The History of Broadband

By Victor Hoang · Updated Oct 24, 2026

In February 1996 the White House held a signing ceremony in the Main Reading Room at the Library of Congress. It said the new Telecommunications Act "will help to develop the nation's information superhighway". The problem was plain. Most people online used a modem on a voice telephone line.

Broadband has no single birthday. Its two physical ancestors are the telephone line of the 1870s and the community antenna television systems of the late 1940s. Its engineering arrived in the 1990s as DSL and cable modems. Its legal frame is the Telecommunications Act signed on 8 February 1996. Its definition keeps moving: the FCC's fixed benchmark went from 4/1 Mbps in 2010 to 25/3 in 2015 to 100/20 in 2024. The throughline: phone and cable companies taught their existing wires new tricks. Then fiber and public money arrived to reach people those wires never served well.

A 1998 modem standard for that line, ITU-T V.90, topped out at 56,000 bit/s downstream and 33,600 bit/s upstream. Broadband is what happened when engineers, regulators and money pushed past that ceiling.

A quick definition first. "Broadband" means a high-capacity connection that is always on, as opposed to dialing a phone number each time. The legal meaning is narrower and keeps changing, because Congress left it to the FCC to decide what counts as "advanced" at any given moment. For the network these connections plug into, start with the pillar, The History of the Internet.

Timeline at a Glance

The Two Wires (1876-1995)

Phone copper and TV coax get taught data

The turning point is a lineman's side business. By the account in the Harvard Business School library's history of cable, John Walson was a lineman for Pennsylvania Power & Light who also owned an appliance store in Mahanoy City, Pennsylvania. To demonstrate television sets that nobody could sell, he strung wire from a hilltop antenna so his customers could pull in Philadelphia stations, 86 miles away. People asked to be connected, and he charged $2 a month. By mid-1948 he had 727 customers, and by 1955 there were about 400 such "Community Antenna Television" systems with 150,000 subscribers in total. Nobody was thinking about the internet. The point is that a second wire network, made of coaxial cable, was now running past homes next to the telephone's copper pairs.

The telephone wire was older and more widespread. Bell's 1876 patent started a network built to carry voices across a narrow band of frequencies. Computers learned to use it through modems, and the history of ARPANET shows how early machines already talked over dial-up lines. The mismatch was baked in. A voice network sets up a circuit for each call, while the internet that grew out of packet switching shares lines among many senders. Squeezing one onto the other worked, slowly.

A third wire was made of glass. In 1966, Charles Kao calculated that a fiber of purest glass could carry light signals over 100 kilometers, against only 20 meters for the fibers of the 1960s, and the first ultrapure fiber followed in 1970. The customer who wanted it was the telephone side: British Post Office representatives visited Corning in the mid-1960s asking for fiber with attenuation of about 20 dB/km. Corning's Robert Maurer, Peter Schultz and Donald Keck delivered 17 dB/km in 1970, when the best bulk optical glasses were roughly 1,000 dB/km. The physics was settled early.

The phone industry's digital line standard was ISDN. ITU's catalogue shows the basic user-network interface recommendation I.430 with an edition dated November 1988, and ITU's I.113 carries the title "Vocabulary of terms for broadband aspects of ISDN" in its June 1997 edition. So the word "broadband" was already in standards-body vocabulary long before consumers used it. ISDN itself never became the mass product; the next idea did.

That idea came from Bellcore. Joseph Lechleider of Bellcore is credited by the National Inventors Hall of Fame with inventing DSL and suggesting the asymmetric variant, and his name is on U.S. Patent No. 5,181,198. The Register's obituary explains the insight: signals crossing between upstream and downstream limited bandwidth, so he proposed that the carrier end transmit at higher power than the customer end, which let downstream speeds exceed upstream ones. Hardware followed. John Cioffi, a Stanford professor, founded Amati Communications in 1991, where his team built a modem that carried more than 6 Mbps over 9,000 feet of telephone line. In 1993 ANSI chose DMT, the modulation Cioffi and his students developed, as the US standard for DSL. The Register quotes Cioffi calling the result "a simple, elegant solution to the problem".

