A search that finds everyone but the subject

Run the experiment that any prospective customer runs. The first Teradata the internet offers is Teradata Corporation of San Diego, ticker TDC on the New York Stock Exchange, which reported $421 million of revenue for the fourth quarter of 2025 and $1.522 billion of annual recurring revenue when it published results this February. The second, for an Indonesian searcher, is PT Tera Data Indonusa Tbk, the maker of Axioo laptops, which raised Rp145.62 billion in its 2022 stock-exchange debut under the ticker AXIO — a listing so oversubscribed that the shares hit the daily price ceiling on their first day of trading. The third is Teradata Megah Corporation of Bandung, a banking-technology consultancy that dates itself to 1992 and holds the teradata.co.id domain, the most natural Indonesian corporate address in the namespace.

Only then, if the searcher is persistent or already knows the exact address, does the fourth claimant appear: PT Teradata Bintang Selaras, an internet service provider operating from a shophouse at Jalan Kapuk Raya No. 1A in Cengkareng, West Jakarta, whose corporate site describes a company founded in 2019 selling dedicated internet, broadband, colocation and network services to businesses and institutions. Let the record state plainly what the crowded search results obscure: this firm has no relationship whatsoever to the American analytics corporation, to the listed laptop assembler, or to the Bandung consultancy. It is a small, licensed, independently owned Indonesian network operator that happens to answer to the most contested seven letters in its industry.

That contest is the economic subject of this essay, because it is not a curiosity — it is a cost line. A retail ISP acquires customers through search, referral and street presence, in roughly that order of scalability. A firm whose name is owned, in every index that matters, by three larger and older entities starts every marketing month with a handicap that compounds: paid search against a US software giant's brand terms is unwinnable, organic search is buried under a stock ticker, and even a procurement officer who has met the company must push past two pages of the wrong Teradata to verify it exists. What makes this case worth a close read is that the company's own paper trail shows it understood the problem early and engineered around it — not by renaming itself, which costs legal money and licence re-papering, but by splitting itself in two. The name that signs the contracts is not the name that sells the product, and the arithmetic of that split tells you more about brand economics at the bottom of the ISP market than any trademark textbook.

The paper trail runs through Kapuk Raya

Identity first, because with four claimants to one name nothing else can be trusted until the entity is pinned. The Indonesian internet association APJII lists PT Teradata Bintang Selaras in its member register as member number 1476, category ISP, city Jakarta Barat — and, tellingly, records its business brand not as Teradata but as TABISNET, with the member domain tabisnet.co.id. The national internet registry corroborates: IDNIC's record for the address block 160.22.206.0/23 shows it registered to PT Teradata Bintang Selaras on 26 June 2024 under the network name IDNIC-TABISNET-ID, with an administrative contact reachable at an @tabisnet.co.id address and a published geolocation feed hosted on the same domain. Global routing tables saw the company's two /24 blocks appear on 4 July 2024, with the aggregate announced from that October. The company's PeeringDB entry was created on 15 July 2024. PANDI, the .id domain authority, shows tabisnet.co.id registered on 31 May 2024 and teradata.net.id on 24 December 2024, the latter through APJII itself as registrar — dates verifiable through the registry's public lookup service, queried for this essay on 3 July 2026. The Wayback Machine's index, consistent with all of this, first captured the TabisNet site in late July 2024 and the Teradata corporate site in January 2025.

Three registry queries failed, and the failures belong in the record. The corporate registry search at the Ministry of Law's ahu.go.id refused the connection outright on the day of writing, so the notarial founding date of the PT could not be independently confirmed; the 2019 founding year rests on the company's own site. The trademark database at pdki-indonesia.dgip.go.id returned an access-denied error to an automated query for the Teradata mark, so whether the American corporation holds registered Indonesian trademarks — and in which classes — could not be verified either way. And the old licence directory at dittel.kominfo.go.id no longer resolves at all, a casualty of the ministry's reorganisation into Komdigi, which means the firm's telecommunications licence must be read through proxies rather than the primary register.

Those proxies are, fortunately, strong. Indonesian rules reserve .net.id domains for entities holding a telecommunications operating licence from the ministry; registrar documentation spells out that the registration file must include the operating permit. A company that holds teradata.net.id — registered, again, through APJII — has necessarily shown a licence to someone whose business is checking licences. Add the APJII membership with ISP legality recorded, the IDNIC direct membership, and the routed address space, and the identity question closes: this is a real, licensed, currently operating ISP, whatever the search engines say.

