger providers. Some of those substitutes are imperfect. Many legacy systems still require IPv4, and clean, stable addresses remain valuable. But the existence of substitutes limits how much a small address holder can charge unless the resource comes packaged with reliability, reputation, geographic fit or a service relationship that reduces operational friction for the customer.

Routing evidence points to outsourced reach, not standalone scale

The routing evidence is more nuanced than the allocation evidence. RIPE has route objects for 45.149.168.0/22 and 2a0f:6600::/29 with origin AS44928. RIPE's AS44928 record identifies the autonomous system as Trixit, assigned to Trixit Holding B.V., with import/export references to AS34968 and AS6461 and maintainer TRIXIT-HOLDING-MNT. BGP.tools and Hurricane Electric's BGP view describe AS44928 as a small IPv6-only network at the time of collection, originating three IPv6 prefixes and no IPv4 prefixes, with 2a0f:6600::/29 labelled as CPS Internetworking BV.

RIPEstat provides the current operational distinction. For CPS's whole IPv4 /22, RIPEstat shows the AS44928 origin first seen in July 2024 and last seen in December 2024, but no current full-feed RIS visibility for the /22 as a whole. Instead, the current more-specific IPv4 routes are the two /23s, each originated by AS2856. RIPEstat reports 45.149.168.0/23 and 45.149.170.0/23 as visible to all 327 IPv4 RIS peers at the query time, with origin AS2856 and RADb route objects. RIPE records identify AS2856 as BT-UK-AS, British Telecommunications Limited, a large UK network.

PeeringDB lists BTnet, AS2856, as a network service provider with regional geographic scope, large prefix counts and traffic levels in the 1-5 Tbps range.

For IPv6, RIPEstat shows 2a0f:6600::/29 originated by AS44928 and visible to 322 IPv6 RIS peers. That means CPS's IPv6 allocation has live visibility through the Trixit-linked AS44928 route. The IPv4 position, however, is currently visible through BT-originated more specifics rather than a CPS-controlled autonomous system. RIPEstat RPKI validation shows both AS2856 /23 IPv4 routes as valid, while the AS44928 route for the CPS IPv6 /29 is unknown in the RPKI validation call.

The IPv4 routes therefore look operationally accepted and clean in current RPKI terms, but they reinforce the point that CPS is not visibly the routing operator for those IPv4 halves.

This matters for supplier dependence. If CPS relies on BT-originated routing for the IPv4 halves, or if the IPXO/Internet Utilities sub-allocation chain is responsible for routing arrangements, CPS may earn value from the resource while other parties control reachability, customer provisioning, abuse response and reputation management. That can be efficient. It can also make CPS a price-taker. The party that owns the end-customer relationship and the party that controls route visibility often have leverage over the resource holder, especially when the resource is one small /22 among many available address blocks.

The AS44928 link also deserves restraint. AS44928 belongs to Trixit Holding B.V., not CPS. The RIPE route object for CPS's /22 was maintained by TRIXIT-HOLDING-MNT, and the route6 object for CPS's IPv6 /29 was maintained by CPS's maintainer. BGP.tools describes the network as Trixit Holding B.V. and lists the CPS IPv6 prefix among its originated IPv6 routes. That is evidence of routing relationship, not ownership by CPS of Trixit or Trixit by CPS. It indicates outsourced or partnered network reach, which is common in small infrastructure operations. It does not prove a consolidated network business.

The business model depends on demand that is not yet public

The viable business models for CPS fall into four buckets. The first is direct managed connectivity or hosting for a small base of Dutch customers. The second is consultancy and network administration, consistent with the company-classification record. The third is resource leasing or sub-allocation, supported by the current IPv4 records. The fourth is long-term resource optionality, where the company preserves scarce IPv4 and uses or disposes of it when economics improve. The public evidence supports the third and fourth buckets more strongly than the first.

Direct service demand would usually leave public traces: a website with business connectivity products, service descriptions, customer testimonials, job postings for network engineers, PeeringDB entries under the company's own name, references to data-centre locations, procurement awards, regulatory listings, service-status pages or customer-facing documentation. In this pass, those traces were not found. That does not mean they do not exist. Small Dutch B2B infrastructure companies can operate through relationships, referrals and legacy contracts with little public marketing.

But the absence prevents a confident claim that CPS has differentiated end-user demand.

