The global name is the opening bid, not the proof

Tianhai Global Network sits in a difficult commercial category: it looks more sophisticated than a local access reseller, but the public evidence does not yet support treating it like a fully documented enterprise carrier. Its core network identity is clear enough. The operating page at https://net.tianhai.info/ says Tianhai Global Network, abbreviated THGN, is AS4842 and is operated by Tianhai InfoTech. PeeringDB lists the network as Tianhai Global Network, also known as THGN, with AS4842, the AS-set AS-TIANHAI, a global scope, an open peering policy, and a modest 1-5 Gbps traffic band. APNIC RDAP lists AS4842 under the China country code with the AS name TH-AS-AP, the registrant Tianhai InfoTech, and contact details at a Changsha address. RIPE public records add Changsha Tianhai Information Technology Co., Ltd. as a China LIR with the unified social credit code 91430104MA4R49JR7U, plus related RIPE-side Tianhai records in the United Kingdom and Hong Kong.

That is enough to show a real number-resource and routing operation. It is not enough, by itself, to show a conventional commercial ISP serving large Chinese enterprises. The economic problem is credibility. A customer buying cross-border connectivity from China into Hong Kong, Japan, North America or Europe is not only buying reachability. It is buying a promise that the seller can keep routes stable, answer abuse and support mail, obey the relevant regulator, document the legal counterparty, manage upstream dependence, and restore service when a small routing mistake becomes a customer incident. A global-sounding name increases the burden of proof because the buyer immediately asks which company is global, where it is incorporated, where the network equipment is located, and whether the public map reflects controlled capacity or lightly rented presence.

Tianhai's public footprint is unusually transparent in some places and unusually thin in others. The network page publishes ASN, prefix limits, peering policy, exchange addresses and NOC mailboxes. PeeringDB publishes six exchange entries and four listed facilities. IPinfo and BGP.Tools show upstream, peer and downstream relationships. RIPE and APNIC publish resource-holder and contact records. A personal profile for Soha Jin links the network to an operator history and describes Tianhai Global Network as a research network with points of presence in Asia, Europe and North America. Chinese and English public traces also show the network being used in education and IPv6 experiments.

The thin part is the commercial surface. The public web pages do not present a rich catalogue of enterprise WAN, MPLS, DIA, cloud-connect, managed-security or service-level products. They do not publish customer references, uptime history, direct sales terms, mainland telecom authorisations, cross-border circuit arrangements, or audited traffic. The root website at http://www.tianhai.info/ is a simple "It works" page with an ICP filing line; the richer network page is a peering and technical information page. That contrast matters. Tianhai may be a credible technical network and LIR-related operation without being a broad enterprise-connectivity provider. It may also have private customers that are not visible publicly. The market should not collapse those possibilities into one story.

The best reading is that Tianhai is a small, developer-led, cross-border network with serious number-resource literacy, enough global peering to be interesting, and a visible China operating identity. The cautious reading is that its enterprise value depends less on the existence of AS4842 than on whether Tianhai can turn a research-and-resource network into a service that a paying buyer can diligence, contract with and rely on.

Identity starts in Changsha, but the records do not stay there

The company identity has to be reconciled before the network can be priced. PeeringDB's organisation record names Tianhai InfoTech, gives http://tianhai.info as the website, places the organisation in Jinhua, Zhejiang, and lists two networks: Tianhai Global Network, AS4842, and Tianhai Europe Network, AS209417. APNIC RDAP for AS4842 uses a different operating address: Rm. 104, Bldg. 7A, 19 Mufeng Rd., Yuelu District, Changsha, Hunan Province, China. The APNIC registrant is Tianhai InfoTech, the administrative and technical role is Tianhai InfoTech NOC, and the contact mailboxes use tianhai.info. APNIC RDAP for 2401:20::/32 repeats the same Changsha contact and names the netblock TH-CN.