One gap remains, and it is analytically interesting rather than embarrassing. The company says 2019; every verifiable trace — domains, addresses, routing, association membership — begins in mid-2024. Five years separate the claimed founding from the network's public birth. The most economical explanation, consistent with how small Indonesian ISPs grow, is that the firm spent those years operating on rented infrastructure: reselling upstream bandwidth under someone else's address space and licence umbrella, invisible to registries because nothing was registered to it. That reading is an inference, not a record, and is flagged as such. But it matters for the economics, because it means the entity now visible is not a startup — it is the moment an established street-level business decided to buy its independence.

Two storefronts, one licence

What the company sells depends on which of its two faces you approach. The corporate site at teradata.net.id speaks enterprise Indonesian: dedicated internet with a 99.5 per cent service-level commitment, broadband for offices, reseller arrangements, colocation, security services and IPTV, aimed at "bisnis, instansi, dan korporat" — businesses, government offices, corporations. No prices appear; dedicated connectivity in Jakarta is a negotiated product. The retail site at tabisnet.co.id speaks another language entirely — "Internet Super Cepat, Murah dan Stabil," super fast, cheap and stable — and publishes a three-line tariff with prices in plain sight. The retail brand's counters claim 5,808 clients, 28 institutional customers, 32 corporate customers and 187 active projects.

The two-storefront structure is the company's revealed answer to the question this essay opened with. TabisNet is, on its face, a compression of the legal name — Teradata Bintang Selaras Net — and everywhere money and infrastructure actually move, TabisNet is the name that appears: on the APJII register, on the IDNIC address block, on the abuse contact, on the geolocation feed, on the retail tariff. The Teradata name survives in exactly two places: the legal shell that signs licences and contracts, and the enterprise-facing domain registered — last of all the company's assets, in December 2024 — inside the one namespace the bigger Teradatas cannot enter. That last point deserves emphasis, because it converts a regulatory formality into a small strategic asset. A .net.id domain requires an Indonesian telecommunications licence; neither the San Diego corporation nor the Bandung consultancy nor the laptop maker holds one. The shophouse ISP therefore owns the only Teradata address in Indonesia's telecom namespace not because it outbid anyone but because the licence gate excluded every richer claimant. Regulation, for once, subsidised the small firm's brand instead of taxing it.

Who pays, then? Three customer layers are visible. The base is retail households in Kapuk and surrounding kelurahan — the counters' 5,808, a figure to be handled with care since it is self-published and unaudited, but directionally consistent with the scale of the address space and the traffic pattern discussed below. The middle layer is the 28 institutions and 32 corporates: offices, schools, clinics and shops buying broadband or low-end dedicated lines, the segment the Teradata-branded site exists to serve. The third layer is other, smaller operators. A post in Indonesia's main neighbourhood-network community on Facebook advertises open partnership under the company's name and its network number — an offer to the RT/RW-net segment, the informal micro-ISPs that wire a few dozen houses at the level of Indonesia's smallest administrative units and buy their bandwidth wholesale from whoever offers the best mix of price and peering. The firm that spent five years, on this essay's reading, as somebody's downstream reseller now recruits downstream resellers of its own. That is the standard ladder of the Indonesian access market, and Teradata Bintang Selaras has climbed exactly one rung of it.

Rp188,000 against Rp250,000: the price list as strategy

The retail tariff is the hardest evidence in this file and rewards close reading. TabisNet's published price list offers three symmetrical, unlimited fiber tiers: LITE 50 at 50 Mbps for Rp188,000 a month, PLUS 150 at 150 Mbps for Rp278,000, and MAX 300 at 300 Mbps for Rp368,000. At exchange rates prevailing this year of roughly Rp16,000 to the US dollar — a conversion offered for orientation only — the entry tier costs about eleven and a half dollars a month.

