Summary

  • Camelot-A is not publicly evidenced as a regional ISP. Russian corporate data identify food retail as its main activity, while its own online terms identify it as the seller behind Yarche! and place it within the KDV group.
  • The company has genuine Internet-number credentials: RIPE records list it as a Local Internet Registry and assign AS208044 to it. Yet RIPE routing observations show no current IPv4 or IPv6 announcements, no visible neighbours and no active address space originated by that autonomous system.
  • AS208044 did originate 193.143.250.0/23 from January 2021 until April 2022. That block is now originated by the wider KDV network, AS62382, indicating consolidation of external routing at group level rather than present-day network independence at Camelot-A.
  • Revenue rose from RUB 100.2 billion in 2023 to RUB 132.0 billion in 2024 and RUB 168.3 billion in 2025, but net profit moved from RUB 1.5 billion to losses of RUB 3.8 billion and RUB 4.0 billion. Growth therefore destroyed rather than created accounting value over the last two years.
  • Public reporting on the 2025 accounts says merchandise purchasing expense rose 28% and labour expense rose 42%. Those are the costs that dominate the judgment; the RIPE fee itself is immaterial, while engineers, access circuits, replacement equipment, security and duplicated capacity are not.
  • The current group-network arrangement is economically more credible than rebuilding a stand-alone public edge. Camelot-A should buy measured resilience for stores and digital commerce, retain number-resource competence as an option, and reject any independence project that cannot show lower total cost or a quantified reduction in lost sales.

Independence is valuable only when it lowers the cost of failure

The economic incentive is straightforward. A retailer with more than a thousand shops, daily deliveries, card terminals, electronic inventory records, an online store and a mobile application cannot treat connectivity as office overhead. A failed link can interrupt payment authorisation, price updates, stock visibility, electronic documents, product-marking checks, loyalty accounts and delivery fulfilment. The retailer pays for the interruption through lost baskets, manual work, spoilage, reconciliation errors and damaged trust. The customer pays through delay or a wasted visit. Suppliers pay when orders and acceptance records stop moving. A connectivity vendor may lose little beyond a service credit that covers only a fraction of the commercial damage.

That asymmetry creates a rational desire for control. An autonomous system can let an organisation express its own routing policy, connect to more than one upstream and move traffic without renumbering every public service. Address resources can reduce dependence on a single carrier. Engineers who understand the retail estate can design failover around business priorities rather than sell a generic access product. In theory, the company captures the benefit of redundancy while forcing competing carriers to sharpen prices.

But owning a number and operating a resilient network are not the same act. Independence brings annual registry charges, engineering salaries, around-the-clock response, security work, monitoring, replacement hardware, spares, software support and the recurring cost of buying physical access from somebody else. No retailer manufactures fibre routes merely by holding an autonomous-system number. A second contract does not create diversity if both suppliers use the same duct, building entrance, power feed or regional backbone. The company continues to bear outages; it has merely changed which costs it pays before and after they occur.

Camelot-A therefore needs a stricter test than whether network autonomy is technically possible. The question is whether another rouble spent on control produces more dependable gross profit than the same rouble spent on store staffing, distribution, refrigeration, shrink reduction, better purchasing terms or selective expansion. That is especially important because the company's recent accounts show vigorous sales growth alongside large losses. A strategy that does not identify the scarce resource it will consume and the cash return it will produce is marketing. On the evidence available, a stand-alone public network for Camelot-A does not yet pass the test.

The company is a grocer, not a proven connectivity seller

The legal identity is unusually important because the network record can mislead readers who begin with it. Camelot-A was registered in Tomsk in July 2007 under OGRN 1077017026580 and tax identifier 7017187800. Public corporate records state that its principal activity is retail sale mainly of food, beverages and tobacco in non-specialised stores. They show LLC Yarche as its sole shareholder. Camelot-A's own procurement portal uses the same registration numbers and Tomsk address, describes Yarche! as a federal proximity-supermarket chain and identifies Camelot-A in the footer.

