Summary
- 10G.KZ LLC is an active Almaty company registered in October 2015, while the network records now attached to it reach further back. RIPE records tie the company to AS25534, created in 2002, and AS61185, created in 2013. The date mismatch proves that the current legal company did not originate the entire operating history; it does not disclose how the resources moved into the present structure.
- The current commercial offer is tangible but geographically narrow. The
10g.kzaddress redirects to Q.Net, whose public site says new connections are available only in Almaty and advertises 100 Mbps, 1 Gbps and 10 Gbps access, daily port rental, metered or unlimited traffic, address rental and support for customer BGP routing. That is evidence of an access business, not evidence of national reach, cloud ownership or data-centre ownership. - Public routing data show two active systems, 5,120 unique visible IPv4 addresses across the principal announced blocks, valid route-origin authorisations for the tested routes and no visible IPv6 announcement. AS25534 currently reaches the wider Internet through Open Media Group and HOSTKEY in observed routing, while AS61185 sits behind AS25534. The resources provide control and scarcity value, but the second system adds no independent upstream path in the observed view.
- The list-price model can produce recurring cash, but it exposes the margin question. A 100 Mbps unlimited connection plus its port costs about KZT717,225 a year before VAT at the current daily prices; a 1 Gbps equivalent costs about KZT3.43 million. The company discloses no active-port count, realised price, traffic mix, churn, gross margin, customer concentration, capital spending or cash flow. Tax payments and address holdings cannot substitute for those measures.
- The judgment changes only with evidence that a diversified base of multi-year 1 Gbps and 10 Gbps customers pays enough contribution after upstream capacity, building access, support and equipment renewal. It weakens if the business relies on a few high-traffic users, one or two engineers, common-path upstreams or price increases that lag costs. The present view is cautious: 10G.KZ controls useful infrastructure, but public evidence does not yet show that control earning returns above the cost of keeping it reliable.
Relevance is the management incentive
The strategic question is not whether 10G.KZ can route packets. The public network proves that it can. The question is whether a small Almaty provider can turn that technical ability into a customer relationship valuable enough to survive three pressures at once.
The first pressure is carrier scale. Kazakhstan's competition authority describes fixed Internet as highly concentrated, with Kazakhtelecom and Kar-Tel, which trades as Beeline, holding a joint dominant position. A World Bank assessment using June 2023 data put Kazakhtelecom at 65.4% of fixed-network subscribers and Kar-Tel at about 22.6%. Those companies can combine access with voice, mobile service, television, cloud products and nationwide account coverage. They can also spread central systems and regulatory obligations across a much larger revenue base. (Kazakhstan competition authority review, World Bank digital-sector assessment)
The second pressure is cloud substitution. A company that once needed a local provider to supply addresses, mail, server access and connectivity can now rent computing by the hour and storage by the gigabyte. Kazakhstan's PS Cloud, for example, publishes a basic virtual server from KZT3,600 a month and separately priced processor, memory and storage resources. Cloud use does not remove the need for a reliable local connection. It does, however, let the cloud supplier capture more of the customer's technology budget while the access provider carries the physical last-mile obligation. (PS Cloud tariff revision)
The third pressure is the economics of readiness. A connection is valuable because it works every day, including when a cable is cut, an optic fails or an upstream route changes. The provider therefore pays for capacity, spares, monitoring and technical availability before knowing whether any particular customer will need them. A large network can pool those costs. A very small provider has fewer contracts over which to spread them.
10G.KZ's rational response is not to imitate a hyperscale cloud catalogue. It is to sell the things scale providers can struggle to deliver economically: a connection to a difficult building, a public address, a routing arrangement, a fast local decision, a technician who understands the customer's setup, or a contract shaped around actual traffic. That niche is real. It is also narrow. Strategy becomes value only when customers pay more for those differences than the company spends maintaining them.