RIPE public records add the more formal Chinese corporate name. ORG-CTIT3-RIPE is Changsha Tianhai Information Technology Co., Ltd., country CN, LIR type, registration number 91430104MA4R49JR7U, and a Changsha address at 19 Mufeng Rd. The RIPE role object THIT-RIPE uses Tianhai InfoTech NOC and the same abuse mailbox. A Chinese company-data page for the same registration number identifies Changsha Tianhai Information Technology Co., Ltd. as established on February 3, 2020, with Jin Shaohai as legal representative and registered capital of RMB100,000. That page is a secondary company-data source, not a substitute for an official filing, but it usefully matches the RIPE registration number.

There are also non-Changsha Tianhai records. RIPE's ORG-TIT2-RIPE is Tianhai Infotech Ltd, country GB, registration number 11782198, address at Kemp House, 160 City Road, London, and remarks describing "a small group of developers" founded in 2013. AS209417, named TIANHAI-EU, is described in RIPE as "Tianhai Experimental Network" and its remarks point back to Tianhai InfoTech, www.tianhai.info, Tianhai Global Network / AS4842, and net.tianhai.info. RIPE's ORG-THIT2-RIPE is Tianhai InfoTech Limited in Hong Kong, registration number 77497759, with an address in San Po Kong. AS57175 is tied to that Hong Kong organisation under the AS name SAKURA-AS and appears in public routing sources as a downstream or related Tianhai network.

These records do not prove one consolidated corporate group. They prove a family of related Tianhai resource and network identities across China, the United Kingdom and Hong Kong. That distinction is important. A buyer can accept a multi-entity structure when the contracts, resource rights and operating duties are clearly assigned. A buyer discounts the structure when it has to infer the commercial counterparty from PeeringDB, APNIC, RIPE, personal pages and route collectors.

The public trail around Soha Jin helps explain the human operating boundary. Soha Jin's personal page lists Tianhai Global Network, AS4842, from April 2016 to the present, and describes the network as originally built by Tianhai InfoTech as a research network. It says the network currently has points of presence in Asia, Europe and North America and connects to HKIX, Seattle IX, KleyReX and other networks for peering. The same page lists Soha Network, AS209306, and other developer and education projects. That is not a corporate filing, but it is a strong operator signal: the public Tianhai network is closely associated with a technically skilled individual and a small developer community.

For a small network, this can be a strength. Customers often prefer a direct path to the person who understands the route policy. For enterprise buyers, it is also a key-person risk. A route community, NOC mailbox or LIR service is only as dependable as the operational process behind it. If Tianhai wants to sell beyond developer and resource-community circles, the identity section of a contract will matter almost as much as the route map.

The routing file is real, but modest

AS4842 is visible from several independent routing views. RIPEstat's AS overview shows the holder as TH-AS-AP - Tianhai InfoTech and the AS as announced. RIPEstat routing status on July 2, 2026 showed one IPv4 prefix, 256 IPv4 addresses, eight IPv6 prefixes and 554 visible /48 equivalents, with high visibility among RIS peers. Its announced-prefix view showed 45.9.11.0/24 and several IPv6 blocks including 2401:20::/40, 2401:20:1330::/46, 2401:20:2104::/46, 2a0e:aa06::/40, 2a13:1800::/48, 2a13:1800:10::/48, 2a13:1800:80::/44 and 2a13:1800:300::/44. BGP.HE showed one IPv4 origin and nine IPv6 origins, with 20 announced prefixes in its broader view. IPIP.NET similarly showed one IPv4 prefix, nine IPv6 prefixes, 256 IPv4 addresses and a set of IPv6 route descriptions tied to Tianhai in China, Hong Kong and Calgary.