The online terms remove any residual ambiguity. They define Camelot-A as the seller operating the Yarche Plus website and mobile application, say the company is part of the KDV group, and describe remote sale and delivery of food and non-food goods. The retailer's supplier page claims more than 1,000 stores, more than 1.5 million customers buying from the chain each day and more than 16,000 regularly purchased products. It says stores do not hold back-room stock and receive everyday products daily. A separate geography page lists five distribution centres: Obukhovo in Moscow region, Zorkaltsevo in Tomsk region, Novokuznetsk, Tolmachevo near Novosibirsk and Omsk. Regional reporting in September 2024 said the chain had passed 1,000 stores and was opening in Tyumen and Yekaterinburg. Later reporting put the estate above 1,300 shops.

This is an operating surface built around fast stock turns, local availability and many small transactions. Camelot-A earns revenue when a consumer buys groceries, not when an outside customer buys an Internet circuit. No public price list, service description, licence disclosure, peering profile or customer reference reviewed for this assessment establishes an ISP, transit, hosting or managed-network product sold by Camelot-A. The absence of such evidence does not prove that no incidental service exists. It does mean that telecom revenue cannot responsibly be included in the valuation argument.

RIPE membership establishes something narrower and still useful. The RIPE NCC member directory lists Camelot-A at Prospekt Mira 20 in Tomsk, with Russia as its service area. The RIPE Database describes the company as a Local Internet Registry and records the same Russian corporate registration number. This shows an administrative and technical capacity to hold and administer Internet number resources. It does not reveal fibre ownership, store circuits, traffic volume, customer contracts, service-level performance, data-centre capacity or telecom gross margin.

That distinction changes who should pay. If the network were a commercial product, connectivity customers would fund engineering, redundancy and equipment renewal through recurring charges. For Camelot-A, the likely payer is the grocery operation. Network spending must be recovered through avoided retail loss, lower carrier bills, greater labour productivity or better digital sales. Those benefits can be real, but they are internal cost economics, not a second revenue engine.

AS208044 records a former edge, not current independence

Camelot-A's strongest network credential is AS208044, registered as CMLT-AS. RIPE records show that the number was assigned in October 2019 and remains assigned. The organisation record is current enough to have been modified in May 2026. These are not scraps from an unrelated namesake: the organisation object carries Camelot-A's Russian registration number and Tomsk address.

The routing evidence is much less expansive. RIPEstat's current overview marks AS208044 as not announced. Its announced-prefix view contains no IPv4 or IPv6 prefixes for the two weeks ending 10 July 2026. The routing-status view reports zero IPv4 addresses, zero IPv6 space, zero observed neighbours and visibility from none of the hundreds of RIPE RIS peers counted in the service. PeeringDB returns no network record for the number. A commercial data service likewise classifies the autonomous system as inactive and shows no hosted domains or addresses, although such commercial summaries should be treated as corroboration rather than authority.

History makes the record more informative. RIPE's routing observations show AS208044 first originating 193.143.250.0/23 in January 2021. The route stayed broadly visible until April 2022. A /23 contains 512 IPv4 addresses, enough for a modest corporate public edge but trivial beside a national store estate. It could support public services, remote access, business systems or translated addresses for a limited set of sites. It could not by itself tell us how the store network was built, how many private addresses sat behind it or whether branch access was diverse.

The route did not disappear from the Internet. RIPE records identify 193.143.250.0/23 as an allocation associated with LLC KDV Grupp, and the current routing view shows it announced by AS62382, KDV's autonomous system. The block is visible to all 327 IPv4 collectors in the current RIPEstat snapshot. The cleanest interpretation is that Camelot-A once originated a KDV-associated block from its own edge and that external origination later moved to the group network. This is an inference from routing history and registry records, not a disclosed corporate explanation.

That move can be economically sensible. A central group edge can pool security tools, engineers, upstream contracts, address administration and spare equipment across a much larger operating base. KDV's public routing policy lists imports from numerous carriers and the current RIPE observation sees thirteen neighbouring networks. Its announced set includes the former Camelot-A /23 alongside other IPv4 blocks. Observed neighbours are not proof of thirteen paid, simultaneous transit contracts; some may reflect regional or historical routing arrangements. They do show that the group network has a materially broader public footprint than AS208044.

Consolidation also moves the failure boundary. Camelot-A may gain upstream diversity without staffing its own public edge, but it becomes dependent on group governance, central change control and shared security. A bad configuration at the common edge can affect several businesses at once. A dispute over budget or ownership can determine which subsidiary receives capacity first. The right comparison is therefore not autonomy against dependence. It is a narrow Camelot-A failure domain with duplicated cost against a broad KDV failure domain with economies of scale.