Resource demand is easier to support. The sub-allocation to IPXO-linked Internet Utilities and the current AS2856 routing of both IPv4 halves show that the space is not dormant in the global routing system. IP2Location classifies at least one address in the block as data-centre/web-hosting/transit usage, with ISP field CPS Internetworking BV, domain cps-i.nl and proxy ASN AS2856. Ping0's history shows AS2856 as the current ASN for the 45.149.168.0/23 half, earlier appearances under Orange and IDC IPXO, and the broader 45.149.168.0/22 registration history under CPS.

BrowserScan shows the 45.149.168.0/23 segment with IPXO identifiers and the RIPE sub-allocation details. These are not primary commercial records, but they are useful market signals. They point to address-market use rather than a simple owner-operated access network.

The risk is concentration. A /22 split into two /23s is too small to support a diversified platform if the only monetisation is address leasing. If both halves depend on a single intermediary or narrow counterparty group, CPS's revenue can be vulnerable to one renewal, one abuse event, one reputation issue or one change in platform policy. A buyer or lessee can replace a /23 more easily than a bespoke managed service embedded in customer operations. For the holder, the smaller the resource base, the more important contract durability becomes.

The missing facts are therefore not cosmetic. Customer count, lease duration, payment terms, rights to withdraw the space, abuse indemnity, escrow or deposit arrangements, whether CPS retains direct contact with end users, and whether addresses are dedicated to one use case would change the economic assessment. Without those facts, the prudent conclusion is that demand exists for the resource but not yet that CPS captures differentiated margin from that demand.

Unit economics are constrained by fees, lease pricing and abuse costs

The unit-economics frame is simple. CPS has a scarce IPv4 input and modest fixed registry costs. The company can earn value if gross proceeds from operating or leasing the resource exceed registry fees, routing/support costs, administrative time, legal/compliance costs, and the risk-adjusted cost of reputation loss. The gross line is bounded by public market prices; the cost line is bounded by operational obligations that do not disappear because the company is small.

RIPE's 2026 fee structure gives a visible floor. An LIR account costs EUR 1,800 annually. New members pay sign-up fees, though CPS is already an existing member. There may be additional charges for independent resources or ASNs depending on the specific resource mix. On a 1,024-address IPv4 block, EUR 1,800 is not a large per-address burden if every address is monetised well. It is a more significant burden if the address block is only partly monetised, if payments are low, or if management time is expensive.

Lease pricing gives the second boundary. IPXO's public material places recent average lease pricing around the mid-30-cent-per-IP-per-month level, while campaign and average-rate pages show standard average rates around $0.38 per IP per month and promotional rates lower. A fully leased /22 at $0.36 per address per month is a small-revenue asset, not a cloud-scale platform. It can cover RIPE fees and some overhead, but it does not obviously support a large engineering organisation. It may, however, be attractive for a lean company if the arrangement is low-touch and payments are reliable.

Abuse handling is the hidden cost. IP addresses used for hosting, VPN, proxy, security-sensitive workloads or high-churn customers can attract spam, scanning, fraud complaints or reputation degradation. A resource holder that delegates addresses still has to care about who uses them, how complaints are routed and how quickly problems are remediated. If CPS is insulated contractually by a platform with strong abuse processes, the risk is manageable. If not, a small holder can find that a few bad users consume disproportionate time and damage block reputation.

The sale alternative is also visible. IPv4.Global auction data showed recent RIPE /22 sales around $16.25 per address in early July 2026. A sale at that level is not enough to fund a large strategic pivot, but it can represent meaningful liquidity for a small private company. Holding the block is rational if management expects steady income, higher future prices, strategic need for the space, or optionality. Selling is rational if operational hassles, buyer power or regulatory overhead exceed the present value of leasing or self-use.

The public record does not show which calculation CPS has made; the sub-allocation evidence suggests management has not treated the resource as purely idle.

Suppliers and counterparties carry more power than the resource holder

Supplier concentration is not only about transit bills. In this case it is about who provides route visibility, who provides demand, who manages address reputation and who can substitute the resource. The current IPv4 evidence points to a chain in which CPS owns the allocated parent /22, Internet Utilities/IPXO is named on the sub-allocated /23s, and BT's AS2856 originates both /23s globally. Each link can be rational. Each link also dilutes CPS's control over the customer-facing economics.

