The dawn of the commercial internet is, by necessity, a story built from fragments. In 1991, the notion of paying a private company for internet access was still novel, and the language to describe this new service was far from settled. Yet, even then, signs of what would become a global industry were emerging. For analysts seeking to track the birth of commercial internet service providers (ISPs), ISIC 6420—Telecommunications—serves as both a baseline and a puzzle. The code is broad, embracing everything from legacy phone companies to the first digital pioneers, so teasing out the true “ISP” cohort requires extra care.
The first step is to extract a registry of firms coded under ISIC 6420 for the country or region of interest in 1991. The list is inevitably eclectic. A handful of incumbents—national telecom monopolies, regional carriers—appear alongside a growing set of new entrants. The trick is to identify those explicitly marketing “internet access” as a commercial service. Trade journals, early advertisements, and the occasional business plan or licensing application from the period can help. Many of the first ISPs were small, often spun out from academic or research network providers, and some operated in a legal gray area before formal licensing caught up to reality.
Once the first commercial ISPs are identified, the next hurdle is measuring uptake. Subscriber counts from 1991 are, frankly, hard to pin down. Public reporting was sporadic. Where numbers exist, they often appear in trade press interviews, investor pitch decks, or, for larger firms, in the earliest regulatory filings. Sometimes, estimates must be triangulated from technical sources: the number of dial-up ports, modem banks, or even customer support staff. Even the largest ISPs in 1991 counted users in the hundreds or low thousands—a reminder of how small the market still was.
Subscriber growth is only half the picture. The true infrastructure bottleneck was backbone capacity—the bandwidth connecting local ISPs to national or international networks. Company technical reports, government filings, and the earliest peering agreements sometimes provide data on trunk line capacity: how much data could, in theory, move in and out of each network. The figures are usually given in kilobits or megabits per second, and, compared to modern standards, are almost comically small. Yet, at the time, a small increase in backbone capacity often enabled a meaningful jump in subscriber numbers.
Overlaying subscriber counts with backbone data provides a sense of both demand and supply. In regions where backbone upgrades came first, new ISPs could scale quickly. Where capacity lagged, even the most aggressive marketers faced technical ceilings. Some firms reported waitlists, “busy signal” ratios, or even public pleas for patience as infrastructure caught up. These signals, though not always quantified, mark the points where commercial enthusiasm ran ahead of the physical network.
Economic context and policy also mattered. Some countries or cities saw faster ISP growth due to proactive deregulation, municipal investments, or partnerships with research institutions. In other places, monopolies resisted new entrants, or regulations lagged, slowing commercial expansion. Matching ISP entry and subscriber growth to these policy shifts adds another layer of explanation to the data.
Throughout, careful documentation is essential. Every assumption—about firm classification, data gaps, or the linkage of subscribers to backbone capacity—should be noted. The history is imperfect, the statistics scattered, but by layering ISIC 6420 firm data with subscriber and backbone figures, a map of early commercial internet services does come into view.
The numbers are modest, the boundaries fuzzy, but the story is there: a handful of firms, a few thousand subscribers, and a physical network just beginning to respond to a surge of commercial curiosity. For those willing to piece together the evidence, 1991 marks the quiet start of something much larger than anyone at the time could have anticipated.