Summary

  • A March 1997 report described a dispute over CNI’s proposed use of ONATEL’s earth station; in August, Reuters reported that CNI had already run a free trial and was serving customers in Bujumbura.
  • CNI’s director put the customer start on 1 August and the public launch later in August, separating technical trial, customer availability and public presentation.
  • A later IANA report says Burundi depended on infrastructure in Zaire for Internet connectivity in 1996, but does not identify the exact route used for CNI’s 1997 service.

“Burundi went online” sounds like a date. The surviving reports instead show a sequence. A national computing agency described a conflict over access to an earth station; its director later said that a free trial had been running for months; paying customers in the capital could connect before a public launch was held. The distinction matters because an announcement can mark public recognition of a service without marking the day its equipment first carried traffic, or the day its first users could buy access.

The institutions in the story were CNI, the Centre National de l’Informatique, and ONATEL, the Office National des Télécommunications. In a 12 March 1997 report, Bujumbura newspaper Net Press said that both organisations wanted to operate the Internet project. It reported that CNI needed ONATEL’s earth station, that ONATEL had refused use of it, and that CNI said it was ready to rent another antenna if necessary. The article attributed ONATEL’s reported refusal to its intention to invest in the same sector. That is a contemporary account of a dispute, not an independently verified finding about the company’s motive.

Net Press also reported that the telecommunications minister had authorised CNI to operate the project. The state-enterprise oversight body SCEP had asked ONATEL to permit station use under a rental arrangement, which was to end once ONATEL’s own infrastructure could serve the project. Net Press said the project was delayed while ONATEL had not responded. Those details identify a control point: having an Internet service plan did not by itself give CNI access to the international facility it expected to use.

But this report does not document a signed lease, the antenna’s configuration, or the route that ultimately carried CNI’s August traffic.

The later chronology comes from a Reuters dispatch dated 1 August 1997, republished by other outlets. CNI director Barnabe Karorero told Reuters that the organisation had offered service free since May on a trial basis, then began offering full Internet access to customers in Bujumbura on Friday, 1 August. He said roughly 80 people were connected. The report described the public launch as planned for later in August. It did not say that the planned event happened on a particular date, nor that service was available throughout the country.

These are three different milestones in a short span. A May trial implies that some service was already operating, although the article does not describe its technical configuration or user count at that time. The 1 August statement concerns customer availability in the capital. The later public launch concerns a public-facing event. Calling all three “the launch” erases the difference between a test, a service offer and a ceremony. Calling 1 August the day Burundi first had any Internet connection also goes beyond what Reuters established.

A later, more technical clue comes from IANA’s 2002 report on the .bi domain. It says that .bi was first delegated in October 1996 and that its initial technical contact was associated with IZPT, an Internet provider in Kinshasa, then in Zaire. IANA states that Burundi depended on infrastructure in Zaire for connectivity at the time. This helps explain that early national service crossed a border in its technical and support arrangements. It does not show that CNI’s 1997 customer link used the same path, or that the .bi domain’s delegation itself provided Internet access.

The August Reuters account gives a glimpse of the service’s narrow early market. Karorero described a mix of foreign aid workers, local businesspeople and government officials among subscribers. CNI reportedly charged $50 for ten hours a month, with additional hours billed separately, or $150 for unlimited monthly service. Those were substantial commitments for a small customer base. Reuters also reported that CNI kept two generators available because of frequent power cuts.

A functioning Internet offer therefore relied on more than an international link: an upstream facility, a local service operator, a way to pay, electrical continuity, and enough support to keep customers connected all had to align.

That evidence does not measure national access, household adoption, traffic quality or the number of institutions using the service. The reported 80 connections are an interview figure, not a census. CNI’s intention to work with the University of Burundi is not evidence that a university network was already serving students. The March dispute and August service report do not close the intervening operational chain. It remains possible that the arrangement changed, that another facility was used, or that the press account omitted technical detail.

The more defensible history is therefore not one ceremonial birthday. It is a ladder of claims with different evidence: a March conflict over a proposed earth-station arrangement; a reported free trial beginning in May; a customer offer in Bujumbura by 1 August; and a public launch that was still planned when Reuters spoke to CNI’s director. The editorial lens of Heng Lu’s “Running-Code Primacy” encourages attention to what a working service can actually do, rather than allowing plans or institutional announcements to stand in for operation.

The historical evidence sets the limit: customers were reportedly served before the planned public launch, while the precise route and nationwide reach remain unproven.

Sources