The current company and the older network are not the same history
The legal identity is reasonably clear. A Kazakhstan counterparty service drawing on government and statistical records gives business identification number 151040018460, a registration date of 22 October 2015, an Almaty address at 597 Seifullin Avenue, office 718, and Askar Almatov as chief executive. Its July 2026 view classifies the company as a private domestic non-financial corporation and places it in the smallest employment band, zero to five employees. It records telecommunications activity as the current main business classification, while noting that source classifications have differed. (10G.KZ counterparty record)
RIPE's current organisation record independently matches the same registration number and building. It names 10G.KZ LLC as a Kazakhstan-based Local Internet Registry, or LIR, at the seventh floor of 597 Seifullin Avenue. The member page says the area serviced is Kazakhstan. This confirms the company as a recognised resource holder. It does not say how many customers it serves, where its cables run, which facilities it owns or how much revenue the resources generate. (RIPE member page, RIPE organisation record)
The network chronology is older. The RIPE record for AS25534 was created on 27 December 2002 under the name INTELSOFT-AS. AS61185 was created on 18 January 2013 under INTELSOFT-WIFI-AS. Both now point to the same RIPE organisation record for 10G.KZ LLC. Routing history for AS25534 also predates the current company's 2015 registration, while 185.146.16.0/22 was allocated in April 2016. (RIPE AS25534 record, RIPE AS61185 record, RIPEstat AS25534 routing status, RIPE RDAP for 185.146.16.0/22)
The safe conclusion is limited. An older Intelsoft network and its resources now sit under the 10G.KZ registry identity. The public records reviewed do not establish whether that happened through a transfer, a reorganisation, a change of name, an asset contribution or another arrangement. They do not identify the consideration paid, any liabilities assumed or the beneficial ownership history. The current company should receive credit for maintaining the resources, not an unsupported claim of having built every part of the network since 2002.
The public brand boundary also requires care. Visiting 10g.kz redirects to qnet.kz. Q.Net's frequently asked questions link back to a 10g.kz rules address, an independent payment page labels Q.Net as Internet service in Almaty and directs users to 10g.kz, and the Q.Net host sits inside address space registered to 10G.KZ. Together these facts support an operational association between 10G.KZ and the Q.Net service. They do not, by themselves, disclose a trademark licence, a separate operating company or the complete ownership chain. (10G.KZ web address, Q.Net questions and answers, Q.Net payment listing)
That distinction matters commercially. A customer signs with a legal provider, not with an address block or a website label. A serious contract review should identify the invoicing entity, the owner of customer-premises equipment, the holder of rights over local fibre and the party responsible for service credits. The public standard contract leaves the provider name blank for completion, so the website association should not be stretched into a legal conclusion that the document itself does not make. (Q.Net standard contract)
The offer is access, traffic and technical control
Q.Net's published offer is compact. The connection page says service is connected only in Almaty. It offers 100 Mbps, 1 Gbps and 10 Gbps access. For 1 Gbps, the customer must provide an SFP or SFP+ port and a specified optical WDM transceiver; for 10 Gbps, it must provide an SFP+ port and another specified WDM optic. New service begins after a contract is signed, the connection invoice is paid and technical feasibility at the address is confirmed. (Q.Net connection terms)
This is not a mass-market promise that every home can order a speed online. It is a building-by-building access proposition. The provider checks whether it can reach the site, decides the work required and charges accordingly. Within an on-net building, the listed connection price starts at KZT5,000 over copper or KZT40,000 over fibre. Outside the coverage area, connection starts at KZT150,000. Those figures are before VAT. The spread is economically important: civil works and building access can consume the margin from a new customer before traffic begins. (Q.Net tariffs)
The revenue model separates several components. Q.Net lists port rental, traffic, address rental, site visits and connection. Metered traffic is KZT0.03 per megabyte. Unlimited traffic on a 100 Mbps port is KZT1,485 per day, while unlimited traffic on a 1 Gbps port is KZT8,900 per day. Port rental is separately listed at KZT480 a day for 100 Mbps, KZT500 for 1 Gbps and KZT2,000 for 10 Gbps. An address costs KZT22 a day. A technician's visit starts at KZT4,000. (Q.Net tariffs)
The posted service conditions explain how the pieces fit together. They say there is no general subscriber fee and the customer pays for services actually received. Metered traffic is charged on the larger of incoming or outgoing usage during the accounting period, not both added together. Unlimited traffic is charged for the time the service is available. Addresses and ports are periodic charges. The provider can support BGP routing for address space supplied by the customer, and it offers both public and private address arrangements. (Q.Net general and technical conditions)
This structure has three attractions. First, it makes low fixed port prices possible while allowing revenue to rise with traffic. Second, it lets a sophisticated customer bring address space and request routing rather than accept only a consumer-style translated connection. Third, daily charging can align bills with time in service more closely than a rigid monthly bundle.