BT is the strongest visible network counterparty in the IPv4 path. RIPE records identify AS2856 as BT-UK-AS and PeeringDB shows BTnet as a large regional network service provider with heavy traffic scale. A small Dutch resource holder does not have symmetric bargaining power with that kind of network. If BT is simply the upstream for a lessee, CPS may have no direct commercial exposure to BT. But the route record still shows that current reachability depends on an operator with far greater scale and many alternative customer relationships.

Trixit Holding B.V. is the visible AS44928 counterparty for the route objects and IPv6 origination. RIPE identifies Trixit as a Dutch LIR with AS44928 assigned since 2015. AS44928's import/export records mention AS34968 and AS6461, and public routing pages describe it as a small IPv6-only network with one upstream and one peer at the time of observation. That kind of specialist routing arrangement can be efficient for CPS, particularly if CPS does not want to run its own autonomous system. But it again means that network operations are not fully internalised.

Internet Utilities Europe and Asia Limited, represented by ORG-IL687-RIPE, is the named organisation for both current IPv4 /23 sub-allocations. The role and maintainer naming are consistent with IPXO/netutils infrastructure. If CPS's economics depend on this relationship, the relevant supplier is not simply a technical upstream. It is a market intermediary that can set or influence terms, handle lessee demand, perform abuse management and compare CPS's block against many other blocks. The resource holder may get passive yield, but the platform or lessee may own the pricing context.

This does not make the model bad. Many small holders should not try to recreate the entire service stack. Outsourcing routing, abuse handling or customer acquisition can be rational. The question is whether CPS receives enough of the value after outsourcing. A small company can earn attractive returns on a legacy or scarce asset if overhead is minimal. It can also become a commodity supplier if every material function sits with a larger counterparty.

Customer concentration is the largest unknown

The hardest missing disclosure is customer concentration. The public evidence does not show whether CPS has one major lessee, several smaller lessees, a managed-service customer base, internal use, or a mixed book. The two /23 sub-allocations both point to the same IPXO-linked organisation. That creates the appearance of concentration at the resource-management layer. It does not prove that end use is concentrated, because the platform or intermediary could serve many underlying users. But from CPS's perspective, concentration at the payment or contract layer would still matter.

If CPS has a single address-leasing counterparty, its renewal risk is high. The counterparty can compare rates, demand better terms, return space if reputation deteriorates, or switch to another block. If CPS has two or more unrelated long-term service customers that require the space for business continuity, the risk is lower. If CPS combines resource leasing with consulting or managed operations, the relationship may be stickier than the address records imply. The public evidence cannot distinguish those scenarios.

The same uncertainty applies to margin. A small resource holder's gross revenue can look clean if the block is leased, but net margin depends on platform fees, tax, bank costs, administrative labour, dispute handling, monitoring, insurance, professional advice and opportunity cost. Creditsafe and Drimble do not provide usable revenue or profit figures in their free public snippets. The LEI record does not disclose financials. RIPE records do not disclose commercial terms. Without those, any margin estimate would be false precision.

The article therefore weights the uncertainty heavily. A strong conclusion would require facts such as recurring revenue by line of business, customer churn, contract length, average lease price, direct versus intermediary payments, support hours, and whether CPS has any non-address recurring service revenue. The current record supports a cautious operating conclusion: CPS has a monetisable resource position, but public evidence does not yet show a diversified demand base or enough differentiated service value to lift it clearly above infrastructure price-taking.

Customer concentration also affects downside. If a single lessee causes abuse complaints or payment disputes, CPS may need to withdraw space, find a replacement lessee, clean reputation and reconfigure records. For a 1,024-address block, one troubled /23 is half the IPv4 inventory. That makes operational diligence more important than it would be for a large holder with thousands of independent assignments. Scale does not just increase revenue; it diversifies mistakes.

Dutch market alternatives set a hard ceiling on pricing

The Netherlands is not an underserved connectivity market where any address holder can command premium access pricing by default. ACM's telecom monitor has described a broadband market dominated by KPN and VodafoneZiggo, with each holding large shares in the fixed broadband market, while fibre adoption continues to rise. Business.gov.nl states that telecom and mail delivery companies in the Netherlands must register with ACM, and that providers have obligations around registration, changes, turnover reporting and related telecom duties.

ACM materials also describe annual fees for providers of public electronic communications activities based on net turnover. A small provider can operate in that environment, but it cannot ignore the scale and regulatory position of incumbents.

