ADSL History: How Copper Became the Backbone of Early Broadband

AT&T has utilized the same copper wire technology since 1887, spanning millions of miles that initially carried only analog voice calls. The advent of Asymmetric Digital Subscriber Line (ADSL) technology remarkably transformed these decades-old copper wires, enabling them to transmit millions of digital bits per second. In the 2000s, ADSL provided fast, always-on internet to millions globally using existing telephone lines, despite not being originally intended for broadband internet. While fiber, 5G, and Starlink dominate today, ADSL played a crucial role in shaping the modern internet.

Beginnings

The 1990s marked the rise of cable TV, which originated in 1948 as Community Antenna Television (CATV). CATV re-transmitted TV signals to households with poor reception. These signals were received via antennas and sent through coaxial cables. Initially, the FCC regulated CATV content tightly, but rules loosened in the mid-1970s, leading to premium channels like HBO. Events such as the "Thrilla in Manila" boxing match in 1975 boosted cable adoption. The 1984 Cable Communications Policy Act deregulated cable TV rates, allowing operators to fund massive build-outs, increasing cable systems in the US by 40% between 1985 and 1990.

Video Dial Tone

The rapid growth of cable in the late 1980s concerned telephone companies like the Baby Bells, which emerged after AT&T's breakup. Cable companies had a direct relationship with customers through coaxial cables delivering video, posing a threat that they might eventually offer other services like phone calls. The Baby Bells felt compelled to deliver video themselves, but antitrust laws prevented them from acquiring cable operators in their service areas.

In mid-1992, the FCC allowed Baby Bells to deliver video content through their wires, creating "Video Dial Tone." This service allowed users to stream movies by picking up a telephone. The telecoms could only transmit content on behalf of third parties, not own it, but this ruling allowed them to enter the streaming business.

They Got Cables. We Got Copper.

Cable companies had an advantage with their coaxial cables, which could carry rich video content. Telecoms, however, relied on legacy copper telephone lines, known as "twisted pairs." These insulated pairs of copper wires twist around each other to reduce interference and have been used by AT&T since the 1880s, resulting in a massive installed base.

A significant portion (99%) of the system's wires are "last mile" connections, typically 1-2 miles long, extending directly into households. These are the most challenging and expensive to replace, as it would involve installing optical fiber and entering homes, a process that can be decades overdue for some lines. Therefore, technicians needed to make Video Dial Tone work entirely over existing copper lines.

The Last Mile is Always the Hardest

Joe Lechleider, a Bell Labs researcher, studied signal processing. In the 1960s, AT&T began converting its network from analog to digital. By the 1970s, the network had digital transmission and switching, but the "last mile" copper lines still required digital-to-analog conversion. Telecom engineers envisioned an end-to-end digital system without replacing every last-mile wire. In 1980, Bell Labs began developing this, leading to the Integrated Services Digital Network (ISDN) international communication standards a few years later.

ISDN established two co-existing voice and data channels on copper lines, each with a bi-directional capacity of about 64 kilobits per second, plus a 16 kilobits per second overhead channel, enabling end-to-end digital signals. While the term "Digital Subscriber Line" (DSL) is debated, ISDN is considered a pioneer, gaining widespread adoption in Europe and Japan. In the US, AT&T did not widely adopt ISDN due to technical reasons and the chaos of its 1984 breakup, which also led Lechleider to Bellcore, a research consortium for the Baby Bells.

An Uneven Idea for Crosstalking

ISDN's slow speeds were a major reason for its limited adoption in the US, leading to the joke "Innovation Subscribers Don't Need." Increasing bandwidth was hindered by crosstalk, specifically "near-end crosstalk" (NEXT). NEXT occurs when signals from one twisted pair electromagnetically interfere with data in an adjacent receiving twisted pair within a cable bundle.

To mitigate NEXT, frequencies for transmitting and receiving data are split. Lechleider proposed an asymmetric split, recognizing that the carrier's receiving end is "louder" due to the convergence of many twisted pairs, while the household end is "quieter" with only one or two pairs. This insight led him to allocate a larger portion of frequencies for downstream data from the carrier to the receiver, and a smaller, lower-frequency portion for upstream data. This asymmetric setup significantly increased overall bits per second without as much crosstalk.

The Bellcore Olympics

Lechleider's asymmetric concept was initially deemed impractical until telecoms entered the movie streaming business with Video Dial Tone. Engineers like Charles Judice recognized that video streaming required more downstream bandwidth and championed asymmetry as a solution.

The next challenge was to agree on a modulation scheme, or "line code," to map digital bits to analog signals for transmission. In 1993, Bellcore held the "Bellcore ADSL Olympics" or "Bellcore Shootout" to determine the ADSL line code standard.