Cable had its own homework. The plant carried television one way, which is no use for data. Time Warner Cable engineers developed hybrid fiber coax in 1993, per CableFax, creating a clean return path that turned the plant into a two-way network. The same article places a test of what became Road Runner in a rebuilt cable system in Elmira, New York, in 1995.

For teams today: the cheapest path to a new capability is often an old asset with one missing feature. DSL did not need new wire, only a smarter idea about power and frequency. Ask what you already own that is one clever fix from a new use.

After the 1996 Act (1996-2000)

Congress asks for advanced services and engineers answer

The turning point is a law. The Telecommunications Act, Public Law 104-104 (S. 652), is described in its own heading as an act to promote competition and reduce regulation in order to secure lower prices and higher quality for consumers and encourage rapid deployment of new telecommunications technologies. Section 706 is titled "Advanced telecommunications incentives". The FCC's 2024 report explains the lasting effect: the Commission must keep inquiring whether advanced telecommunications capability is being deployed to all Americans in a reasonable and timely fashion. Every later benchmark fight starts from that sentence.

Picture a household in 1998 with one phone line. Going online meant tying it up, and the best case was the V.90 standard approved by ITU on 1 September 1998: up to 56,000 bit/s down and 33,600 bit/s up. This is the world the first web browsers were built for. Pew's trend series starts in March 2000 with 1% of US adults reporting home broadband.

Both rival wires answered at once, and both needed a standard to become a product. On the phone side, ITU published the first edition of G.992.1, ADSL transceivers, in July 1999. On the cable side, a DOCSIS timeline from Multichannel News (via NextTV) records that CableLabs adopted DOCSIS as a working project name in December 1996, the industry formalized its modem certification plan in November 1997, and Thomson and Toshiba received the first DOCSIS 1.0 modem certifications in March 1999, with Cisco first to qualify a cable modem termination system. Certification mattered for the same reason it did in Wi-Fi: a buyer could pick any certified modem and expect it to work.

These years were also the dot-com years, and the physical network was overbuilt for a demand that arrived later than the promoters said. The glut of idle backbone fiber is part of the story in the dot-com bubble. The access network, the last mile to the home, was the part that still had to be solved.

For teams today: a standard on paper is not a product. DSL and DOCSIS both needed a second step, an interoperability process, before customers could buy with confidence.

The Duopoly Years (2001-2009)

Phone and cable compete, and the rules favor both

The turning point is a court decision. On 27 June 2005 the Supreme Court ruled 6-3 in National Cable & Telecommunications Association v. Brand X that the FCC's classification of cable modem service as an information service, not a telecommunications service, was a lawful construction of the Communications Act. That meant cable modem providers were exempt from mandatory common-carrier regulation. The owners of the two existing wire networks, cable and telephone companies, were now competing under that frame.

Adoption grew fast inside that frame. Pew's fact sheet says home broadband increased rapidly between 2000 and 2010, from 1% of US adults in March 2000. The equipment kept improving too: NextTV's timeline records that the first DOCSIS 3.0 modems were certified in May 2008, and the first 3.0 headend equipment in December 2007.

Meanwhile a third competitor laid glass to the door. Verizon launched FiOS TV in Keller, Texas, on 22 September 2005, and by its first anniversary the service had reached more than 80 communities across seven states. Fiber to the home had reached paying customers, thirty-five years after the 1970 glass. The web that ran on these connections also changed shape. The history of the World Wide Web explains how pages grew from documents into applications, and heavier pages needed the faster pipe. So did video, which is where the creator economy found its first paying platform.

For teams today: a classification decision can shape a market for ten years. When your industry's rulebook is being written, the technical question of what a product is can matter more than the product.

The Plan Years (2010-2019)

A national goal, and a bar that keeps moving

The turning point is a document. On 16 March 2010 the FCC released Connecting America: The National Broadband Plan. According to a Congressional Research Service report, its first goal called for 100 million US homes to have affordable access to actual download speeds of at least 100 megabits per second, with upload speeds of at least 50 Mbps, by 2020. The goal is often called "100 squared".

The plan sat beside a quieter change: the legal definition kept climbing. The FCC's 2024 report summarizes the history in one line each, and the three benchmarks are easy to compare.