Set that against the nearest branded comparator with a published national tariff. Biznet, the most aggressive of Indonesia's private fiber challengers, lists its entry package HOME 0D at Rp250,000 a month for 50 Mbps. Same headline speed, two published price tags: Rp188,000 against Rp250,000. The unbranded operator sells the identical retail unit at a 24.8 per cent discount to the branded one — Rp62,000 a month, or roughly Rp744,000 a year per subscriber, left on the table relative to what an established name charges for the same bandwidth in the same city. That observed pair is the closest thing the record offers to a market price for a name, and it anchors this essay's thesis: at the bottom of the Jakarta access market, brand equity is worth about a quarter of the monthly bill, and a firm with no usable brand must hand that quarter back to the customer.

Two qualifications keep the comparison honest. First, part of the gap is structural rather than reputational: Biznet carries a national backbone, television bundles and a large marketing establishment in its price, while TabisNet carries a shophouse and aerial fiber through kampung streets. The pair measures the combined premium of brand and infrastructure depth, and the record does not allow a clean split between the two. Second, TabisNet's symmetry and unlimited quota are genuinely aggressive product terms at this price point; the discount buys the customer more than parity. But both qualifications cut the same way analytically. A firm that cannot monetise a name must compete on the only two dimensions left — price and product generosity — and both are visible in this tariff. The price list is what a brand handicap looks like when it is written down.

The upper tiers tell a second story. The step from 50 to 150 megabits costs the customer Rp90,000; from 150 to 300, another Rp90,000. Priced per megabit, the ladder falls from Rp3,760 a month at the entry tier to Rp1,853 in the middle and Rp1,227 at the top — a two-thirds decline in unit price across the range, which is the signature of a network whose capacity costs are dominated by the drop cable and the port, not the traffic. Bandwidth, once a fiber strand is lit, is nearly free at the margin for the operator — the pricing ladder is a revenue-extraction curve, not a cost curve, and its gentle slope says the company is optimising for upsell volume in a price-sensitive neighbourhood rather than skimming a premium segment. Everything about the tariff — round numbers ending in eight, a lucky digit in the local market; WhatsApp-first sales contact; installation promoted on short-video platforms — describes a business built for the kampung and ruko economy of northwest Jakarta, not for the glass towers eight kilometres southeast where the other Teradatas' customers sit.

The arithmetic of 512 addresses

Now assemble the machine, stating at each step what is evidence and what is inference. The evidence: a published three-tier tariff starting at Rp188,000; a self-published counter of 5,808 clients; an address inventory of exactly 512 public IPv4 addresses and no IPv6, per the registry and measurement records; a regulated levy structure that takes 0.5 per cent of gross telecommunications revenue as the operating-rights fee and 1.25 per cent as the universal-service contribution, both set by ministerial regulation; nine exchange ports and two transit relationships visible in routing data; and a traffic curve that a measurement service classifies as a residential access pattern peaking between ten and eleven at night Jakarta time.

The inference, built conservatively on that floor. If every one of the 5,808 counted clients paid only the cheapest tier, retail revenue would be about Rp1.09 billion a month, or roughly Rp13.1 billion a year — call it $800,000. A blended average nearer Rp220,000, allowing for the upper tiers and the institutional lines, would push the figure toward Rp1.28 billion a month. The counter is the weakest link in this chain: it is marketing copy, not an audited subscriber count, and this paragraph should be read as an upper bound on the retail base rather than a measurement of it. The same site's corporate twin claims "2.5M happy user," a figure five hundred times larger that is best understood as an unedited web-template counter — and which, incidentally, is its own small piece of evidence about how much scrutiny the corporate storefront expects to receive.

The cost side is thinner but bounded. The two regulated levies together take 1.75 per cent of gross revenue — on the floor estimate above, about Rp19 million a month, or Rp3,290 of every entry-tier subscription — before the company buys a single megabit. The percentages are worth respecting even at this scale, because they are the only cost in the structure that scales perfectly with revenue and can never be negotiated down; they are also the mechanism through which the ministry knows, to the rupiah, how big this company really is, even when the public does not. Transit comes from two upstream providers, with adjacencies visible to national wholesalers including NAP Info Lintas Nusa and Parsaoran Global Datatrans; the prices of those contracts are negotiated and unpublished, and no figure for them is asserted here. What the record does show is a company working hard to minimise that line: nine exchange ports mean that traffic to national content and to other Indonesian networks flows at the cost of port fees rather than per-megabit transit, which for a network whose evening peak is streaming video is the difference between a viable margin and none. The 512 public addresses against a claimed 5,808 clients imply roughly eleven subscribers sharing each address through translation at the network edge — standard practice for Indonesian ISPs, and a reminder that the firm's scarcest capital asset cost it an IDNIC membership rather than a spectrum auction. Even the web hosting whispers frugality: both domains sit on the parking nameservers of a budget shared-hosting provider, per the PANDI records. Depreciation on the fiber plant itself — the cables over the streets of Kapuk, the splitters, the customer terminals — is invisible to this method and is the largest unobserved number in the file; aerial builds in dense kampung run cheap per home passed, but "cheap" here is an industry generalisation, not a company figure, and is flagged as exactly that.