The current evidence favours the latter. Keeping AS208044 assigned preserves an option at low direct registry cost. Re-activating it would make sense only if Camelot-A can demonstrate that the group edge cannot deliver the required routes, recovery times, security separation or commercial leverage. No such demonstration is public.

Sales expanded while value creation reversed

The financial record is the coldest check on the independence story. A published series based on Russian accounts shows Camelot-A revenue of RUB 54.9 billion in 2019, RUB 55.8 billion in 2020, RUB 64.6 billion in 2021, RUB 81.8 billion in 2022, RUB 100.2 billion in 2023, RUB 132.0 billion in 2024 and RUB 168.3 billion in 2025. On the sales line, this is formidable expansion: revenue more than tripled in six years and rose about 68% in only two.

The bottom line refuses to validate the same story. The series shows losses of RUB 1.45 billion in 2019, RUB 2.12 billion in 2020 and RUB 1.01 billion in 2021; profit of RUB 361 million in 2022 and RUB 1.51 billion in 2023; then losses of RUB 3.79 billion in 2024 and RUB 4.02 billion in 2025. The implied net margin improved to roughly 1.5% in 2023, collapsed to negative 2.9% in 2024 and was still negative 2.4% in 2025. Revenue growth narrowed the loss ratio slightly in the latest year, but it did not restore value creation.

The two-year contradiction is stark. From 2023 to 2025, Camelot-A added about RUB 68.2 billion of annual revenue while net income deteriorated by roughly RUB 5.5 billion. Across the full 2019-2025 series, the simple sum of reported annual results is a cumulative loss above RUB 10 billion. That is not a cash-flow calculation and says nothing about dividends, leases or asset investment. It is still a warning against praising scale without asking what each new rouble of sales costs.

Public reporting on the 2025 statements identifies the main pressure points. Merchandise purchasing costs rose 28%, close to the 27.6% increase in revenue, while labour expense rose 42%. Headcount reached about 16,600 after increasing by roughly 3,000. If those headcount figures are comparable, staffing grew about 22% while labour expense rose much faster, implying a sizeable rise in cost per worker as well as more workers. This is where the margin went, not into the EUR 1,800 annual RIPE membership charge or the EUR 50 fee per autonomous-system assignment.

Those registry fees are worth stating because they expose a common category error. The right to administer numbers is cheap relative to a RUB 168 billion retailer. Operational independence is not. A useful edge requires routers, firewalls, optical equipment, software rights, monitoring, engineers who can respond at night, physical access to independent carriers, security assessment and a replacement cycle. Branch resilience multiplies that expense across stores and warehouses. The membership line can be immaterial while the architecture it enables remains expensive.

Rough scale calculations show both the opportunity and the limit. Dividing 2025 revenue by 365 produces average sales of about RUB 461 million a day. Dividing again by 1,300 stores produces about RUB 355,000 per store per day, although the store count is a period-end approximation and reported revenue may include activities not evenly distributed across shops. One hour of company-wide sales flow is around RUB 19 million on a simple average. Preventing a severe central outage can therefore justify meaningful investment. Preventing a one-hour interruption at one ordinary store cannot justify a private backbone on its own.

The decision must be made by failure domain. Central ordering, pricing, payment interfaces, identity, product marking and distribution systems can affect hundreds of stores simultaneously and deserve strong redundancy. A single low-volume branch may be better served by two commodity links, one fixed and one mobile, with an offline mode for critical functions. Network control is most valuable where commercial loss is concentrated; uniform technical prestige wastes money.

Reliability earns its return through retail operations

Camelot-A's public materials reveal several places where dependable data movement could protect gross profit. The supplier terms require structured documents at shipment, including transport or universal transfer records, invoices and quality certificates. For goods subject to mandatory marking, suppliers must send electronic dispatch and transfer documents when shipping. Missing documents can cause refusal at the warehouse. Veterinary records may need correction within four hours, including weekends. The terms also set penalties and reimbursement duties for late, incomplete, incorrectly marked or unacceptable deliveries.