It also allocates risk. A customer on metered traffic carries usage volatility. A customer choosing unlimited service pays a high recurring amount whether its port is busy or quiet. The provider carries the risk that a genuinely busy unlimited customer consumes expensive upstream capacity. The two sides need traffic measurement they both trust. Q.Net's customer portal provides traffic and balance information, while its questions page explains account-level traffic visibility and the balance threshold. (Q.Net questions and answers, Q.Net online payment)
Prepayment improves the provider's working-capital position. The published terms define a disconnection threshold equal to 30 days of recurring address, port and other periodic charges. Service can be suspended when the balance falls below that threshold and resumes after the customer tops up. This limits the accumulation of receivables, although it can make the service less forgiving for a customer whose payment process is slow. A provider this small has a strong reason to prefer cash before traffic: one unpaid high-capacity account can otherwise absorb both wholesale cost and engineering attention.
The most consequential limitation appears in the technical conditions. The stated access speed is guaranteed only as far as the provider's node. It is not a guarantee of throughput to every destination on the global Internet. That is normal network engineering, but it defines the product boundary. The customer buys a local access path plus whatever quality the provider's upstream and peering arrangements deliver beyond it.
Address space is useful inventory, not a profit statement
The current routing footprint is substantial enough to matter. RIPEstat observed AS25534 announcing seven IPv4 routes representing 5,120 unique addresses on 10 July 2026. The visible set includes the 217.15.176.0/20 and 185.146.16.0/22 aggregates and more-specific routes inside them. It observed no IPv6 announcement. AS61185 announced one IPv4 route, 185.146.18.0/23, covering 512 addresses, also with no IPv6. Because that /23 sits inside the /22 associated with the same organisation, it should not be added again when estimating the company's unique visible stock. (RIPEstat AS25534 routing status, RIPEstat AS61185 routing status, RIPEstat announced prefixes for AS25534)
The tested routes have a sound security signal. RIPEstat's route-origin validator reports the 217.15.176.0/20 announcement by AS25534 as valid. It also reports the 185.146.18.0/23 announcement by AS61185 as valid. A valid route-origin authorisation helps networks reject an accidental or malicious announcement with the wrong origin. It does not protect customer accounts, stop an outage, prove path diversity or guarantee that every more-specific route is configured correctly. (RPKI result for AS25534, RPKI result for AS61185)
The economic value of the IPv4 stock is real but easy to exaggerate. RIPE NCC exhausted its general IPv4 pool in November 2019. A qualifying member that has never received an allocation can now seek only a /24, or 256 addresses, from a waiting list when space is recovered. Larger needs are met through transfers or address-sharing technology. Existing clean address space therefore gives an operator flexibility that a new entrant may lack. (RIPE NCC IPv4 run-out explanation)
Scarcity does not create demand. At the posted KZT22 daily rental, one address produces KZT8,030 of annual list billing before VAT if rented every day. Multiplying that price by all 5,120 unique visible addresses gives a theoretical ceiling of about KZT41.1 million a year. That is not a revenue estimate. Network, broadcast and infrastructure addresses are not all billable; some customers use private space; utilisation is unknown; discounts may apply; support and abuse handling cost money; and a public address is usually attached to a wider access service. The calculation shows the limit of resource value, not the likely result.