Those counts are small compared with a commercial backbone. They are meaningful for a research, peering and IPv6-oriented network. The network has enough address space and route visibility to act as an internet participant, support downstreams, run anycast or content-adjacent services, and expose live behaviour to other networks. It does not have the public address depth of a mass-market ISP or cloud platform. IPinfo classifies the AS as hosting, shows 256 IPv4 addresses, notes anycast on at least one IP, counts two hosted domains, and lists the network as not currently ranked in its own ranking view. CAIDA ASRank gives AS4842 a rank of 7,623, a customer cone of three ASNs, an AS degree of 14, two inferred customers, 10 peers and two providers. Again, that is a functioning network, not a large carrier.

The exchange footprint explains the "global" claim more than the address inventory does. PeeringDB lists operational exchange entries at HKIX in Hong Kong, Seattle IX, Seattle IX Jumbo, EVIX, KleyReX in Frankfurt and YYCIX in Calgary. The speeds range from 100 Mbps at KleyReX to 10 Gbps at YYCIX, with 1 Gbps entries at HKIX, Seattle IX and EVIX. PeeringDB lists facilities at China Unicom (Hong Kong) Global Center, TGT Hong Kong Data Centre 2, Equinix TY6 in Tokyo and Arrow Calgary DC2. The network's own page lists HKIX, Seattle IX and KleyReX and publishes IPv4 and IPv6 peering addresses. The differences between the network page and PeeringDB are not necessarily contradictions; small networks often update one public surface faster than another. But the mismatch reinforces the need to verify live ports, paid cross-connects and route-server status before relying on the map.

Upstream and peer counts vary by source. IPinfo lists 30 peers, three upstreams and six downstreams, with upstreams including Hurricane Electric, The Constant Company and Free Range Cloud. BGP.Tools' public snippet described AS4842 as peering with 298 other networks and having four upstream carriers, and showed downstreams including TWD2 Education and Research Network, Soha Network, RBB InfoTech, Tianhai InfoTech Limited, Kano Network and Shanghai Luogu Network Technology Co., Ltd. RIPEstat's neighbour view showed 20 observed neighbours, including left-side provider-like relationships with Hurricane Electric, The Constant Company and Free Range Cloud, and right-side customer-like relationships with AS24239, AS209306, AS135671, AS135674, AS151464 and AS57175.

The exact counts matter less than the shape. Tianhai has a small origin footprint, a visible IPv6-heavy routing profile, an open peering posture, and a handful of related or customer-like downstreams. It is more credible as a specialised network, LIR-adjacent resource operator and technical peering participant than as a large China enterprise ISP. Any commercial thesis must start there.

What Tianhai seems to sell, and what it does not publicly prove

The public network page is not a broadband sales page. It is a peering page. It tells other networks how to peer with Tianhai, gives the ASN and AS-set, points to PeeringDB, and exposes NOC, abuse and support mailboxes. It says AS209417 is also part of Tianhai Network but is used for experiment and research only. It also says Tianhai does not provide LIR service publicly, while adding that people can contact Soha to inquire about ASN and IPv6 block service in the RIPE NCC region from a partner. That language is unusually direct. It positions Tianhai as technically open but commercially selective.

That matters for revenue logic. A conventional enterprise carrier sells dedicated internet access, private circuits, cloud connectivity, managed security, DDoS protection and service-level assurance. Tianhai's public pages do not show that bundle. A number-resource and research-network operation earns value differently: through memberships, resource sponsorship, routing support, IPv6 allocations, transit or peering assistance, community trust, and the ability to help technically literate customers obtain and operate resources that would be harder to manage alone. A third-party ZX LIR service page lists Changsha Tianhai Information Technology Co., Ltd. among RIPE LIR partners and publishes pricing tables for APNIC and RIPE IPv6 PA blocks and AS numbers. That is not Tianhai's own price list, and it should not be treated as a direct Tianhai tariff. It is still a market signal that Tianhai appears in the resource-service ecosystem around small-network operators.