Put the pieces together and the shape of the business emerges: perhaps a billion rupiah a month of gross revenue, less than two per cent of it owed to the state off the top, transit disciplined by peering, addresses stretched eleven ways, hosting bought by the year — a machine with no expensive component except the fiber in the air and the technicians who climb to it. Small, but not fragile in the way outsiders assume: the cost structure has almost no fixed rent extracted by anyone with pricing power over it.

Nine exchange ports and two rented roads out

For a company of this size, the interconnection record is unusually rich, and it is where the wholesale ambition shows. PeeringDB lists the network present at nine internet exchanges: IIX-Jakarta — the association-run exchange that comes bundled with APJII membership — plus JKT-IX, DE-CIX Jakarta, EdgeNxT, AIX, OpenIXP, ILIX, each at ten gigabits, a one-gigabit port at Omadata's exchange, and a hundred-megabit port at DE-CIX ASEAN in Singapore. The listed facilities run from APJII's own data centre and ProDC in South Jakarta to an Omadata site in Surabaya, at the other end of Java.

Read economically, that inventory is over-built for 5,808 households in Cengkareng — deliberately. An eyeball network's peering saves money in proportion to how much of its traffic terminates domestically, and Indonesian exchanges are cheap to join; a rational micro-ISP takes every free or near-free port it can reach. But a Surabaya facility presence and a Singapore port serve a different purpose: they are the shop window for the partnership offer in the neighbourhood-network forums. What an RT/RW-net operator in the outer kampung buys from a firm like this is not raw transit — Telkom's wholesale arm or any of a dozen aggregators would sell that — but a blend: domestic routes at exchange cost, international routes through the upstream contracts, address space to borrow, and an operator on the licence side of the regulatory line willing to stand between an unlicensed reseller and the ministry. The nine ports are marketing collateral in that sale, and rather more honest collateral than the "2.5M happy user" counter.

The dependency structure this creates is asymmetric in a way worth stating precisely. Downstream, the company's retail customers can leave with a month's notice; there are no visible lock-ins beyond installation, and the switching cost in a neighbourhood with competing aerial fiber is a technician visit. Upstream, the company depends on two transit providers for every route its peering cannot reach — which in practice means most of the international internet — and on the exchanges' continued low pricing for its domestic cost advantage. None of these suppliers is a monopolist; Jakarta's wholesale transit market is among the most competitive in Southeast Asia, and the count of licensed operators paying the telecommunications levy grew from 609 in 2020 to 1,457 in 2024, most of them ISPs buying inputs from the same wholesale pool. Supplier power over this firm is therefore modest. The real dependency is regulatory: every layer of the structure — the ASN, the addresses, the .net.id domain, the licence that RT/RW-net partners are effectively renting — exists at the pleasure of a ministry whose levy base more than doubled in four years and whose incentive to formalise, inspect and consolidate the bottom of the market grows with it.

The name tax, itemised

Return now to the name, with the business fully in view, and itemise what the collision actually costs. Start with search, the channel where the tax is levied first. The corporate storefront competes for the word Teradata against a NYSE-listed company whose quarterly results alone generate more indexed press in a week than a Cengkareng ISP will generate in its corporate lifetime; against the Axioo maker's investor-relations apparatus and ticker feeds; and against a Bandung consultancy that has had thirty-four years to accumulate backlinks to teradata.co.id. Organic discovery of the enterprise brand by an uninformed buyer is, on the observable evidence of any results page, approximately impossible. Paid search against those incumbents means bidding on a brand term whose commercial intent is dominated by data-warehouse buyers — traffic that would be expensive and almost entirely wasted.