The two autonomous systems deserve the same discipline. Operating an AS gives 10G.KZ control over routing policy and makes customer BGP support credible. A second AS can separate services or histories. It does not automatically create resilience. RIPEstat's current neighbour view shows AS61185 with only AS25534 on its upstream side. If AS25534 loses all external reach, AS61185 has no independently observed path around it. (RIPEstat neighbours for AS61185)
Nor is resource holding free. RIPE NCC's 2026 charging scheme sets the annual LIR contribution at EUR1,800, with additional charges for relevant independent resources and AS assignments. That fee is modest beside network construction, but it is fixed, euro-denominated overhead for a company billing customers in tenge. Registry work, security contacts, route records and abuse response also require staff time. (RIPE NCC charging scheme 2026)
The missing IPv6 announcement is more strategically important than the annual fee. IPv4 remains commercially useful, but dependence on it can push growth toward address sharing and rental. IPv6 would give the company a long-term addressing path, improve its position with technically demanding customers and reduce the impression that the network is being maintained mainly as a legacy asset. No public IPv6 route was visible in the reviewed routing data. That is not evidence that the company has no IPv6 allocation or private capability; it is evidence that the two public systems were not visibly originating IPv6.
List prices show a recurring model, not realised economics
The Q.Net tariff card makes useful arithmetic possible. At the prices posted on 10 July, a 100 Mbps unlimited traffic charge plus the 100 Mbps port costs KZT1,965 a day. Over 365 days that is KZT717,225 before VAT. The 1 Gbps combination costs KZT9,400 a day, or KZT3.431 million a year. A 10 Gbps port alone costs KZT730,000 a year, but the page does not publish a corresponding 10 Gbps unlimited traffic price. A 10 Gbps customer's full bill therefore cannot be inferred from the port line.
Q.Net announced that the 100 Mbps port rental will rise from KZT480 to KZT500 a day on 1 August 2026. That is a 4.2% increase for this component and lifts the annual 100 Mbps combination by KZT7,300 if the traffic price stays unchanged. Kazakhstan's annual inflation was 10.3% in June 2026, while the National Bank's base rate stood at 17% from 8 June. One tariff line moving by less than headline inflation does not prove a margin squeeze, because traffic, labour and equipment costs move differently. It does show why price discipline matters when financing and replacement capital are expensive. (Q.Net tariff notice, Kazakhstan June 2026 inflation, National Bank base-rate decision)
The 1 Gbps list price looks attractive only after the cost stack is known. The provider must pay for the fibre or leased circuit to the building, optical equipment at each end, aggregation capacity, router ports, upstream traffic, power, monitoring, billing, tax, support and periodic replacement. Some of those costs are fixed; some rise with usage; some arrive in foreign currency. The gross margin on an idle on-net 1 Gbps port can be very different from the margin on an off-net port running close to capacity.
The connection fee does not necessarily recover construction cost. KZT40,000 for an on-net fibre installation may cover a short drop, technician time and an optic. KZT150,000 for an out-of-coverage building may be only a starting price. Building permission, ducts, poles, road crossings and landlord charges can cost much more. If the customer signs for several years, the provider can recover the gap through recurring margin. If the contract is short or easy to cancel, the provider is financing the customer's location.
Metered traffic can protect that margin, but it can also deter use. At KZT0.03 per megabyte, one decimal terabyte costs KZT30,000. The 100 Mbps unlimited traffic line of KZT1,485 a day equals the metered price of about 49.5 gigabytes per day, roughly 1.5 terabytes over 30 days. The 1 Gbps unlimited line equals about 296.7 gigabytes per day, roughly 8.9 terabytes over 30 days. Customers below those levels may prefer metering; heavy users may prefer unlimited service. The provider needs the mix to price peak capacity, not just monthly volume.