Home adoption held in the low 60s. Pew's fact sheet lists several 2010 readings of 60% to 64% of US adults with a home broadband subscription. And the 25/3 line set in 2015 held for most of a decade: the FCC later said it was the benchmark "relied on in the last seven reports", even as the speeds many providers marketed substantially exceeded it.

The era also showed the limits of a speed target. A benchmark can say what a good connection is. It cannot say who gets one, or at what price. The wireless last meter inside the home, covered in the history of Wi-Fi, improved faster than the street-level wire that fed it, and the Ethernet cable behind the router stayed the quiet standard for inside-the-building wiring.

For teams today: a target that cannot be measured by its owner will drift. The plan's goal was written in actual speeds, while the 2012 availability figures in the same CRS report counted advertised speeds.

The Funding Years (2020-present)

Gigabit cable, public grants and a higher bar

The turning point is the vote that set the new bar. On 14 March 2024 the FCC adopted its 2024 Section 706 Report, concluding that fixed broadband needs download speeds of at least 100 Mbps and upload speeds of at least 20 Mbps to count as advanced. It said it could no longer conclude that 25/3 supports "advanced" functions, and it noted that it was weighing Congress's action and other federal and state programs. Two commissioners dissented.

The same report counted who was left out. As of 2022, the FCC says approximately 24 million Americans lacked access to fixed broadband at 100/20 Mbps, with rural areas accounting for nearly 80% of that total, or almost 19 million people. The number is a reminder that the wires were built where the money was, not where the people were.

The public response is money. NTIA's Broadband Equity, Access and Deployment program is a $42.45 billion federal grant program funded by the Infrastructure Investment and Jobs Act.

The technology kept advancing on the old wires too. CableLabs released the DOCSIS 4.0 specifications on 26 March 2020, putting the industry on a path toward speeds of 10 Gbit/s downstream and up to 6 Gbit/s upstream. The coax that Walson's customers watched television on in 1948 is, in principle, still carrying the internet. For how this plumbing connects to the rest of the story, see the Internet timeline.

For teams today: infrastructure progress is not uniform. If your service assumes a fast connection, your market is whoever has one, and that is a smaller group than the headline averages suggest.

Key People

Key Facts

• The Telecommunications Act became law on 8 February 1996 as Public Law 104-104, with Section 706 on advanced telecommunications incentives.
• ITU approved the 56 kbit/s V.90 modem standard on 1 September 1998.
• Pew's series starts at 1% of US adults with home broadband in March 2000.
• The FCC raised its fixed benchmark from 4/1 Mbps in 2010 to 25/3 Mbps in 2015, then to 100/20 in 2024.
• About 24 million Americans lacked fixed broadband at 100/20 Mbps in 2022, per the FCC.
• BEAD is a $42.45 billion federal grant program.

The Pattern Behind the History

We call the sequence the Old Wire, New Rulebook Pattern. It is our synthesis of the history above, not a research finding.

1. Reuse the wire already in the ground. DSL rode telephone copper and DOCSIS rode television coax, both after a clever fix.
2. Standardize, then certify. G.992.1 and DOCSIS turned lab results into products that different makers could ship.
3. Let the law choose the competitors. The 1996 Act and the 2005 Brand X decision shaped who could build and who had to share.
4. Move the finish line. The FCC's benchmark rose from 4/1 to 25/3 to 100/20 as use changed.
5. Fund the gap. When markets left millions without service, public money arrived through programs such as BEAD.

What It Teaches Teams Today

• Look for the old asset. The biggest broadband leaps came from existing wires with one missing feature, not from new networks.
• Certify what you standardize. DOCSIS needed certification waves, not just a specification, before a modem was a safe purchase.
• Know who writes your rules. The classification of a service in 2005 shaped a decade of competition.
• Define "good" in writing, and revise it. A moving benchmark is honest. A fixed one eventually protects the status quo.
• Plan for the people at the edge. If your product assumes a fast, always-on connection, check how many of your customers actually have one.

Broadband shows how a network gets built: old wires get new tricks, rulebooks decide who may compete, and the finish line keeps moving. For the question of who built the whole network, read who invented the Internet.