For CPS, this means direct access-network competition would be difficult unless the company has a narrow niche. Incumbents and larger business providers can bundle access, voice, mobile, security, cloud interconnect, managed LAN, SD-WAN and support. Cloud providers can bundle compute with global network services. Hosting providers can include addresses in monthly plans. IP brokers can supply address space without asking customers to buy broader Dutch network services. The practical substitute set is wide.

That does not eliminate opportunity. SMEs often need continuity, human support, stable addressing and practical network administration more than they need global hyperscale. A small Dutch company can win if it solves messy customer problems that large platforms treat as low priority: migration from legacy hosting, preserving static addresses, coordinating DNS and mail reputation, managing firewalls, or supporting small offices through provider changes. But those services need proof. They require customer relationships and operational know-how, not just a RIPE record.

Cloud competition matters because it changes buyer expectations. BEREC's cloud and edge computing work cites high hyperscaler shares in cloud markets, including a 72 percent regional market share for hyperscalers in Europe in the referenced Synergy data. OECD cloud competition work similarly notes concerns around large cloud providers, switching frictions and concentration. That environment pushes small infrastructure companies toward two strategies: specialise locally or supply scarce inputs to larger ecosystems. CPS's public record looks closer to the second strategy.

If it wants to be valued as the first, the missing evidence would be customer specificity, service reliability, and local operational depth.

The Dutch market also has wholesale-access history and regulatory complexity. European Commission and ACM-related broadband documents have repeatedly examined wholesale access, KPN's position, cable constraints and business-service markets. Those records are not about CPS directly, but they establish a competitive backdrop: network access and business connectivity in the Netherlands are shaped by infrastructure scale, regulation and incumbent economics. A small resource holder cannot assume that holding addresses alone creates local network power.

Regulation turns small scale into fixed overhead

Regulation is not the main risk because CPS is small; it is a risk because small companies have fewer transactions over which to spread compliance. RIPE membership itself carries registry duties. Public telecom activity in the Netherlands can trigger ACM registration, turnover reporting and fee obligations. Address leasing can create abuse, sanctions, know-your-customer and contractual diligence concerns. RPKI and routing hygiene require technical attention. None of these obligations is impossible, but each turns into fixed overhead.

The current resource record shows some attention to route hygiene. The two live IPv4 /23s are RPKI-valid for AS2856 in RIPEstat's validation data. That is positive. The IPv6 /29 routed by AS44928 returns unknown in the RIPEstat RPKI validation call, which is not the same as invalid but is less strong than a valid status. The existence of valid RPKI on the IPv4 routes suggests that the current operating arrangement has at least addressed origin validation for the active IPv4 halves. In a world where networks increasingly filter invalid routes, that matters.

The compliance downside is reputational. An address holder linked to third-party use must ensure that abuse contacts work and that delegated users do not damage the block. RIPE records include an abuse contact for CPS and separate abuse contacts for the IPXO-linked sub-allocations. That separation is operationally useful, but it does not remove CPS from the ownership chain. If the addresses are used in ways that attract complaints, the company's name can still appear in WHOIS-derived data, geolocation records and reputation tools.

The sanctions and geopolitical risk is limited in the evidence reviewed. CPS is Dutch, active in the LEI system and tied to RIPE-region resources. No sanctions-specific public finding appeared in the source set. The relevant geopolitical risk is more about cross-border routing and counterparties. Current IPv4 origin is British Telecommunications, the sub-allocation organisation is UK-based, and the company is Dutch. That is a normal European network arrangement, but it means the business is exposed to more than one jurisdiction's operational and legal expectations.

Small scale also affects resilience. A large provider can maintain dedicated compliance, abuse and routing teams. A small holder may rely on one technical principal, an intermediary or outsourced support. RIPE's records list a named administrative/technical contact for CPS, while third-party contact pages also associate Stefan Baltus with CPS. Concentrated technical knowledge is common in small infrastructure firms, but it is a risk if customers require continuity. A management handover, illness, dispute or loss of a key outsourced relationship can affect service quality faster in a small company than in a large one.

Unofficial signals support a cautious leasing thesis

Unofficial market signals should be used carefully. IP2Location, Ping0, BrowserScan, BGP.tools, IPinfo and similar services collect, infer and display network data, but they are not contract records. They can lag, misclassify geography, conflate holders with users, or display stale route history. Still, when multiple independent network-intelligence pages align with primary RIPE and RIPEstat data, they can help interpret the commercial pattern.