Procurement friction is the second levy. An Indonesian purchasing officer running a routine diligence pass on "Teradata" must distinguish the counterparty from a US corporation with an Indonesian enterprise presence, a Tbk with public filings, and a Bandung firm with three decades of banking references — and the failed public lookups documented above mean the officer cannot resolve the question cheaply from the trademark or company registries either. Every hour of that confusion is a cost the company imposes on its own buyers, which is the most regressive place a brand tax can fall.

The trademark shadow is the third, and the least quantifiable. Whether Teradata Corporation holds live Indonesian registrations in the relevant classes could not be verified — the intellectual-property database refused the query, and that gap is stated here as a gap. What can be said structurally: the American firm polices its mark globally, Indonesia's trademark regime is first-to-file with well-known-mark protections that have favoured foreign plaintiffs in publicised cases, and a telecommunications company using the identical word in an adjacent technology category is not a fanciful target. The one asset the challenger cannot easily take is the licence-gated domain: PANDI's rules put teradata.net.id behind a wall that only a licensed Indonesian operator can cross, so the likeliest legal endgame would contest the name on the sign, not the address on the network.

It is worth pausing on how a collision this bad happens at all, because Indonesian company law manufactures them at scale. Government regulation on company naming — PP 43 of 2011 — and the Ministry of Law's naming terms for the registration system require a wholly locally owned perseroan terbatas to carry a name of at least three words, in Indonesian, not identical or confusingly similar to an already-registered company name. The test is run against the domestic company register, not against the world's trademarks or the world's search indexes. So a founder in 2019 who wanted a name that smelled of data infrastructure could take the most famous data word in enterprise computing, append two euphonious Indonesian words — Bintang Selaras, roughly "harmonious star" — and receive a perfectly legal, perfectly registrable name that no algorithm at the ministry would ever flag, because no other Indonesian PT had taken it. The registry's uniqueness test and the market's findability test are different tests, and the space between them is where thousands of Indonesian companies live, invisible behind global namesakes their incorporation papers never checked for.

Against those three levies, observe what the company has already done, because its own conduct is the best available estimate of the tax's size. It built the retail business — the layer where customer acquisition is search- and referral-driven — under a different name entirely, and put the real counters, the real tariff and the real contact details there. It let the Teradata name persist only where names are cheap: on the legal shell, where renaming means notarial deeds, licence amendments across the ministry, the association and the registries, and the re-papering of every upstream contract; and on an enterprise storefront whose customers arrive by referral rather than search anyway. The revealed calculation is that the name is worth changing at the margin where marketing happens and not worth changing at the core where paperwork happens. That is, incidentally, the correct calculation. A full legal rename would buy the firm nothing its dual structure has not already bought — the retail brand is clean, the licence assets do not care what they are called — and would cost real money and weeks of administrative risk. The only event that flips the arithmetic is a trademark demand against the legal name itself, at which point the deed change happens under duress and the TabisNet brand, already carrying the customers, absorbs the shock. The company has, in effect, pre-positioned its own escape route.

What the neighbourhood chatter is worth

The unofficial record around this firm is thin, young and consistent, and each strand carries a different weight. The retail brand markets where its customers live: a short-video account pushing installation promotions, Instagram posts making the case that stable internet matters, the WhatsApp numbers doing the actual selling. That is exactly the acquisition mix a kampung fiber operator should run, and its existence corroborates the business model more than any single claim on the website. A travel firm's reel documents a staff gathering in Bali around the turn of 2025 — a small signal, but companies that cannot make payroll do not fly their staff to Bali, and the timing sits right after the year in which the firm bought its network independence. The partnership post in the RT/RW-net community group, discussed above, signals that the wholesale layer is live and hunting. And the near-total absence of complaint chatter — no visible outage threads, no consumer-forum pile-ons of the kind that attach to Jakarta's larger ISPs within minutes of a fault — is genuinely ambiguous: it is what excellent service looks like, and also what a customer base that complains in closed WhatsApp groups rather than public forums looks like. The prudent reading takes it as absence of evidence, not evidence of quality.

What would settle each signal is knowable. The subscriber counter would be settled by the annual report every licensed operator files to the ministry with its levy payment — gross revenue, reported under penalty, from which subscriber scale can be triangulated; those filings are not public, but their existence means the truth is one regulatory disclosure away. The Bali signal would be settled by headcount evidence — sustained hiring in the job boards, which at the moment show the company only faintly. The service-quality question would be settled by six months of crowdsourced speed-test and outage data for the network's address space, which accumulates on its own as the base grows. None of these tests is available today, which is itself the finding: the company is young enough, and small enough, that the public record is still mostly what it says about itself, corroborated at the edges by registries that do not take dictation.

