[ih] History of Tier 1 Networks

[Vint Cerf]

thanks for many memories, Olivier.

I joined MCI in 1994 and was part of the planning for merger of the two
companies. We built two networks: MCI.NET (commercial internet) and vBNS
(very Broadband Network Service) for NSF in 1995 as the NSFNET was shut
down. We were planning to merge with BT but around Feb 1998, as I recall,
the BT Board dropped its offer by about 20% and the MCI Board accepted the
Worldcom bid instead. That was a disaster as several people have noticed.
Upon that merger, MCI.NET was divested to Cable and Wireless because
anti-competition rules led to the requirement that MCI Worldcom divest
either MCI.NET or UUNET. By 2002, MCI was bankrupt thanks in part the tax
fraud involving mistreatment of optical fiber lease costs (among other
things, perhaps).

v


On Mon, Apr 27, 2026 at 10:33 PM Olivier MJ Crépin-Leblond via
Internet-history <internet-history at elists.isoc.org> wrote:

> Dear James,
>
> I never imagined this question being asked in the "Internet history",
> but here's my $0.02 based on some personal experiences at the time.
> Disclaimer: this is not the "definitive" history of the development of
> these companies, but based on some of my memories, some of which might
> be biased, and some faded, and I apologise for such potential inaccuracies.
>
> My comments are interspersed in your text below.
>
> On 24/04/2026 08:47, James Bensley via Internet-history wrote:
> > [ ... ]
> >
> > The main question I want to understand is how the current set of Tier 1
> networks came to be Tier 1 networks (why these networks and not a different
> set of networks?).
> >
> > To clarify my query; below is the list of Tier 1 networks which I think
> it could be said are derived from state funded telcos within a certain
> country:
> >
> > - AS701 Verizon / UUNET (from the USA via Bell Atlantic)
>
> A brief history of UUNET is fairly summarised on the Wikipedia page:
> https://en.wikipedia.org/wiki/UUNET
>
> What helped UUNET at its inception was that it took over most of the
> dial up UUCP traffic from !seismo! "Seismo" was a machine at the Centre
> for Seismic Studies in Northern Virginia and was operated by a gentleman
> named Rick Adams. By virtue of its location in the UUCP bang paths,
> seismo was one of the most important nodes in the UUCP world in the late
> 80s.
> As usual, at some point the "accountants" questioned why seismo was paid
> for by university funds to perform communications that had nothing to do
> with seismic studies. Rick Adams therefore started UUNET and transferred
> the traffic to that.
>
> As usual, the initial "idea" was, let's do it as not-for-profit, but
> soon enough, it was clear that it was receiving a lot of longer distance
> dial up calls and leased lines were needed for some of the main UUNET
> backbone traffic (some using the NSFNET, some using other already
> existing networks that were typically US gov funded) - with email
> playing a small part but Usenet news making up a lot of traffic, with an
> explosion of newsgroups including some alt.binaries that were great
> bandwidth hoggers.
>
> Soon enough UUNET launched a commercial venture, AlterNet and this
> changed the game. When faced with threats of disconnection from the
> NSFNET because of newly created Acceptable Use Policies, UUNET got with
> PSINET and CERFnet, both fearing disconnection, to create the first
> Commercial Internet Exchange - CIX. More info on this on:
> https://en.wikipedia.org/wiki/Commercial_Internet_eXchange
>
> CIX was in California (I had assumed Washington DC but the Wiki page
> says otherwise?) but UUNET was on the East Coast.  Almost
> simultaneously, a sister to the Palo Alto Internet eXchange (PAIX) was
> created in Reston - Metropolitan Area Ethernet East (MAE-East) born in
> 1992. (or my memory fades on this, perhaps MAE-East was there before
> PAIX) More on https://en.wikipedia.org/wiki/MAE-East -- if someone
> remembers, it would be appreciated.
>
> By then, most of the dial up UUCP traffic to UUNET had switched to
> leased lines which made MAE East big pretty quickly as a very reliable
> interconnection point between commercial and non commercial networks -
> the world's first large scale Internet Exchange Point (IXP). Each of
> these developments were not without the Cassandras predicting the
> "immediate death of the Internet". Thanks to dozens of really determined
> individuals, the whole thing held together and thrived, especially in
> the face of the WWW suddenly multiplying exploding traffic - and
> pornographers finding the Web to be an excellent way to make.money.fast.
> (pun intended)
>
> As widely known, UUNET was Verizon's entry point into the Internet,
> somehow. The pioneer spirit was really that of Rick and his friends.