This is a business in which data quality and physical flow meet at the loading bay. Connectivity does not create demand for yoghurt or bread, but a failed document exchange can stop compliant goods from entering inventory. A delayed stock update can create an empty shelf in a format that advertises freshness and daily replenishment. A stale price can produce a dispute at checkout. An unavailable marking service can slow acceptance. A broken route to a payment processor can turn a willing buyer into abandoned revenue.

The online channel extends the dependence. Camelot-A's terms support customer registration, a mobile application, product catalogues, online orders, payment information, order identifiers and delivery status. The delivery page lists service windows across many cities and neighbourhoods. Customer data include contact details, order history, device and browser information, online activity and location. That creates obligations for availability, privacy, security and accurate recovery. It also creates a direct commercial choice: whether to own the network path, buy it, or combine both.

Yet none of these functions requires Camelot-A to originate a public route under its own autonomous-system number. A well-designed service can use multiple carriers, cloud or domestic hosting, content delivery, replicated applications, store-and-forward messaging and offline transaction rules while the public edge remains at group level. Conversely, an active ASN does not prevent a database failure, defective application release, warehouse power loss, mobile-network restriction or fibre cut before the first diverse junction.

The return on reliability should therefore be measured in operational outcomes. Relevant figures would include minutes of checkout unavailability, failed payment attempts, orders cancelled for technical reasons, shelf gaps caused by delayed replenishment messages, warehouse acceptance time, delivery substitutions, support hours, losses during each incident and the share of incidents attributable to carrier concentration. A proposal for autonomy should state which of those measures will improve, by how much, and at what total cost.

Without those figures, the most defensible model is selective redundancy. High-impact central sites merit physically diverse access, tested failover, spare equipment and strict separation of administrative control. Distribution centres merit redundant paths because one interruption can impair many stores. Branches merit diversity proportionate to sales and local carrier conditions. The online service merits application-level resilience independent of the corporate network. The assigned ASN can remain an option rather than a trophy.

Upstream diversity is bought, not declared

Who pays for independence? Ultimately, grocery customers and suppliers do, through the prices and trading terms that fund Camelot-A. The retailer can absorb some cost by becoming more efficient, but a loss-making company cannot pretend capital is free. If network spending does not reduce another expense or prevent a larger loss, it deepens the deficit.

Who benefits? Store teams benefit when checkout and inventory systems stay available. Customers benefit when payment, pricing and delivery work. Suppliers benefit when orders and documents arrive. The group can benefit from lower unit costs if expertise and upstream capacity are shared. Carriers may benefit too, because multi-homing usually means the customer keeps at least two paid relationships rather than eliminating suppliers.

Who bears the downside? Camelot-A bears fixed payroll and equipment costs even when no failure occurs. It bears the integration risk when old store equipment meets a new design. It bears security risk if more routing and administrative capability expands the attack surface. It may bear foreign-exchange and procurement risk when replacement hardware or software is difficult to source. Employees bear the operational burden when failover procedures are incomplete. Customers still bear inconvenience when physical access routes share a hidden dependency.

The public KDV routing record shows why group buying can be superior. AS62382 currently originates several aggregate and more-specific IPv4 routes. Its registry policy names a range of upstream networks, and RIPE sees a broad set of neighbours. At group scale, traffic and engineering can be pooled. A larger buyer can negotiate with carriers across regions, maintain common security policy and hold spares that serve several facilities. Camelot-A alone would have to reproduce part of that capability before it achieved any incremental benefit.

But the group arrangement must not be accepted on faith. Supplier concentration can move inside the corporate perimeter. Camelot-A may depend on KDV's central network team as heavily as it once depended on a carrier. The remedy is commercial clarity: documented service objectives, cost allocation, incident reporting, change windows, recovery tests and the right to establish a separate edge if the shared service repeatedly fails. Independence has bargaining value when it is a credible option, not when it is an unused number with no engineers or access contracts behind it.

Physical diversity remains the hardest and most valuable proof. For each high-impact site, Camelot-A should know which carrier owns the last mile, which ducts and poles are used, where paths first converge, which power systems feed the equipment and whether mobile backup relies on the same regional transport. Carrier logos are not a topology. The economic buyer should pay for verified separation, not for the appearance of two invoices.