Public financial visibility is the central weakness. The counterparty record reports tax payments of KZT12.44 million in 2024 and KZT11.13 million in 2025, a 10.5% decline, with KZT8.03 million displayed for 2026 at the July observation point. Tax payments are not revenue, profit or cash flow. They can include several taxes and timing effects, and a partial 2026 figure must not be compared with a full year. The series shows continued fiscal activity, not commercial returns. (10G.KZ counterparty record)
The same record's zero-to-five employment band is more directly relevant, though still incomplete. A small legal entity can use contractors, related companies, outsourced field teams and supplier support. The figure does not count those people. It nevertheless raises a diligence question: who monitors the network, handles night incidents, visits customer sites, manages routes, bills customers and covers holidays? If the answer is a tightly experienced team, the company can be efficient. If the answer depends on one person, the low overhead is also a continuity risk.
Revenue growth, if disclosed, would still be limited public evidence. A provider can grow sales by reselling more transit at thin margins or by funding long local builds for customers that can leave. Value creation requires contribution after direct network cost, cash collection, maintenance capital and a return on the equipment and routes committed. The missing measures are recurring revenue, gross profit by service, active ports, average traffic, customer acquisition cost, contract duration, churn, bad debt, top-customer exposure, maintenance spending and free cash flow.
Suppliers hold much of the economic leverage
Current observed routing places two external networks on the upstream side of AS25534: Open Media Group and HOSTKEY. The public RIPE policy record, last modified in 2018, instead names TransTelecom and Smartnet as upstreams and lists several peers and downstreams. The difference may reflect normal commercial change or incomplete observation. It also shows why a current route view and a contractual network map matter more than an old registry description when assessing resilience. (RIPEstat neighbours for AS25534, RIPE AS25534 record)
Open Media is a credible local supplier and competitor. It advertises business Internet from 100 Mbps to several gigabits, dedicated links in Almaty and Astana, 24-hour monitoring and support, and a 99.9% service level. Buying from Open Media can give 10G.KZ local reach without duplicating every backbone investment, but it places another commercial layer between 10G.KZ's customer and national or international capacity. (Open Media business network, RIPEstat neighbours for AS25534)
HOSTKEY's appearance as an observed upstream can diversify commercial and route exposure relative to a single Kazakhstan wholesaler. It does not prove physical path diversity out of 10G.KZ's Almaty site. Two routes can share the same building entry, metro fibre, border crossing or power supply. Customers buying resilience need a diagram of physical paths and failure domains, not just two network names. (RIPEstat neighbours for AS25534)
Data-centre evidence is similarly bounded. PS.kz's Almaty data-centre page lists 10G.KZ among the local providers connected to its facilities, alongside national and city networks. That supports interconnection presence. It does not show that 10G.KZ owns the building, rents a particular rack, has a second site or stores customer data there. PS.kz also markets colocation, dedicated servers and administration directly, so it can be supplier, meeting point and substitute at the same time. (PS.kz Almaty data centre)
Supplier leverage reaches the customer premises. The published Q.Net conditions say equipment installed at a subscriber location can remain the provider's property and must be returned when service ends. That gives 10G.KZ control over standardisation and replacement, but it ties up capital in devices spread across buildings. Spares, truck visits and obsolete optics are part of the cost even when the headline product is only a port.
Customer concentration is unknowable and therefore material
No public source reviewed provides an active customer count, a named customer list or a distribution of contract value. The four-digit example in the Q.Net contract-number format is not a subscriber count. The online payment page proves that customers can replenish accounts by card; it does not reveal how many accounts are live. The absence of disclosed references does not mean there are no substantial customers.
The likely economic customer is clearer than the count. A household seeking ordinary Internet can buy a national bundle. A technically capable business, hosting operator, gaming venue, office centre or network user may value a public address, BGP support, high upload capacity, traffic visibility or a direct fibre arrangement. The Q.Net requirements for SFP optics at 1 Gbps and 10 Gbps point toward customers able to manage business-grade equipment. They do not exclude individuals.
Concentration can make a small network look better before it makes it fragile. One large 10 Gbps user can justify an upstream upgrade and generate meaningful recurring cash. It can also dictate price, create a sharp traffic peak and strand capacity when it leaves. A building owner can open access to many tenants; losing that building agreement can remove them together. A reseller or hosting customer can appear as one contract while representing many end users, concentrating credit risk even when demand beneath it is diverse.