The simplest commercial reading is therefore not "regional ISP" in the household-broadband sense. It is a small network service and number-resource platform with China, Hong Kong, North America and Europe routing presence. Its customers are more likely to be technically capable networks, developers, education projects, small AS operators, IPv6 experimenters, resource holders or niche connectivity users than ordinary consumers. The public downstream list supports that reading. TWD2 Education and Research Network, Soha Network, RBB InfoTech, Kano Network, Shanghai Luogu Network Technology and Tianhai InfoTech Limited are not mass-market access brands. They look like research, developer, education or small-network relationships.

For enterprise connectivity, that is both a limitation and a possible wedge. A large enterprise does not usually want a boutique resource sponsor as its primary international network. But a technically sophisticated customer with a narrow need might value Tianhai's specificity: a Hong Kong exchange path, an IPv6 route, a small anycast footprint, a China-adjacent operator who understands research networks, or help making an ASN reachable through a few global points. The business is not about scale first. It is about proving that small scale is deliberate, documented and supportable.

The buyer is another network, not a home subscriber

Tianhai should be analysed through a wholesale and technical-buyer lens. The public evidence does not point to last-mile broadband sales, consumer support desks or a household fibre footprint. It points to AS operators, IPv6 users, research communities, developers, resource holders and small networks that need someone to make global routing work. That buyer is very different from a family comparing fibre plans. It already knows what an ASN is, understands why an IRR route object can break reachability, notices when an IPv6 path hairpins through the wrong continent, and may care more about a direct NOC answer than a glossy service catalogue.

This customer shape changes the economics. A home broadband operator wins by spreading fibre build, customer-premises equipment and call-centre cost across dense streets. Tianhai's public model, by contrast, is closer to concentrated technical trust. One paying resource customer may need careful onboarding, route filtering, LOA paperwork, abuse handling, renewal reminders, and route-leak avoidance. The revenue per customer can be higher than a consumer internet account, but the support burden can also be heavier because the customer is technically capable enough to make unusual requests. A small AS customer may ask for a specific IPv6 delegation size, a specific import policy, an RPKI explanation, a transit handoff through a preferred location, or help reconciling RIPE, APNIC, IRR and PeeringDB records. That is high-touch work.

Tianhai's open peering posture is useful here. Open peering does not mean free transit for everyone; it means the network is willing to exchange traffic where the economics and policy allow. For a small downstream or resource customer, open peering can improve reachability without buying capacity from every large carrier. But it is not a substitute for paid upstream diversity. Route-server peering can improve paths to many networks while still leaving the customer dependent on upstreams for the rest of the internet. Tianhai's ability to explain that distinction is part of its commercial value. Unsophisticated buyers may hear "global" and expect universal carrier-grade reach. Technical buyers know that the value is in route choice, fast correction and accurate policy.

The visible ICP line on the root website also points toward a light public-web presence rather than a marketing-heavy service surface. A company that wanted to sell broad managed connectivity into the Chinese enterprise market would usually publish more buyer-facing pages: industries served, service-level terms, escalation hours, security commitments, certifications, licence references and case studies. Tianhai publishes a minimal website, a peering page and technical route evidence. That is not a bad sign if the business is intentionally narrow. It is a warning sign only if the brand is being priced as a full enterprise connectivity platform.

The right buyer, therefore, is one whose problem is specific enough for Tianhai's strengths to matter. A small education network needing IPv6 reachability, a developer community running an experiment, a small AS looking for resource help, or a network that wants a China-adjacent Hong Kong path may find Tianhai more useful than a large provider's anonymous portal. A multinational enterprise trying to consolidate offices, cloud links, security appliances and compliance reporting would need a much heavier public proof package.