Customers are diffuse, but the market still sets the price

Camelot-A does not appear to face classic business-customer concentration. The claim that more than 1.5 million customers buy from the chain each day implies a highly distributed consumer base rather than reliance on a few accounts. That reduces the risk that one buyer can cancel a contract and remove a large share of revenue. It does not create pricing power.

Food retail customers can switch at the level of each basket. Competitors include national proximity chains, discounters, regional grocers, marketplaces and rapid-delivery services. The consumer does not need to terminate an agreement; the consumer walks to another shop or opens another application. Network reliability matters, but so do shelf price, location, freshness, assortment and waiting time. A technically elegant store with an uncompetitive basket still loses.

The revenue numbers suggest that Camelot-A has been effective at adding selling capacity or nominal sales. Regional reports describe expansion into the Urals, and the network passed from more than 1,000 stores to more than 1,300 in public accounts of its scale. Inflation also raises nominal grocery revenue without proving greater physical volume or better margin. The accounts do not disclose enough to separate new-store sales, like-for-like volume, price inflation and online contribution. That missing split is central: revenue growth from openings can consume capital and labour before stores mature, while inflation can make the top line look stronger as purchasing costs rise in parallel.

X5 provides a useful benchmark, not a direct valuation comparable. The much larger Russian retailer reported 2025 revenue of RUB 4.64 trillion, adjusted earnings before interest, tax, depreciation and amortisation before lease-accounting effects equal to 6.1% of revenue, and a 2.0% net margin. Its staff cost was 8.6% of revenue; lease expense 3.9%; utilities 1.9%; and customer-delivery cost 0.8%. Its net margin still fell by 0.77 percentage points despite 18.8% revenue growth. The lesson is not that Camelot-A should match every ratio. It is that even a leader with purchasing scale, a large digital business and disclosed productivity programmes operates within a few points of error.

Camelot-A's negative 2.4% net margin in 2025 sits on the wrong side of that narrow range. Returning merely to its own 2023 net margin of about 1.5% would require an improvement around four percentage points from 2025, equivalent to roughly RUB 6.5 billion at the latest revenue level. No plausible saving on RIPE fees or public routing alone approaches that amount. Network investment belongs in the solution only if it improves larger cost pools: labour time, shrink, delivery efficiency, payment success, stock availability or carrier procurement.

Supplier bargaining is another source of margin, but it has limits. Camelot-A's published supply terms are detailed and allocate many costs of non-compliance back to vendors. They cover delivery timing, product quality, electronic records, marking, rejection, re-sorting and penalties. That demonstrates purchasing discipline and a digital control surface. It also shows why reliability benefits suppliers: accurate orders and rapid acceptance reduce disputes. Pushing every cost outward, however, can narrow the supplier pool or return through higher quoted prices. The best network reduces shared friction; it does not merely improve the retailer's ability to issue penalties.

Labour and equipment, not registry charges, decide the case

The 42% increase in labour expense reported for 2025 is the most important unit-economic signal. Camelot-A added about 3,000 employees while sales and geography expanded. Retail labour covers stores, warehouses, delivery, support and administration; public data do not isolate network staff. The company therefore cannot assume that bringing more connectivity work in-house is a free use of existing people.

A competent independent edge needs scarce skills. Routing policy must be maintained. Firewall and remote-access rules must be reviewed. Certificates, domain services and address records must be accurate. Security incidents require investigation. Hardware fails outside business hours. Carriers need escalation. Changes need rehearsal and rollback. A small team can operate a modest edge, but resilience against absence and turnover requires depth. If the company depends on one knowledgeable employee, it has replaced supplier concentration with key-person concentration.

Equipment refresh is equally unforgiving. Routers and firewalls have finite support lives. Capacity requirements rise with encryption, richer applications, cameras, analytics and online traffic. Spare units need compatible software and tested configurations. Sanctions and export controls affecting advanced technology, electronics, encryption and certain telecom items can narrow vendor choice or complicate support for Russian enterprises. The effect on any specific Camelot-A purchase is not public, so it should be treated as procurement risk rather than a confirmed shortage. It still raises the value of standardisation, longer planning horizons and domestic support capability.