>
> > - AS1299 Telia / Arelion (from Sweden)
>
> No idea --- but I sense there's a link to the first Arpanet link to
> Sweden with SUNET and NorduNET academic networks picking up from there
> and creating a large online market for Internet in Sweden very early on.
> A fairly good history is explained on the relevant Wikipedia page:
> https://en.wikipedia.org/wiki/History_of_the_Internet_in_Sweden
>
> I can only imagine that Telia benefited from this early strong local
> market.
>
> > - AS2914 NTT (from Japan)
>
> NTT's entry in the Internet was already alluded to in that they
> purchased Verio Internet. Verio Internet was primarily a hosting
> provider starting in 1996. I knew them well as my company was one of
> their first customers. Just a small account with them, but they were
> incredibly reliable and their console for managing domain names,
> nameservers, web servers etc. was excellent.
> In parallel, NTT was funding telecommunications research at Imperial
> College London with a Japanese research assistant being in the same lab
> as me and through him, I met with the then CEO of NTT Europe circa
> 1994-5. At the time a group of us young students had a business plan for
> an ISP that would rival the UK's only large commercial ISP PIPEX. (
> https://en.wikipedia.org/wiki/Pipex )
>
> We presented a full business plan to NTT Europe that included the
> creating of one of Europe's first commercial Internet overlay.
>
> But there was a snag. NTT Europe was a small operation mostly providing
> leased lines and telephone for financial services clients in Europe. NTT
> Japan was a national Telecom Company (TelCo) but focussed solely on
> National connectivity. The international connectivity market in Japan
> was run by KDD Corporation, a company created in the 1950s for Japan's
> international telecommunications. Thus, in a deeply conservative country
> with very strong social and business protocols, it was inconceivable
> that NTT could offer a telecommunication service that would provide
> publicly accessible international telecommunications. We had several
> meetings with a range of top level NTT officials, each with protocol,
> interpreters, etc. only to end with a refusal from NTT Europe to grant
> our small group £2M and invest no more than £5M into an Internet
> network. I remember the speech of the top man: "Internet is a hobbyist
> network. The risk is too high" -- and the risk-averse Japanese company
> ended up making a complete u-turn five years later after huge hubbub and
> restructuring in the late 90s as to why NTT is not a player at all in
> the Internet. They had to purchase Verio for $5.5 billion in August 2000.
> Yes, 5 years to go from £5M to £5Bn - that's how much corporate mistakes
> cost.
> Thus no entrepreneurial score for NTT, just very very deep pockets to
> the rescue.
>
> > - AS3320 Deutsche Telekom (from Germany)
> > - AS5511 France Telecom / Orange (from France)
>
> These two are inter-related. Soon after our failure with NTT, in 1996 I
> read that France Telecom, Deutsche Telekom and Sprint Corporation had
> signed a joint venture named "Global One". Sprint had risen in the US to
> be a major player. They had plenty of clients on X.25 networks that they
> transferred to TCP-IP and had plenty of international telephone lines.
> So Sprint brought the Internet knowledge into the Global One Alliance. I
> immediately got a meeting with the top guys at France Telecom in Paris a
> stone's throw away from the Gare Montparnasse to push for an evolved
> version of the project. Sitting at the table were the top people from
> France Telecom, VTCOM their multimedia division, Transpac their data
> network. And things were not looking good: the VTCOM guys were defending
> the Minitel which was French and which people needed to pay for use.
> Transpac defended their track record in running a very stable large
> scale X.25 network and had some aversion to TCP-IP. Here again, the top
> guy's quote to remember was: "the Internet is a pipe dream for "les
> américains" - how will you ever make any money if the services you offer
> are free? You cannot make a penny giving services out for free. Oh and
> it's all in English, which will never work in France" - minitel was
> raising a lot of money through videotex services whereas you'd need to
> pay every time you dialled in, except for the free "annuaire" -
> directory enquiries. Still - they quietly launched "wanadoo" as a
> side-track, with a young team that they were predicting would suffer...
> Turning to the topic of the Global One Alliance the folks scoffed at
> Sprint Communications (they are tiny) and at Deutsche Telekom (we'll
> show the Germans who's the boss).
>
> With no outright recognised leader, the Global One Alliance failed