The correct diligence measures are top-one, top-five and top-ten shares of revenue and gross profit; contract expiry dates; cancellation rights; traffic at peak hour; receivable days; and assets dedicated to each account. Customer concentration should be measured after pass-through traffic cost. A large account with high revenue and almost no contribution can be less valuable than ten smaller on-net customers that rarely call support.
Contract durability is also unclear. The posted standard form contains an initial end date in 2016 and then annual automatic renewal unless either side gives notice. Its continued publication in that form makes it a weak guide to a new 2026 agreement. The current commercial schedule, service credits, termination period and minimum commitment need to be read in the executed contract, not assumed from a legacy template. (Q.Net standard contract)
Real substitutes put a ceiling on price
10G.KZ does not compete only with another small ISP offering the same port. It competes with several ways of solving the customer's problem.
Beeline Business advertises an office package with fixed Internet at up to 300 Mbps, mobile lines, a virtual telephone exchange and router rental. The full listed price for the three-line version is KZT28,800 a month before VAT, with a 50% promotional price for qualifying on-net customers through the end of 2026. Q.Net's 100 Mbps unlimited traffic and port total KZT58,950 over a 30-day month before VAT. These are not equivalent services: one says “up to,” one separates traffic and port, and neither public page fully describes contention, upload rate or service credits. The comparison still shows what a price-sensitive office sees. 10G.KZ must earn a premium through dedicated performance, routing flexibility, local response or difficult-building access. (Beeline Business package, Q.Net tariffs)
Kazakhtelecom represents the scale alternative. Its audited 2025 consolidated statements reported KZT568.2 billion of revenue, KZT129.5 billion of gross profit and KZT1.405 trillion of assets. Those group figures include far more than fixed access and should not be compared directly with 10G.KZ. They illustrate the difference in purchasing power, network breadth and ability to finance upgrades. (Kazakhtelecom 2025 audited results)
Open Media is the closer operational substitute. It sells business and operator links, advertises multi-gigabit capacity and serves both Almaty and Astana. A customer that values a local specialist can ask both companies for a design. A carrier needing wholesale capacity can also buy from Open Media directly rather than through 10G.KZ. The current observed routing relationship means 10G.KZ may sometimes buy reach from the same company against which it competes.
PS.kz offers another route. A customer can colocate equipment or rent a server in its Almaty data centre and buy connectivity there, reducing the need for a 10 Gbps circuit to its own premises. A cloud-first customer can move applications into PS Cloud and purchase a smaller resilient access line. That shifts spending away from access capacity but can increase the value of low-latency, reliable connectivity. 10G.KZ benefits from cloud adoption only if it remains the preferred path to the cloud.
Mobile and satellite access create a backup substitute rather than a full equivalent. Kazakhstan had 19.09 million mobile subscribers with Internet access and 3.325 million fixed Internet subscribers at the end of 2025. A business may use mobile connectivity for failover or for a small site where fibre construction is uneconomic. It will not receive the same public-address, routing and sustained-throughput characteristics by default. The existence of an easy backup nevertheless raises the standard for what a premium fixed line must deliver. (Kazakhstan communications statistics for 2025)
The market is growing, which helps all credible providers. Kazakhstan's communications-service volume rose from KZT1.340 trillion in 2024 to KZT1.475 trillion in 2025, while the physical-volume index for Internet services reached 112.7%. Growth does not guarantee attractive margins for a small operator. It can invite more fibre builds, promotions and bundled offers. 10G.KZ needs differentiated demand, not merely a rising national total.
