Skip to content

USB: The Port for Almost Everything

Abstract

On January 15, 1996, seven companies published the specification for a Universal Serial Bus, one rectangular socket meant to replace the zoo of serial, parallel, PS/2, SCSI, and game ports on the back of every PC. Its driving force was an Intel architect named Ajay Bhatt, who had to sell the idea inside his own company before he could sell it to an industry, and who became, thirteen years later, the only connector designer ever portrayed as a rock star in a television commercial. USB started slowly, was rescued by the one major computer maker that hadn’t helped design it, defeated a technically superior rival from Apple, and ended up as the most successful peripheral standard in computing history, so universal that the European Union eventually wrote it into law. Along the way it killed more ports than any technology before or since.

The Cable Chaos

A personal computer in 1994 was a museum of incompatible connectors: DB9 and DB25 serial ports for modems and mice, a Centronics parallel port for printers, PS/2 ports for keyboard and mouse (identical in shape, electrically different, and unforgiving if swapped), SCSI in several mutually hostile variants for scanners and external drives, a game port for joysticks, and proprietary connectors for everything else. Most required the right cable, the right expansion card, manually assigned interrupts, and a reboot. For the industry’s plan to make the PC a mass-market appliance (and, at Intel, to give ever-faster processors something to do) the back panel was a standing embarrassment.

Ajay Bhatt and the Seven Companies

Ajay Bhatt, an Indian-born computer architect who had joined Intel in 1990, concluded that the PC would never become a true consumer appliance until it had one self-configuring, hot-pluggable port for everything. The idea was not obviously a winner inside Intel (connectors were nobody’s glamour project) but Bhatt evangelized it, wrote the first specification drafts, and became the effort’s public architect. Intel built the first USB support chips in 1995, and a consortium of seven companies (Intel, Compaq, DEC, IBM, Microsoft, NEC, and Nortel) published USB 1.0 on January 15, 1996.

The design goals read like a direct negation of the port zoo: hot-pluggable (no reboot), self-identifying (the device tells the computer what it is), power-supplying (5 volts from the bus, no wall adapter for a mouse), and universal (one connector for keyboard, printer, camera, and disk). Two speeds served different classes of device: 1.5 Mbit/s “Low Speed” for keyboards and mice (sufficient for them to this day) and 12 Mbit/s “Full Speed” for everything else.

Bhatt’s later career made him the rare standards engineer with a public face. He went on to contribute to PCI Express and platform power-management architecture, accumulated over 130 patents, and in 2009 became the subject of an Intel television ad (part of its “Sponsors of Tomorrow” campaign) in which an actor playing him strode through an office to screaming fans. The real Bhatt received the European Inventor Award in 2013 and India’s Padma Shri in 2025. The one design decision he could never live down (the rectangular Type-A plug that fits only one way but looks like it fits two) has its own entry: USB Was Designed to Be Easy to Plug In. It Wasn’t.

The Slow Start and the Apple Rescue

USB 1.0 landed with a thud. Few devices appeared; Windows 95’s USB support was late and fragile; PC makers kept soldering on the legacy ports alongside the new one, giving peripheral makers no reason to switch. USB needed someone reckless enough to remove the alternatives.

That someone was the one major computer maker not among the seven founders. In August 1998, Apple (fresh from Steve Jobs’s return) shipped the iMac G3 with two USB ports and nothing else: no ADB, no Mac serial ports, no SCSI, and, most controversially, no floppy drive. Peripheral makers who had ignored USB for two years suddenly faced millions of iMac owners who could connect nothing old, and a wave of translucent, iMac-colored USB printers, scanners, and drives followed within months. Together with the cleaned-up USB 1.1 specification (September 1998) and Microsoft’s parallel “legacy-free PC” push, the iMac turned USB from a checkbox into the default. The standard’s decisive adopter was a company with no seat at its founding table.

USB 2.0 (April 2000) raised the ceiling to 480 Mbit/s (forty times faster) just in time for the gadget decade: USB flash drives displaced the floppy disk, external hard drives displaced SCSI, and every camera, MP3 player, and phone charged from the same port. Later revisions kept multiplying: USB 3.0 (2008, 5 Gbit/s), 3.1 (2013, 10 Gbit/s), 3.2 (2017, 20 Gbit/s), and USB4 (2019, 40 Gbit/s; 80 Gbit/s in its 2022 revision), a four-thousand-fold speed increase over the original Full Speed, through a compatible family of plugs.

⚠️ Dead End: FireWire, the Better Bus That Charged Rent

USB’s one serious rival came from Apple itself. FireWire (standardized as IEEE 1394 in 1995, begun at Apple back in 1986) was by most technical measures the better design: 400 Mbit/s when USB managed 12, peer-to-peer operation (two camcorders could talk without a PC in between), guaranteed-bandwidth isochronous transfers ideal for video, and up to 45 watts of power per port when USB delivered 2.5.

It lost anyway, and the reason has become a standards-war parable: FireWire charged rent. Implementing it meant licensing a pool of over 250 patents at roughly $0.25 per unit, trivial per device, poisonous at PC-industry volumes and PC-industry margins, and doubly unattractive next to USB, which Intel promoted with essentially royalty-free licensing and built into its chipsets. FireWire kept the professional niches where its design genuinely mattered (digital video camcorders were nearly all FireWire, and the first iPods synced over it) while USB took the other 99% of the market. USB 2.0’s 480 Mbit/s erased the headline speed advantage in 2000, camcorders went memory-card, the iPod went USB, and Steve Jobs (whose company had invented FireWire) declared it dead in 2008. The better architecture lost to the cheaper ecosystem: the same lesson as Token Ring vs. Ethernet, one layer up the cable.

USB-C and the Long Arm of the Law

The reversible USB-C connector (2014) fixed the eighteen-year-old orientation complaint and consolidated data, video, and up to 240 watts of power into one plug, which made it, for the first time, plausible as the charger for everything. The European Union had been pushing that direction since a 2009 memorandum of understanding herded phone makers toward micro-USB; when voluntary agreement frayed (Apple kept its proprietary Lightning connector) the EU stopped asking. A 2022 directive made USB-C legally mandatory for smartphones sold in the EU from December 28, 2024, and for laptops from 2026. Apple moved to USB-C with the iPhone 15 in 2023, a year ahead of the deadline, ending Lightning after eleven years.

It was a genuine novelty in computing history: a connector standard completed not by market victory but by legislation, the port so universal that a government made universality compulsory.

Legacy

USB did what its name promised, which almost no “universal” standard ever does. It erased the serial port, the parallel port, PS/2, the game port, consumer SCSI, the floppy disk (as transport), and eventually its rival FireWire and the proprietary phone charger. Billions of USB devices now ship every year; the humble 1996 Low Speed mode still carries most of the world’s keystrokes. And the standard’s history keeps its founding irony intact: designed by seven companies to tame the PC’s back panel, it was made ubiquitous by Apple, perfected as USB-C, and finally universalized by an act of the European Parliament.

📚 Sources