Here, the alignment is meaningful. IP2Location shows an address inside the CPS range with ISP CPS Internetworking BV, domain cps-i.nl, proxy ASN AS2856 and usage type data centre/web hosting/transit. Ping0 shows current AS2856 visibility for 45.149.168.0/23, previous appearances under Orange and IDC IPXO, and a company-history view that keeps the broader /22 tied to CPS while showing a current /23-level Internet Utilities/IPXO label. BrowserScan displays the 45.149.168.0/23 segment with IPXO details and the RIPE sub-allocation fields. BGP.tools shows AS44928 as small and IPv6-only, with the CPS IPv6 /29 among its originated prefixes.

The signal is not that CPS is secretly a hyperscale cloud company or that it sells proxy services. The signal is narrower: CPS's IPv4 inventory appears to be in the address-market infrastructure, where sub-allocation, intermediary management and BT-originated routing matter. That supports the thesis that CPS may be earning resource-holder value. It also supports the concern that CPS's differentiation may be weak if the customer sees the product as replaceable IPv4 capacity.

This is where the "below cloud scale" phrase becomes economically concrete. Cloud scale is not only about owning data centres. It is about controlling demand, automation, support, billing, reputation, route operations and product attachment at large volume. CPS does not show that scale publicly. The company may have a rational lean model, but lean is not the same as powerful. The more functions are performed by IPXO-linked infrastructure, BT routing, Trixit routing or other counterparties, the more CPS looks like an input supplier rather than a platform.

The unofficial signals do help reject one alternative: the address space does not look abandoned. It is visible, structured, sub-allocated and routed. That is better than a dormant-resource story. The downside is that active use through stronger counterparties can still be economically thin for the holder. Activity is not the same as attractive margin.

What would change the judgment

The current judgment is cautious: CPS Internetworking BV has a real resource-holder footprint and likely monetisable IPv4 value, but the public record does not prove enough differentiated demand to conclude that the company earns durable above-commodity returns. The cost base may be small, which helps. The resource may be clean and useful, which helps. But routing and sub-allocation signals point to reliance on larger counterparties, and missing customer/margin disclosure prevents a stronger conclusion.

Several facts would change that view. First, evidence of multi-year direct contracts with Dutch SMEs, managed-service customers or hosting users would show that CPS owns demand rather than merely supplies address space. The most persuasive evidence would include contract duration, renewal history, SLA terms and customer dependence on CPS-specific operational knowledge. Second, disclosure of lease economics would matter: actual price per address, platform or broker fees, payment reliability, rights to reclaim the space, and abuse indemnity would show whether address monetisation is attractive or merely marginal.

Third, proof of diversified counterparties would reduce concentration risk. If the two /23s are ultimately serving many stable end users through enforceable arrangements, the current sub-allocation structure is less concerning. If one platform or lessee controls most revenue, the business remains fragile. Fourth, route-control evidence would matter. CPS running its own autonomous system, maintaining its own PeeringDB profile, or showing direct upstream diversity would indicate deeper operating capability. The current public record instead shows AS44928/Trixit and AS2856/BT as key routing actors.

Fifth, a visible service proposition would help. A website describing managed network operations, SME continuity, security, IPAM, migration support or Dutch business connectivity would not by itself prove margin, but it would support a thesis of customer-specific value. The absence of such material leaves the company looking like a registry and resource-management case. Sixth, RPKI and abuse-process evidence across both IPv4 and IPv6 would improve the risk profile, especially if CPS can show that third-party use is monitored and contractually controlled.

The fact pattern that would most clearly overturn the price-taker thesis is a bundled service model: CPS owns the address space, directly serves a set of sticky Dutch business customers, controls routing through redundant providers, charges for managed continuity rather than raw addresses, and uses the IPv4 block as a retention tool rather than a standalone commodity.

The fact pattern that would confirm the price-taker thesis is the opposite: most revenue comes from a single address-leasing or sub-allocation counterparty, pricing follows platform averages, CPS has little visibility into end users, and operational problems are handled by intermediaries whose terms CPS cannot materially influence.

On today's evidence, the second pattern is closer to what can be verified, though not conclusively proven. CPS is not an empty name; it is a real Dutch company with RIPE membership, a scarce IPv4 block, IPv6 resources and active routing signals. But below cloud scale, owning a scarce input is only the opening move. The value is earned when management attaches that input to differentiated demand, durable contracts and controllable operations. The public record does not yet show that CPS has made that conversion at a scale or quality that would protect it from the economics of a small infrastructure price-taker.