Crowded streets, cheap switching

The competitive geography is unforgiving and worth stating without romance. Cengkareng is not underserved territory: aggregator listings for the district count the national brands present — Telkom's IndiHome, Biznet, MyRepublic, First Media, the state electricity utility's Iconnet — before any of the local operators are tallied. Below the brands sits the informal layer the company itself recruits from: hundreds of RT/RW-net micro-operators wiring alleys the trucks cannot reach, any of which can become a competitor or a customer depending on whose wholesale offer wins the month. Above it sits mobile substitution — Indonesian data bundles are among the cheapest per gigabyte anywhere — and, at the exotic margin, satellite service now inside the same levy regime that binds terrestrial operators. Switching costs across all of it are a technician visit and an installation fee.

The customer's side of that ledger explains the churn arithmetic every operator in the district lives with. The national brands acquire with promotional pricing that resets after the introductory months; the local operators acquire with permanently low list prices and same-day service from a technician who lives nearby. A household choosing between a heavily advertised first-year discount and TabisNet's flat Rp188,000 is making a bet about its own patience with post-promotional bills, and the local operator's offer is, in effect, the promise that the price on the banner is the price forever. That is a defensible retail position precisely because it is unglamorous: it cannot be copied by a national brand without repricing the brand's entire base.

In that structure, the firm's defensible ground is exactly two things. First, physical presence: aerial fiber already strung through specific streets, where the marginal cost of connecting the next house is lowest for whoever's cable is already on the pole, and where a shophouse office two kilometres from the customer beats a call centre in a tower. Second, the licence stack: the bundle of permissions, addresses and interconnection that lets it stand upstream of operators too small to hold their own. Neither moat is deep, but neither is decorative — both are the kind of asset that must be rebuilt street by street or filing by filing by anyone who wants to take the position. The risks map onto the same two assets. Operationally, aerial plant in a flood-prone coastal lowland of northwest Jakarta lives one bad monsoon or one municipal cable-tidying campaign away from an expensive month — a general hazard of the district, noted here as context rather than a company-specific record. Regulatorily, the wholesale-to-informals business depends on the ministry's continued tolerance of the RT/RW-net layer; a hard formalisation push would either destroy the downstream market or massively enlarge it, depending on whether the small operators are shut down or forced to buy exactly the licence-fronting service this firm sells. Geopolitics barely touches a business this local; the exchange-rate pass-through on international transit priced in dollars is the widest external channel, and peering discipline narrows it.

What would move the judgement

The judgement as it stands: PT Teradata Bintang Selaras is a real, verifiably licensed, structurally frugal neighbourhood fiber operator of perhaps a billion rupiah a month in gross revenue, which has solved the worst brand collision in its industry not by fighting for a name it cannot win but by quietly moving its commercial life to a name it invented — and the Rp62,000 monthly gap between its tariff and the branded comparator is the visible price of fighting for customers without a name at all. Several discoveries would move that assessment, in either direction.

A corporate-registry extract, when the public search becomes reachable, that contradicts the 2019 founding or reveals ownership links to any larger operator would reopen the identity file — the analysis above treats the firm as independent because nothing in the registries contradicts it, not because independence is proven. A trademark search, once the intellectual-property database answers queries again, showing live Teradata registrations in telecommunications classes held by the US corporation would convert the trademark shadow from a structural risk into a datable one, and would make the legal rename this essay judges unnecessary today a matter of time. Ministry data — or a levy dispute — revealing gross revenue far below the tariff-times-counter estimate would say the 5,808 figure is aspiration rather than approximation, and would shrink the wholesale story with it. Conversely, evidence of address-space growth beyond the current 512, an IPv6 deployment, new exchange ports outside Java, or a quickening cadence of partnership offers in the operator forums would say the wholesale rung of the ladder is bearing weight, and that the firm's next public record will be written in the registries before it is written on any website. That has, after all, been the pattern so far: every important fact about this company appeared in a registry first and in its marketing second. Watch the registries.

Evidence register