> pretty quickly. It is only through a generation change, with all of the
> people present in the room retiring 5-10 years later, that France
> Telecom finally understood the stakes and I suspect the same went of
> Deutsche Telekom - but they had to pay the high price of acquisition and
> development at full pace instead of being there from the early days and
> thus stupidly losing an early mover advantage. (Wanadoo did grow
> quickly, but it was much to their surprise.)
>
>
> > - AS6453 Telstra (from Australia)
>
> This one's an interesting one which I do not have first hand
> information, but the story I have heard on many an occasion (and that I
> never doubted) is that it all came to one determined telecom engineer
> whom you all know, who fought every bit of the way for Australia to have
> a satellite link to the Internet for the Australian Academic and
> Research Network (AARNet) after munnari.oz UUCP node was set up.
> https://en.wikipedia.org/wiki/Internet_in_Australia
>
> Again - the big Telcos at the time had no clue, but thankfully Telstra
> was born out of entrepreneurial pioneering.
>
> > - AS6762 Telecom Italia Sparkle (from Italy)
> > - AS6830 Liberty Global (comprised from the M&A of many European
> incumbents)
> > - AS7018 AT&T (from the USA)
> > - AS12956 Telxius / o2 (from Spain)
>
> I have no story for any of the above.
>
> >
> > I think the following Tier 1s have their roots as being either fully
> privately funded or only minorly state funded (sometimes indirectly through
> M&As):
> >
> > - AS174 Cogent
> > - AS3257 GTT
> > - AS3491 PCCW / Console Connect
> > - AS3356 Lumen / Colt
>
> This is another personal story. In 1995, Colt (City of London Telecom)
> was a very small, new operator of telecommunication services, mostly
> telephone lines, in the City of London. In 1993 in the early days of UK
> telecommunication deregulation, it was awarded a license to compete with
> British Telecom and Cable and Wireless in voice and data communication.
> I met with the then CEO and leadership team of this small firm a few
> weeks after they had just signed up their 100 client in the City of
> London. They only dealt with corporate clients. At the time, the
> majority of their investment went into digging the roads around the City
> of London, laying cables and fibre to services new customers, one block
> at a time. We presented a plan for Internet Services for the Financial
> Industry in the City - which was firmly rejected two weeks later. The
> top guy's quote to remember was: "We asked our 100 clients about their
> interest in the Internet. Not a single client is interested. Not a
> single one! In fact some of the clients are telling us the Internet
> would be a security risk to their operations and they definitely do NOT
> want to get connected." - I could not convince him or any of his
> colleagues otherwise.
> Again, it took a generation change and new management with much capital
> injection to make COLT what it is today. They needed a huge sum for a
> bailout agreement in 2001 and only then did they grow with different
> management. Thus, no pioneering spirit in them in 1995.
>
>
> > I am told that some expanded into global connectivity due to a desire to
> provide better connectivity for their home market. What puzzles me is why
> some of these state funded telcos did this and some didn't.
> >
> > I'm told that NTT is one such example, which is now one of the biggest
> global IP carriers (Nippon Telegraph and Telephone wanted to improve
> connectivity and prices for the Japanese market), whereas SingTel (from
> Singapore) didn't; their monopoly in Singapore is still very strong today,
> but they've mainly staid within their “region”.
>
> See above. I think that NTT thrived the moment it was accepted that they
> could provide international connectivity back home. If I remember
> correctly, NTT were, at the time (late 1990s) the largest company in the
> world by its real estate value so they had very deep pockets.
>
> >   BT (from the UK) is another example. BT didn't go on to become a Tier
> 1 whereas most of it's Europe neighbours did.
> >
> > I'm guessing that the peering arrangements that lead to establishment of
> the Tier 1 tier where solidifying around the early 90's. At this time BT
> was operating Tymnet which later morphed in BT Global Services. I think
> they were very welled placed to enter into this realm, but didn't for some
> reason.
>
> The story of BT is a very sad turn of events, and ultimately a huge
> waste of amazing people.
> The UK was a pioneer of Internet Connectivity, thanks to many people but
> of course with the first link with Peter Kirstein at UCL playing a big
> role. The folks at the University of Kent at Canterbury channelled the
> UK's international UUCP traffic; JANET the UK Joint Academic Network,