Regulation and geography raise the cost of independence
Internet access in Kazakhstan is a regulated communications service. The communications law establishes the rights and obligations of providers and users, while the service rules require Internet access to be provided under a contract and customer relations to operate in Kazakh and Russian. A 2026 law also updates licensing conditions for Internet access, IP telephony and trunked radio. The exact effect on 10G.KZ depends on its licences and services, which were not disclosed in the public material reviewed. (Kazakhstan law on communications, communications service rules, 2026 telecommunications law notice)
Compliance has direct cost. Kazakhstan's rules require network operators to fund technical capabilities for lawful operational-search and counter-intelligence functions. Personal-data law imposes collection, processing, protection and cross-border transfer duties. A small provider must maintain the necessary equipment, records, access controls and response procedures even though the cost is spread across far fewer customers than at a national carrier. (network requirements for lawful functions, Kazakhstan personal-data law)
The geography of international capacity also limits independence. A World Bank sector assessment described Kazakhstan's international links through Russia, China, Uzbekistan, Kyrgyzstan and Turkmenistan and identified restrictions in the cross-border connectivity market. A new 370-kilometre subsea fibre project between Kazakhstan and Azerbaijan is intended to add an alternative Europe-Asia route and improve resilience. The designated principal operators are Kazakhtelecom and AzerTelecom International. (World Bank digital-sector assessment, Kazakhstan government update on the Caspian cable)
That cable could help 10G.KZ if it creates genuinely competitive wholesale capacity and a path that avoids common terrestrial risks. It could do little for the company's margin if access is expensive, available only through dominant wholesalers or still shares the same Almaty facilities and local fibre. The strategic benefit depends on purchase terms and physical routing, not the existence of a national project.
Financing conditions add another constraint. A 17% national base rate makes debt-funded network expansion demanding. Imported routers, optics and servers expose a tenge revenue base to foreign-currency replacement costs. The company can protect itself through customer prepayment, supplier terms, conservative expansion and price review clauses. It cannot remove the need to replace equipment eventually.
This is why resource control and capital discipline must be considered together. The company already holds more visible IPv4 space than a new RIPE member can readily obtain. Buying more addresses or equipment without contracted utilisation would turn a strategic asset into idle capital. Refusing all investment would preserve cash temporarily while service quality and technical relevance decay. The right investment is a port, route or local build tied to durable contribution, with enough spare capacity for resilience but not for status.
Informal signals show continuity, not momentum
The unofficial public footprint is sparse. A third-party payment service still identifies Q.Net and 10g.kz as an Almaty Internet service, but provides no customer volume or performance evidence. It is useful for continuity and discoverability, not for service quality or market share. (Q.Net payment listing)
The stronger signal is the company's own operational behaviour. Q.Net posted tariff notices in November 2025 and June 2026, acknowledged an electronic VAT invoice error in February 2026 and maintains online card payment and account functions. That is consistent with a live subscriber business handling routine billing. It does not establish growth, satisfaction or technical performance. (Q.Net news, Q.Net online payment)
The site also carries legacy features: a standard contract with a 2016 initial end date, technical conditions whose filename points to 2023, and a network registry policy last modified in 2018 that differs from current observed upstreams. Old documents can reflect a stable service rather than neglect. They can also increase sales friction because a customer cannot tell which terms are current. Updated contracts, a current service-level schedule and a current interconnection description would be inexpensive evidence of operating discipline.
No broad body of customer discussion was found that could support a conclusion about reputation. That absence should not be converted into praise or criticism. Small business connectivity is often purchased privately and discussed only during incidents. The appropriate response is direct reference checking, not sentiment invented from silence.
The facts that would change the judgment
The first decisive fact is contract-level contribution. 10G.KZ should be able to show recurring access, traffic, address and support revenue by customer cohort, less upstream capacity, third-party local loops, facility charges and direct field work. A growing top line with shrinking contribution would confirm price-taker status. Stable or rising contribution per active port would show that customers pay for differentiation.
The second is utilisation. The useful measures are active 100 Mbps, 1 Gbps and 10 Gbps ports; peak and average traffic; committed upstream capacity; headroom at each aggregation point; and the proportion of visible address space actively assigned. A 10 Gbps port sold at a low base price has little value if traffic is sporadic and support-intensive. A well-used port under a multi-year commitment can anchor network investment.