The three competitors were:

• CAP (carrierless amplitude/phase) from AT&T.
• QAM (Quadrature amplitude modulation) from Broadcom Corporation. Both CAP and QAM are similar, using established algorithms to encode digital data into an analog signal over a single frequency band.
• DMT (Discrete multitone) from Amati, a startup founded by Stanford professor John Cioffi, inspired by Lechleider. DMT, similar to Orthogonal Frequency-Division Multiplexing (OFDM), splits the 1.1 megahertz copper frequency band into 256 slices, each about 4 kilohertz wide. It then splits and encodes data using a method like QAM, transmitting it in parallel across these frequency slices.

DMT's multi-carrier approach won the competition, proving 50% more efficient and allowing significantly more data to be pushed through copper wire, achieving speeds up to 6 megabits per second. DMT's adaptability was another key factor. It could adjust the bit-per-hertz rate in different frequency slices to optimize for signal-to-noise ratios, especially in higher frequencies where copper attenuation is worse. This flexibility allowed DMT to probe and optimize for the unique characteristics of millions of copper loops, approaching the theoretical limit of data transmission.

DMT's drawbacks included its complexity, unfamiliarity among engineers, and computational intensity, requiring powerful silicon to manage 256 parallel channels. Amati's entry was a prototype, and the industry needed a company to produce DMT standards-compliant silicon and gear for telecoms. Alcatel stepped in.

A Small Team Inside Alcatel

In the early 1990s, Alcatel NV, the world's second-largest telecommunications firm, competed with giants like Ericsson and AT&T. A small team at Alcatel's research center in Antwerp, which focused on broadband access via coax cable and optical fiber, began a special project. While fiber offered the best performance, its high cost made it impractical for widespread use, leaving copper to fill the gap. At the time, copper's maximum data rate was limited to ISDN's 144 kilobits per second.

During a coffee break in 1990, Martin De Prycker and Willem Verbiest, two Alcatel researchers, discussed a symmetrical broadband DSL system prototype they had seen at Bellcore. This refrigerator-sized prototype demonstrated the possibility of 1.5 million bits per second over twisted pair lines, hinting at fast broadband internet over ordinary copper.

De Prycker and Verbiest secured permission to form a small, autonomous three-person team to work on DSL technologies. Despite lacking experience in twisted pair transmission, the team, after studying IEEE journals, began their work in January 1992.

The Pivot to DMT

Initially, the Alcatel team focused on single-carrier transmission line codes like QAM and CAP. However, in March 1993, news broke that DMT had won the Bellcore Olympics. This decision was controversial, and the debate over line codes continued for years, with some providers like AT&T sticking with CAP. Nortel, committed to optical fiber, also rejected DMT-based copper ADSL, developing a proprietary hybrid system called Etherloop.

The Alcatel team was surprised by DMT's victory. Verbiest recalled their certainty that CAP would become the standard, only to learn upon returning from a meeting that DMT had been chosen. Alcatel decided to pivot, licensing and adopting DMT despite their limited knowledge, recognizing an opportunity to influence the ADSL standard. The team rapidly expanded to over 100 people.

In 1995, less than two years after the DMT pivot and before the final ADSL standard was finalized, Alcatel introduced the first DMT-based ADSL chipset, consisting of an analog ASIC, a digital signal processor, and a channel processor, designed by Alcatel and fabricated on a 0.7 micrometer node. In October 1995, Alcatel demonstrated an end-to-end prototype at the Telecom Geneva Fair. By mid-1997, Alcatel had a video streaming product ready, but customers no longer needed it.

Video Dial Tone Fails

As late as 1996, industry experts predicted Video Dial Tone would be a success, envisioning on-demand movies streamed to homes. While the video rental market was valued at $12 billion then and Netflix is prevalent today, Video Dial Tone failed for several reasons.

The product itself was not terrible, offering hundreds of "channels" with menus for services like faxes and video mail. ADSL performed its function well. Bell Atlantic's CEO Ray Smith praised ADSL in a 1995 interview, highlighting its excellent quality, virtual VCR capabilities, and digital, server-based nature. Beta testers confirmed the product's quality, with channel-changing lag under a second and visual/audio quality comparable to broadcast satellite.

The installation economics also offered advantages over cable. Smith noted that while ADSL installation cost more per house, it was only done after a sale, unlike cable, where not every house in a covered area subscribed.

However, Video Dial Tone failed due to a lack of content. Telecoms, prohibited from owning content, struggled to entice movie studios to lease channels. The primary cause of its demise was regulation. Before 1996, the FCC's inability to establish coherent rules for Video Dial Tone led several carriers to abandon their plans. The 1996 Telecommunications Act lifted ownership restrictions, allowing telecoms to own cable systems, thus eliminating Video Dial Tone's original purpose and seemingly dooming ADSL.