> ran on X.25 and span the whole of the UK. There were multiple numbers of
> other networks, with the National Physics Lab (NPL) being a Pioneer.
> This is all explained in
> https://en.wikipedia.org/wiki/Internet_in_the_United_Kingdom
>
> It all worked really well, even though there was a real struggle in the
> early 90s in regards to replacing X.25 with TCP-IP but that's another
> story.
>
> Anyway, BT provided all of the leased lines at the time, across the
> country, plus a good microwave network. It engaged in cutting edge
> research in long distance fibre optic. Its labs in Martlesham Heath,
> which I visited in the early nineties as part of my PhD research, had
> incredible projects like optical routers and demonstrations of
> e-commerce and even an AI digital assistant that could make flight and
> hotel bookings. Wow!!!
>
> But because of the telecom deregulation in the UK, it was blocked by the
> UK government from commercially developing any of these services. BT had
> to be "weakened" to let the competition gain ground for a "fair
> competition" - otherwise, in my opinion, BT would have probably crushed
> all competition out there.
>
> It is only in around 1995 that it started offering individual Internet
> access, with a very clumsy initial marketing campaign that was just a
> set of internal documents as to how you could use the Internet. Yet, it
> quickly became one of the UK's largest ISPs, quickly passing Demon
> Internet, the independent ISP that had stated in the early 90s. By the
> late nineties, BT had fully understood the Internet stakes. I must admit
> that my company was a sub-contractor for BT Syncordia and we did lots of
> work on TCP-IP private networks for banks, government, transport, which
> required very high quality that BT could supply but that most of the
> other UK operators at the time were unable to reach.
>
> Soon enough in 1997-8, BT was interested in purchasing MCI through a
> multi million dollar acquisition after it had a good relationship with
> MCI through the Concert Communications Strategic Alliance it had forged
> with MCI in 1994. This is another story that probably requires its own
> chapter but some others here are probably much better suited to tell it
> than me. I was on the BT side, when WorldCom bid $37 billion for MCI, a
> tragedy for us. The sum was deemed to be unreasonable thus BT dropped
> its bid. I used the word "tragedy" because as it happened, MCI WorldCom
> ended up having a massive $11 billion accounting fraud scandal (another
> story!!! https://en.wikipedia.org/wiki/WorldCom_scandal ) and bankruptcy
> in 2002. It ended up being acquired by Verizon in 2005.
>
> BT never really recovered from this slap in the face. Costs in preparing
> the buyout of MCI were signed off as a huge loss. The "accountants" took
> control. By then competition, helped by the UK Government deciding to
> allocate large corporate contracts to new entrants in the UK at the
> expense of BT (some new entrants not playing fair and ending up in their
> own huge corporate scandals - more side stories here). The last
> surviving branch of BT Syncordia, its UK operations, were shut down in a
> massive restructuring by the year 2000 with the loss of tens and tens of
> thousands of jobs overnight. I remember receiving a call from one of the
> top honchos asking me if I could hire him - and I had to admit to him
> that throughout these years, although we were doing so much work for
> them, we were a very very small company that could not afford him.
> For a more complete history of the BT tragedy, read:
>
> https://www.encyclopedia.com/books/politics-and-business-magazines/bt-group-plc
>
> Ultimately, I blame the "accountants" and cost-cutters for screwing BT
> up. The best of the best people were all headed to the chopping board
> under a mountain of paperwork.
>
> So that's how BT did not end up as a Tier 1 Internet Provider. They did
> have the pioneering spirit, but perhaps were over-ambitious and also
> unlucky.
>
> I hope this provided a bit of entertaining reading. Some of it might be
> faded as it's all from memory.
>
> I deliberately did not include names of some people because their
> failure to recognise the Internet's potential would be deeply
> embarrassing for them. <sigh>
>
> Kindest regards,
>
> Olivier
>
> --
> Internet-history mailing list
> Internet-history at elists.isoc.org
> https://elists.isoc.org/mailman/listinfo/internet-history
> -
> Unsubscribe:
> https://app.smartsheet.com/b/form/9b6ef0621638436ab0a9b23cb0668b0b?The%20list%20to%20be%20unsubscribed%20from=Internet-history
>


-- 
Please send any postal/overnight deliveries to:
Vint Cerf
Google, LLC
1900 Reston Metro Plaza, 16th Floor
Reston, VA 20190
+1 (571) 213 1346


until further notice