The alternative is not helpless resale. Camelot-A can buy managed routing from multiple carriers, use the KDV public edge, retain internal architecture and security skills, and require transparent incident data. It can own customer-premises equipment where that improves control without operating every upstream session. It can use an independent specialist for design review while keeping day-to-day support under contract. It can activate AS208044 only at high-impact sites or as a migration tool if the group service ceases to meet its needs.

Each alternative consumes a different scarce resource. Full autonomy consumes engineers and capital. A managed service consumes vendor margin and may create lock-in. Group centralisation consumes internal bargaining power and can create a shared failure domain. Commodity access with application-level resilience may be cheap but leaves some network failures outside the retailer's control. There is no ideology in the choice. The winning design is the one with the lowest total cost for the required commercial recovery time.

State ownership and sanctions add governance risk to technical risk

The operating judgment cannot exclude the 2025 change in control around KDV. A Russian court ordered the KDV holding transferred to state ownership after designating its former beneficial owners as an extremist association. Reuters reported that the group included more than ten factories and that Yarche! had more than 1,300 stores. Interfax reported that the Moscow City Court left the decision in force in February 2026. A report on Camelot-A's 2025 accounts says the Russian Federation became the company's beneficiary on 1 October 2025.

This is not a footnote to network planning. A new owner can alter capital allocation, approved vendors, security requirements, internal charging, executive incentives and the boundary between subsidiary and shared systems. Centralisation under AS62382 may become more attractive if the state owner prioritises group control. Separation may become more attractive if resilience, accountability or future restructuring requires clear technical boundaries. Public records do not disclose the new owner's detailed network policy, so either outcome remains possible.

The RIPE framework creates a separate cross-border dependency. RIPE NCC is based in the Netherlands and says it complies with European Union sanctions. It explains that Internet number resources are treated as economic resources for sanctions purposes. For a sanctioned holder, RIPE freezes registration rather than use: the holder cannot acquire or transfer resources, but RIPE does not automatically deregister them or terminate membership. RIPE also notes that banking checks can affect invoicing and payment. Its 2026 fee page gives special billing documentation information for Russian members.

There is no public evidence reviewed here that Camelot-A's resources are frozen. The point is conditional. A Russian entity's number-resource strategy now depends not only on technical competence and domestic law but also on Dutch association rules, European sanctions, payment rails and due diligence. Keeping records current and ownership evidence clear has option value. So does avoiding a design that assumes resources can always be transferred quickly during a restructuring.

Domestic compliance expands the availability requirement. The supplier terms refer to electronic veterinary records, alcohol-control records and the Chestny Znak mandatory-marking system. The online terms invoke Russian personal-data law and describe extensive customer information. These obligations make systems and connectivity more important, but they also increase the cost of poor security and inaccurate data. Independence that improves uptime while weakening access control is not reliability; it is a different form of expected loss.

Public reviews are warning signals, not operating statistics

Unofficial commentary offers only weak evidence, but it can identify questions worth testing. Customer reviews indexed by Yandex include positive comments about proximity, assortment and acceptable delivery timing. Reviews on another consumer site describe the convenience of home delivery while some users characterise the delivery service as immature. These are self-selected experiences, not a representative satisfaction survey, and they do not isolate network failures from picking, inventory, courier or customer-service problems.

Employee-review sites carry a similarly mixed and more negative set of signals. Posts describe long shifts, physical intensity, combining checkout and goods-receiving duties, close monitoring and uneven local supervision; some also praise pay in smaller cities. Identities and circumstances are not verified. The reviews cannot establish company-wide labour conditions. Their relevance is narrower: they are consistent with a retailer in which store labour is stretched and process complexity can fall on frontline staff.

That matters to the network decision because a brittle technical design creates manual work. When inventory, payment or marking systems fail, employees reconcile, re-enter, explain and queue. A reliable design can reduce that burden. A complicated failover procedure can increase it. Camelot-A should treat incident-related support calls, manual corrections and overtime as part of network cost, rather than looking only at carrier invoices.

The Google Play listing confirms that Yarche Plus is an active delivery application and advertises delivery within an hour. It also notes that data practices can depend on use, region and age. The listing proves a customer-facing digital channel, not its profitability or reliability. The useful market signal is that users can compare the experience immediately against large rivals. Availability is necessary to compete, but an application rating or anecdote cannot tell Camelot-A whether an ASN should be active.