The Internet Boom

The early 1990s saw the rise of the World Wide Web and the Mosaic browser, making the internet user-friendly and driving adoption, with about 20 million US households online by the mid-1990s. However, access was a problem. Most people used voiceband modems, which converted digital data into analog signals over phone lines.

These modems seized phone lines, dialed ISPs, and established a voice circuit, producing the iconic static and screeches. Voiceband modems used only about 3.3 kilohertz of frequency, resulting in slow speeds, topping out at 28.8 kilobits per second (56K modems didn't arrive until late 1997). They also tied up phone lines and incurred call charges, causing frustration for users and their families. A 1996 study by Pacific Telesis found that 16% of local calls failed to connect due to high internet usage, compared to a historical 1% failure rate.

ADSL did not block phone lines. Recognizing the growing demand for internet, Alcatel quickly pivoted ADSL from video to broadband. There was significant technical overlap, as both required small uploads for content requests and large downloads. However, telecoms, perhaps wary after Video Dial Tone's failure, were initially hesitant to embrace this pivot, questioning the technology's readiness and their ability to support millions of customers.

Alcatel's ADSL Ascendancy

Rising competition eventually forced telecoms' hand. Cable firms, having invested billions in new infrastructure, faced challenges from satellite TV and realized they needed to offer internet. The 1996 Telecom deregulation also created "Competitive local exchange carriers," obligating the Bells to lease their copper lines and equipment to startups, who then resold them to ISPs.

This compelled the Baby Bell telecoms to develop their own internet strategy. To offer broadband internet, they needed to purchase and install ADSL equipment, including:

• Customer premises equipment: A splitter and ADSL modem installed in the household. The splitter separated analog voice signals from data signals, and the modem facilitated internet communication.
• Telecom equipment: The Digital Subscriber Line Access Multiplexer (DSLAM), a large server-like device in the telecom's central office, received signals from connected households and routed them to the appropriate ISP.

Standardization and group purchasing were crucial for cost reduction, similar to how cable providers lowered cable modem costs. In August 1996, four Baby Bells (Ameritech, BellSouth, Pacific Bell, and SBC Communications), serving 65% of the US population, formed the Joint Procurement Consortium (JPC) and issued a request for proposals.

Alcatel aggressively pursued the contract and was shortlisted alongside Westell and Ericsson. In October 1996, Alcatel won based on price and its comprehensive end-to-end solution, from modems to DSLAMs, marking a significant victory for its DSL equipment division.

Commercial deployments in the US began around 1999. Aggressive advertising and sales by the Bells led to approximately 2 million DSL subscribers by 2000, nearly doubling the following year, and reaching almost 12 million by 2004, keeping pace with cable's 15 million. ADSL became the primary internet access technology for the Baby Bell telecoms, overshadowing other DSL technologies like HDSL, SDSL, IDSL, and RADSL, which were eventually encompassed by the term xDSL, and ultimately, ADSL.

Going Around the World

The JPC's buying power provided initial scale in the US, but Alcatel, as a global firm, quickly expanded its sales to telecoms worldwide. By 1999, they had shipped a million lines. DSL made significant inroads overseas, particularly in:

• Singapore: Singtel was an early adopter, conducting trials in the late 1990s and early 2000s, though they later transitioned to fiber due to the country's small size.
• Korea, Japan, and Taiwan: These countries, with dense apartments and legacy copper loops, were well-suited for ADSL. Korea rapidly adopted DSL, reaching over 10 million users by 2002.
• Europe: In the early 2000s, DSL became the leading broadband technology in OECD countries, especially where cable internet was less prevalent and regulators compelled incumbents to open their lines to DSL-only startups.

DSL's subscriber base peaked in the late 2000s and early 2010s with approximately 360 million lines. Alcatel maintained a 40-50% market share during this period and continued to enhance DSL transmission speeds. VDSL2, for instance, could achieve up to 200 megabits per second.

While DSL has since been superseded by faster fiber or cable technologies and is considered a legacy technology in decline, its decline is slow. AOL, for example, only phased out its dial-up service in September 2025. Even now, over a hundred million DSL subscribers still access the internet via copper wires. With its decent speeds and deep international penetration, DSL is expected to remain in use for a considerable time.

Conclusion

DSL holds a unique place in internet history. While not the initial internet experience for many, nor the current access method, its near-miraculous repurposing of existing copper infrastructure brought hundreds of millions online. Alongside cable modems, DSL helped establish the internet as the always-on, ever-present experience it is today. Its speed enabled people to first experience the rich content that fueled the growth of major internet businesses. This was a significant achievement for a technology not originally designed for web browsing, a fortunate accident that accelerated the development of one of history's most powerful tools.