The unit economics are unforgiving

The unit economics of cross-border enterprise connectivity would be difficult for Tianhai unless the company chooses customers carefully. Revenue would come from recurring fees for transit, IPv6 or ASN resource service, hosted routing, anycast, BGP support, remote peering, port access, or a small managed connectivity bundle. The cost base would include exchange ports at HKIX, Seattle IX, EVIX, KleyReX and YYCIX; data-centre presence or remote-hands arrangements in Hong Kong, Tokyo and Calgary; upstream commits to providers such as Hurricane Electric, The Constant Company and Free Range Cloud; RIPE and APNIC membership or sponsorship costs; IP resource administration; abuse handling; route-filter maintenance; compliance paperwork; monitoring; and the labour of answering customers who often know enough BGP to ask hard questions. The margin only works if customers pay for trust, support and route quality rather than raw bandwidth. A buyer who only wants cheap transit can go to a larger carrier or cloud provider. A buyer who needs careful IPv6 routing, a cross-border research path, or resource-management help may pay more per unit because the alternative is institutional friction and engineering time.

Supplier concentration is the quiet risk. If a small network uses a few upstreams and several exchange route servers, route quality can change quickly when one provider changes policy, a port congests, an IX session drops, or a route-server filter misbehaves. For ordinary web hosting, a short impairment may be tolerable. For cross-border enterprise connectivity, the same event becomes a service failure. The public evidence does not show private backup circuits, capacity reservations, SLA-backed upstreams, or physical redundancy between all listed points. Tianhai may have them privately, but a customer cannot assume them from PeeringDB alone.

The flip side is agility. A small route-savvy operator can move faster than a large carrier when the buyer needs a narrow fix. Tianhai's public records show validated abuse contacts, published NOC and support mailboxes, open peering, and a track record in IPv6 experiments. The operator-led model can be effective when the customer base is technically adjacent and the problem is specific. It weakens when the customer base is nontechnical, demands contract-heavy support, or expects carrier-grade reporting every month.

Switching costs would be high for the right customer. A downstream AS that uses Tianhai for reachability must update BGP sessions, route objects, IRR entries, RPKI origin expectations, filters, monitoring, and perhaps customer documentation before moving. A small network relying on Tianhai for IPv6 resources has even more friction because the address plan itself may be tied to the service. Those switching costs can produce sticky revenue if service is reliable. They can also create reputational damage if the provider is unresponsive. In this niche, support is not a soft add-on. Support is the product.

Route quality is the other way Tianhai can earn a premium. The 2020 TWD2 note is revealing because the benefit was not abstract global reach; it was a specific path improvement from Japan toward CERNET2 through Tianhai at HKIX. That is the kind of problem where a small network can create value. If a customer has a few important destinations and Tianhai can steer traffic cleanly through Hong Kong, Seattle, Calgary, Tokyo or Frankfurt, the customer may care less about total traffic band and more about whether the relevant route works. For China-linked technical users, path predictability can matter more than headline bandwidth.

This also explains why IPv6 is central. IPv6-heavy networks often have different economics from IPv4-heavy hosting. Tianhai has only 256 visible IPv4 addresses in several public views, but it has a much larger IPv6 story through APNIC and RIPE allocations. IPv6 space is not scarce in the same way as IPv4, so the value is not merely ownership of numbers. The value is knowing how to allocate, register, announce, secure and troubleshoot them across multiple registries and route collectors. For a small customer, that expertise can be worth more than a raw /48 or /44. For Tianhai, it means the revenue should be tied to service quality and documentation, not just address size.

RPKI and abuse handling become part of the same trust equation. Public views such as IPIP.NET and BGP.Tools show RPKI-valid indicators for several Tianhai prefixes, while APNIC and RIPE records show abuse contacts. These are basic but important signals. A small network that leases resources or carries downstreams can quickly suffer reputational harm if customers announce the wrong prefix, ignore abuse, or let stale route objects linger. The cost of keeping the network clean is labour, not just bandwidth. That cost can be invisible in a port-speed table but decisive in whether the business scales.

