Summary

  • In 1997, investors proposed a company that would sell Internet exchange access for commercial traffic, building on a WIDE project whose NSPIXP exchanges KDD described as experimental and academic.
  • JPIX’s plan attached reliability to a funded service and a two-site design. Its announcement records that intention; it does not prove the promised resilience or reveal who governed access.

The operating model was the story

On 23 June 1997, KDD announced that it and seven other companies would establish Japan Internet Exchange Co., or JPIX. Sixteen investors were expected to form the company in July. The release did not present another research network. It described a company that would provide exchange service to Internet service providers, carry commercial traffic and respond to the growth in Internet use.

The comparison in the announcement was pointed. Japan’s two major exchanges then in full operation, NSPIXP1 and NSPIXP2, were described as basically academic facilities run by the WIDE project for experimental purposes. JPIX would operate alongside that work, not erase it: KDD said the new company would cooperate with WIDE. What the announcement proposed to change was the institutional basis for dependable commercial service.

That difference became clearer a dozen years later. In a 2009 JPNIC interview, JPIX president Yoshiki Ishida recalled that NSPIXP2 depended in part on voluntary effort. More users, he said, were willing to pay for exchange connectivity as a service, and ISPs therefore invested in a fully commercial version of NSPIXP2. His account is a participant’s later recollection, not minutes from the 1997 negotiations. It nevertheless explains why a company mattered: someone had to carry the recurring obligation to maintain and scale an exchange as customer traffic grew.

The June release made that obligation concrete through a physical plan. The first JPIX switch was to be installed at KDD’s Otemachi building. A later expansion to KDD’s Shinjuku building was intended to distribute risk and load across the exchange functions. KDD also said the company would serve ISPs and content providers and seek users in Asia and the United States. These were plans, not performance measurements. The public records reviewed here provide no 1997 uptime series, customer contracts, fee schedule or traffic graph with which to judge whether the design delivered its goals.

Ownership was distributed, though not evenly. KDD’s release listed 16 shareholders and capital of 250 million, without naming the currency in its English text. KDD held 24 per cent, the largest listed stake; five other firms held 10 per cent apiece, Nihon Cisco Systems held 8 per cent, and the remaining nine held 2 per cent each. KDD was an anchor investor, not a majority owner. That table tells us who supplied capital in the plan. It does not tell us the board’s voting rules, customer admission criteria or which network controlled its own routing policy.

The dates also distinguish an announcement from an operating service. KDD’s June statement targeted an October launch. JPIX’s current history places the company’s establishment in July and the start of its IX port service in November. The later date is the better marker for service beginning; the earlier one remains evidence of what the founders expected when they announced it. By March 2000, the company says, it provided a Gigabit Ethernet interface—a later scale milestone, not evidence of the capacity available at launch.

The transition was therefore not a clean break between “volunteer” and “business.” JPIX’s own founding announcement coupled commercial traffic with cooperation with WIDE. The shared technical purpose continued while its funding and service obligations acquired a company wrapper. Nor did an exchange operator become the Internet’s routing authority: the reviewed records do not show JPIX deciding what each connected network announced or accepted. They show a narrower and more durable control surface—the facility, the port service, the investment required to expand it and the terms on which participants could connect.

That is why JPIX’s 1997 history is more than a corporate founding date. An experimental exchange can prove that networks are able to meet; a commercial service asks who will pay to keep that meeting point dependable as use grows. In Japan, the answer was a consortium of ISPs and infrastructure firms, with KDD as the largest named investor and WIDE still in the picture. The public evidence establishes the proposed bargain and the service’s start. It leaves the actual prices, reliability record and distribution of operating authority open.

Sources