The realistic alternatives are incremental, shared and measurable

Camelot-A has four broad choices. The first is to remain entirely dependent on one group or carrier design. It is cheap to administer but leaves pricing, change control and outage response concentrated. Given the scale of the store estate, that is difficult to defend unless the underlying service already includes verified physical diversity and strong recovery performance.

The second is full stand-alone autonomy: re-activate AS208044, obtain or assign suitable address space, contract multiple upstreams, staff the edge and separate public services. This maximises formal control. It also duplicates capability already visible at KDV level and creates new fixed costs while Camelot-A is loss-making. The evidence does not support it today.

The third is group centralisation with hard internal commitments. KDV continues to originate the shared address space and buy upstreams, while Camelot-A defines critical services, receives incident and route data, tests recovery and retains the contractual or technical right to separate. This captures scale while preserving leverage. It is the strongest default on current evidence, although ownership change makes transparent governance essential.

The fourth is selective independence. Camelot-A keeps group routing for normal traffic but adds diverse access, separate security zones, independent name resolution or a limited stand-by edge for the systems whose failure would affect many stores. Distribution centres and central commerce services receive more protection than ordinary branches. This approach spends capital where the avoided loss is largest and can be expanded only after measured results.

The choice should be reviewed against realistic alternatives every year. Carrier prices change. Store geography changes. KDV's internal service may improve or deteriorate. Hardware support changes. Online sales may become a larger share of revenue. An assigned but inactive ASN has option value precisely because Camelot-A does not need to make an irreversible choice today.

The company should also separate revenue growth from reliability value. More stores increase total dependence on connectivity, but they do not automatically improve the return on a central edge. The return rises when a common failure can interrupt more profitable transactions, when internal control shortens recovery, or when multi-carrier buying lowers recurring cost. It falls when new stores are low-volume, carrier paths are not truly diverse, or group services already provide the same capability.

Five disclosures would change the judgment

First, current network maps and incident records could show that AS208044 is used in ways not visible to public routing collectors, or that Camelot-A controls private interconnection across stores and warehouses. Private control can be economically important even when the public ASN is inactive. Evidence of measured outage reduction and lower total carrier cost would strengthen the case for independence.

Second, segment accounts could reveal telecom or network-service revenue. A recurring external revenue line, with customers, pricing and gross margin, would change the payer from grocery operations alone and justify a different capital base. No such line is established publicly today.

Third, store and channel economics could show how much revenue and gross profit are lost during connectivity incidents. Payment decline rates, order cancellations, downtime minutes and manual-recovery cost would turn reliability from an assertion into an investment case. A small number of high-severity failures could justify more control even if average availability appears good.

Fourth, details of the KDV shared service could prove or disprove concentration. Physical path maps, current upstream contracts, recovery tests, staff coverage and internal charges would show whether Camelot-A receives genuine diversity or merely a central dependency. The observed breadth of AS62382 is encouraging but insufficient.

Fifth, a credible return-to-profit plan could establish room for investment. Camelot-A needs to explain how sales growth will translate into gross and operating margin after merchandise, labour, leases, utilities, delivery, shrink and finance costs. Network spending should appear as a quantified contributor to that plan, not as a substitute for it.

The conclusion: preserve the option, do not fund the symbol

Camelot-A had a visible autonomous edge and no longer does. Its former /23 is now originated by the wider KDV network, which has the broader upstream and address footprint. The company's own public identity is grocery retail. Its latest revenue is large, but two years of losses show that scale is not paying for itself. Those facts produce a clear decision.

Camelot-A should not spend scarce capital rebuilding a stand-alone public network merely to claim independence. It should keep AS208044 and its RIPE competence current, because the option is inexpensive and may become useful under changing ownership or carrier conditions. It should demand measurable resilience from the KDV network, verify physical diversity at high-impact sites, and buy selective separation where incident economics support it. It should direct the remainder toward the larger sources of lost margin: merchandise cost, labour productivity, store execution, shrink and delivery economics.

The benefit of redundancy belongs to customers, employees, suppliers and the retailer. The bill belongs to Camelot-A. The downside also belongs to Camelot-A when duplicated equipment, specialist labour and support contracts do not prevent enough loss to cover their cost. Until the company can disclose that return, network independence is an option worth preserving, not a strategy worth funding.