Regulation is not an appendix in China cross-border networking

China-based network credibility cannot be judged only by packet paths. Jurisdiction is part of the product. Tianhai's APNIC and RIPE China records point to Changsha, while PeeringDB points to Jinhua, the root website displays an ICP filing line with Zhe ICP Bei 15007961-1, the RIPE-side Tianhai Infotech Ltd record points to the United Kingdom, the Tianhai InfoTech Limited record points to Hong Kong, and exchange/facility records point to Hong Kong, Tokyo, Calgary, Seattle and Frankfurt. That is a broad jurisdictional footprint for a small network.

A buyer needs to know which entity signs the contract, which entity controls the route objects, which entity invoices, which entity handles abuse, which entity holds the RIPE or APNIC relationship, and which law governs disputes. A Chinese enterprise customer would also care whether any service touching mainland users, content or cross-border traffic requires ICP, ISP, value-added telecom, cross-border data-transfer or other telecom-related permissions. The public evidence here shows number-resource registration and an ICP line on the root website. It does not show a public mainland enterprise telecom service licence for cross-border corporate connectivity. That absence does not prove non-compliance; it limits what can be inferred from public sources.

The Hong Kong records are especially important. Hong Kong is a natural place for China-facing international connectivity because it sits at the boundary between mainland demand and global internet exchange. Tianhai's PeeringDB facility entries include two Hong Kong sites, and RIPE records show Tianhai InfoTech Limited in Hong Kong. AS57175 appears as a related Hong Kong network under the Tianhai maintenance boundary. That can be useful because Hong Kong facilities and exchanges make a China-adjacent network more internationally usable. It also creates a diligence question: is Hong Kong merely a route point, a corporate entity, a customer, a downstream, or a separate operating arm? Public routing and registry records point in the direction of relatedness, not a fully explained corporate structure.

The United Kingdom record should be treated with similar care. ORG-TIT2-RIPE says Tianhai Infotech Ltd is a small group of developers founded in 2013 and uses a London address. AS209417 is an experimental network that imports from AS4842, The Constant Company and Amazon, and exports AS209417 to those providers. The Tianhai network page says AS209417 is used for experiment and research only. That makes it inappropriate to use AS209417 as proof of commercial European enterprise service. It is better read as a research and resource trail that strengthens the technical story while adding another legal identity to reconcile.

Market signals point to a technical community, not a mass market

The strongest unofficial signal is the operator ecosystem around Tianhai. Soha Jin's public profile places AS4842 in a long-running personal technical history, connected to Luogu, open-source projects, Soha Network and education or research activity. TWD2's 2020 post says TWD2 Education and Research Network reached a strategic cooperation with Tianhai Global Network after Tianhai became a peering partner with CERNET2, and that TWD2's Japan-node traffic to CERNET2 would pass through Tianhai at HKIX to reduce latency. A 2022 Wandai Blog post says a hardware-router class experiment used AS4842 Tianhai Global Network and AS24239 TWD2 Education and Research Network to provide a global IPv6 testbed, with Soha supporting the internet-connectivity part as AS4842 NOC. Those are not customer references in a commercial-sales sense, but they are useful signals of actual use by technically literate communities.

The LIR-resource chatter points the same way. Tianhai's own page says it does not publicly provide LIR service, while a third-party ZX page lists Changsha Tianhai among RIPE LIR partners and publishes resource-service prices. A separate personal blog post about becoming a BGP player recommends net.tianhai.info for RIPE-related IPv6 and ASN service, contrasting it with another APNIC route. This is informal market evidence, not audited revenue. It suggests Tianhai is known in the small-AS and IPv6 hobbyist or semi-professional community as a place to ask about resources. That reputation can be commercially valuable if Tianhai wants a niche service business. It can be commercially limiting if the company wants to be perceived as a formal enterprise carrier.

There is little visible consumer reputation. That is not a weakness for this kind of network; consumer review sites are not where BGP buyers usually speak. The more relevant reputation questions are whether route objects are maintained, abuse mail is answered, downstreams are filtered correctly, RPKI is kept clean, and peering sessions are stable. Public sources show some positive indicators: validated APNIC contacts, RPKI-valid prefix signals in BGP/IPIP views, public exchange entries, and long-running operator history. Public sources do not show ticket response times, customer churn, traffic graphs, invoice quality, dispute history or incident postmortems.