The third is retention and concentration. Renewal rates after price changes, average contract age, top-customer shares of revenue and gross profit, and cancellation rights would show whether recurring billing is durable. The judgment improves materially if no customer, reseller or building can remove a damaging share of contribution. It weakens if one high-traffic account funds an upstream commitment that cannot be reduced when the account leaves.
The fourth is costed resilience. Evidence should identify independent upstream contracts, separate fibre routes and building entries, power arrangements, router redundancy, spare optics, escalation coverage and measured restoration times. AS61185 needs an independent path if it is presented as resilience rather than service separation. A second network number on the same router and upstream does not change the failure domain.
The fifth is a credible staffing model. The zero-to-five employment band can support an efficient local operator if contractors, on-call coverage, documentation and succession are robust. It becomes a warning if one engineer controls routing, customer history and emergency response. Buyers should ask who can make changes at night, who covers leave, how supplier escalations work and how often recovery procedures are tested.
The sixth is capital and cash conversion. Current accounts should separate operating profit from equipment purchases, customer-funded construction and maintenance replacement. Positive cash after tax, working capital and recurring maintenance would show that the network funds itself. Growth financed by customer prepayments while old equipment ages would not.
The seventh is commercial evidence for the resource stock. Address utilisation, address-related revenue, abuse workload and customer demand for BGP would show whether registry status contributes more than administrative cost. Public IPv6 deployment would demonstrate that the company is using scarce IPv4 as a bridge, not treating scarcity as the whole strategy.
The eighth is documentation of the legal and operating boundary. Executed customer forms should name the provider; current licences should match the services; the Q.Net brand relationship should be clear; and rights to fibre, equipment and facilities should survive a supplier dispute. The older network history should be reconciled with the 2015 company so customers and creditors know which assets and obligations sit where.
A specific positive pattern would change the present view: at least several dozen diversified business contracts, including a meaningful group of multi-year 1 Gbps or 10 Gbps customers, producing positive contribution after wholesale capacity and local-loop cost; low churn after the August price change; no severe customer concentration; independent physical upstream paths; documented round-the-clock response; and positive cash after equipment replacement. The exact customer number can vary with contract size, but all parts of the pattern must hold together.
The negative pattern is equally specific: one or two customers carrying most gross profit, frequent discounting to fill ports, traffic growth without repricing, a shared physical path behind nominally different upstreams, overdue router refresh, no tested staff cover and customer contracts shorter than supplier commitments. Those facts would make the company an infrastructure price-taker even if reported revenue were rising.
Control is valuable only when the contracts are better than the costs
10G.KZ has more substance than a name in a member list. It is a registered Almaty company with a live access offer, current billing activity, two operating autonomous systems, a meaningful IPv4 footprint, valid route-origin protection and observable upstream reach. Its Q.Net terms expose a commercially intelligible model: connect buildings, rent ports and addresses, charge for traffic, take payment in advance and support customers that need more routing control than a mass-market bundle provides.
The same evidence sets the limit of confidence. Service is publicly restricted to Almaty. The legal entity sits in the smallest employment band. Current revenue, profit, cash, customers and utilisation are undisclosed. One system depends on the other in observed routing, no IPv6 route is visible, and the published network and contract documents carry old elements. Larger competitors can bundle more at lower apparent prices, while local specialists and data centres offer credible alternatives.
Who pays is therefore narrow: organisations for which local fibre, public addressing, BGP, high sustained throughput or direct technical access is worth more than a national bundle. Who benefits can be balanced: customers receive flexibility, while 10G.KZ earns recurring value from assets and knowledge already in place. Who carries the downside is less balanced. The company commits to capacity, equipment and readiness; the customer bears the business interruption when the design is not genuinely diverse; wholesalers usually cap their exposure through contract terms.
The present judgment is cautious rather than dismissive. Resource-holder status gives 10G.KZ an option to serve differentiated demand. It is not proof that such demand exists at a profitable price. The management task is to avoid spending like a cloud provider while proving that local control earns more than commodity access margins. Until contract contribution, concentration, utilisation, physical diversity and maintenance cash are visible, the company should be treated as a capable small network whose value remains unproven, not as a scaled infrastructure platform.