Competitors and substitutes are larger, clearer and less flexible

Tianhai's competitors depend on which product a buyer thinks it is buying. For ordinary China cross-border enterprise connectivity, the substitutes are China Telecom, China Unicom, China Mobile, PCCW Global, NTT, GTT, Cogent, Lumen, RETN, Telstra, Equinix Fabric, Megaport, Console Connect, local Hong Kong carriers, managed SD-WAN providers and hyperscale cloud networking. Those providers bring scale, contracts, finance teams and formal SLAs. They also bring process, pricing opacity and slower custom work. Tianhai cannot beat them on raw scale. It can only win where a buyer values speed, technical directness, IPv6/resource expertise or a narrow peering path.

For LIR and resource services, the competitor set is different: APNIC and RIPE members offering sponsored resources, small-network consultants, BGP-player communities, cloud VPS providers that allow BGP, and networks such as Hurricane Electric or Vultr that provide accessible routing environments. In that market, Tianhai's China identity and operator reputation may be a differentiator. A Chinese-speaking network engineer may prefer a Tianhai path because it combines familiar language, APNIC/RIPE literacy and a real AS. But the customer still has to ask whether the resource agreement, renewal terms, abuse process and transfer rights are strong enough.

For research and education networks, Tianhai's substitute is often direct institutional connectivity. A university lab or student project can use CERNET, a university partner, cloud credits, a friendly AS operator, or a commercial VPS provider with BGP. Tianhai's advantage is that it appears to have done this kind of work before. Its risk is that research users may not pay enough to sustain a multi-continent footprint unless the network also has paying resource or connectivity customers.

The strategic question is therefore not whether Tianhai can become a national carrier. The public evidence does not suggest that ambition. The question is whether Tianhai can preserve a credible niche: China-linked, technically literate, IPv6-forward, cross-border enough to be useful, and small enough to give customers direct attention. That is a real position, but it requires discipline. Listing more global points than the business can support would damage trust faster than it creates revenue.

What would change the judgement

Several facts would raise confidence. First, a clear corporate map showing how Tianhai InfoTech, Changsha Tianhai Information Technology Co., Ltd., Tianhai Infotech Ltd in the United Kingdom, Tianhai InfoTech Limited in Hong Kong, AS4842, AS209417 and AS57175 relate to one another. Second, public service terms that identify the contracting entity, renewal rules, abuse handling, resource-transfer rights and support hours. Third, a current network status page or looking glass that confirms live points, upstreams, route communities and incident history. Fourth, licence or filing evidence for any mainland China commercial telecom service beyond website hosting or resource administration. Fifth, customer or community references that distinguish paid commercial use from research participation. Sixth, facility or exchange evidence that separates active ports from historical or low-usage presence.

Facts could lower confidence too. If the PeeringDB facilities are stale, the global map becomes weaker. If most downstreams are related experiments rather than independent paying customers, the customer-cone story becomes less commercial. If the LIR-service market is mostly informal and low-priced, support burden could exceed revenue. If the Hong Kong and UK entities are merely resource labels without operational contracts, buyers will discount the structure. If abuse mail is slow, the network's reputation cost can rise quickly because small hosting and resource networks are judged heavily by abuse response.

The current judgement is balanced. Tianhai Global Network is not a paper network. It has a visible AS, APNIC and RIPE records, IPv6-heavy route visibility, public peering, a small set of facilities and exchange points, and a credible operator-community history. But the same public record shows a company whose commercial identity is less mature than its routing identity. A buyer should price Tianhai as a specialised technical network that can be useful in China-linked cross-border and resource contexts, not as a conventional enterprise carrier until the contracts, licences, operating duties and live capacity are made as visible as the ASN.

Evidence register

The following public sources anchor the analysis: