U.S. Robotics

A Chicago company that did for modems what Formula 1 did for ordinary cars: its own HST protocol, which the sysops of electronic bulletin boards bought for speed rather than for compatibility. It won on speed but not on the standard, and in 1997 it sold itself to the networking giant 3Com at the very peak of the modem era, a few years before broadband internet displaced dial-up.

In brief

U.S. Robotics is a modem company from Chicago, founded in 1976 by five acquaintances from their student days who each chipped in $200 for an acoustic coupler. The company grew into one of the largest players in the North American modem market not through openness, as its competitor Hayes did, but through the opposite: a closed but fast protocol, HST, sold first and foremost to a professional audience, the sysops of electronic bulletin boards. Its own microchip instead of other companies' chips gave it a real speed advantage, and the same engineering groundwork, brought down in price, produced the budget Sportster and made USR the dominant brand in the mass market. In 1997, at the peak of the modem era, the company was bought by the networking giant 3Com (the deal was announced at $6.6 billion, and on the closing date it was worth $8.5 billion), but only three years later 3Com itself took the asset apart, spinning off Palm and the remainder of the U.S. Robotics business into separate companies, without ever taking a place in the broadband internet market that made dial-up modems unnecessary.

How it started (the founders)

Casey Cowell, a 23-year-old economics graduate of the University of Chicago, dropped out of his doctoral program and returned to Chicago. Together with his classmate Paul Collard and three more acquaintances (according to one summary, Steve Muka, Stan Metcalfe, and Tom Rossen estimate), they each put in $200 and, according to the same summary, rented a windowless room above an army surplus store on Lincoln Avenue in Chicago. The company's name refers not to their own specialty but to a favorite book: "U.S. Robots and Mechanical Men, Inc." from Isaac Asimov's I, Robot. The first product was an acoustic coupler cast in homemade molds; the first year brought in $50,000 in sales, half of it from reselling other companies' equipment rather than from a product of their own. The company took up modems in the usual sense only in 1979, when FCC decisions allowed equipment from companies other than AT&T to be connected to the telephone network, the same regulatory shift on whose wave the future competitor Hayes had started a year earlier.

Year-by-year timeline

1976: Cowell and four acquaintances from the University of Chicago found U.S. Robotics in Chicago with $200 each fact
1979: after the change in FCC rules, the company releases its first modem as a stand-alone product fact
1986: Courier HST is introduced, a proprietary protocol with trellis coding at 9,600 bps, four times faster than standard 2400-baud modems at roughly twice the price; it runs at full speed only with another HST modem fact
1989: HST is pushed to 14,400 bps fact
1992-03: HST is pushed to 16,800 bps fact
1993: launch of the budget Sportster line on the same DSP architecture as the expensive Courier; aggressive retail expansion and pricing make USR the dominant supplier in the mass market fact
1995-09: purchase of Megahertz Holding (PC Card), ISDN Systems Corporation, and Palm Computing (for $44 million) fact
1996: the PalmPilot comes to market (base model $299) and takes the leading market share among handheld organizers; by the same year USR controls, by various estimates, 22–27% of the North American modem market, with profit up 158% over the year estimate
1996-10: announcement of the x2 technology, USR's first step in the 56K race fact
1997-01: Cowell steps down as president of USR (keeping the posts of chairman and CEO), a month before merger talks begin fact
1997-02-26: 3Com and USR sign the Amended and Restated Agreement and Plan of Merger; the exchange ratio is 1.75 shares of 3Com for one share of USR; the deal is announced as a $6.6 billion stock swap fact
1997-06-12: the deal closes; because 3Com's share price rose over the four months between announcement and closing, the actual value on the closing date is already $8.5 billion fact
1998: the ITU approves the V.90 standard, compatible with neither USR's X2 nor Rockwell/Lucent's K56flex, and the 56K war ends in victory for a third option fact
2000-03: 3Com spins off Palm as a separate public company; at the IPO the market values Palm at $53 billion fact
2000-06: 3Com separately spins off the brand and the remainder of the U.S. Robotics business, which goes on selling dial-up modems in a niche

Lesser-known but significant facts

The company's name is a literary reference, not a technical one: the five founders named the firm after the fictional corporation from Asimov's I, Robot, not after anything connected with real robotics, which the company never worked on.
Technically, HST did not get along with other makers' modems: it delivered its full speed of 9,600–16,800 bps only when paired with another HST modem, meaning that both the sysop and the caller had to buy USR specifically to get the benefit, and that explains why USR handed out large discounts to sysops in particular, since they were the end of the line that had to be won over first.
The Sportster was not a separate development but the same chip, only cheaper: the budget line of 1993 used the same DSP architecture as the premium Courier for sysops and corporate customers, so one engineering asset was sold twice at different prices.
Palm was bought for $44 million, and five years later it alone was worth $53 billion at its IPO, roughly 1,200 times the purchase price and more than six times the final value of the entire 3Com–USR deal ($8.5 billion); the most expensive asset came to 3Com as an add-on to the modem business.
Cowell left the post of president exactly one month before talks on selling the company began, in January 1997, while remaining chairman and CEO; the chronology in the legal document does not explain the reason, but the coincidence in timing is noticeable.

Legend vs. the record

Legend: U.S. Robotics simply made modems cheaper than Hayes. That is how the story gets simplified after the fact, especially next to the account of the inexpensive Sportster. The record shows the opposite at the start: the flagship HST protocol of 1986 cost roughly twice as much as an ordinary 2400-baud modem and was closed and incompatible with other makers' devices, a premium rather than a budget product, aimed at the narrow professional audience of BBS sysops, with large targeted discounts for them in particular. USR became cheap only seven years later, in 1993, when the same engineering groundwork (the DSP chip) was brought down in price into the Sportster. Verdict: first an expensive closed protocol for professionals, and only later, as a separate step, a cheap mass-market line; these are two different stages of the strategy, not a single pricing policy from the very beginning.
Legend: X2 won the 56K war. People sometimes write this, remembering X2 as the more widely promoted brand of that era. The record shows that neither X2 (USR) nor the competing K56flex (Rockwell/Lucent) became the final standard: in 1998 the International Telecommunication Union (ITU) adopted V.90, a protocol incompatible with both earlier formats, which became the common denominator of the industry. Verdict: neither of the two warring sides won the war; a third, regulatory option won, and it forced both companies to make their equipment all over again.
Legend: the merger with 3Com was a success. As far as the deal price goes, yes: USR shareholders got $8.5 billion at closing against the announced $6.6 billion, because 3Com's stock rose. The record shows a very different picture over the long run: already in 2000, less than three years later, 3Com itself took the combined company apart, spinning off Palm as a separate company (the market valued it at $53 billion, more than six times the price of the entire 1997 deal) and separately spinning off the remainder of the U.S. Robotics business, which went on selling dial-up modems. Neither 3Com nor the former USR ever took a position in the approaching broadband era; according to an independent assessment, 3Com had routers to sell to broadband customers but missed out entirely on selling the modems themselves of the new generation (DSL/cable). For 3Com itself the end was its purchase by HP in 2009 for $2.7 billion, less than a third of the final value of the 1997 merger. Verdict: the deal was financially favorable for the USR sellers at the moment, but strategically it did not create the promised networking giant; it fell apart faster than it could justify itself, and both sides missed the next technology cycle.

The first growth lever

The growth lever for U.S. Robotics was not an open standard (that was the move of its competitor Hayes) but the exact opposite strategy: a closed but genuinely faster protocol, sold first to the most demanding and best-paying part of the market. HST in 1986 gave four times the speed of the standard modems of the time, but it ran at full speed only when paired with another HST modem, a technical limitation that turned into a business tool: for a BBS network to actually speed up, the sysop and his regular callers had to switch to USR collectively, and that is exactly why the company offered sysops aggressive personal discounts, since they were the side of the network that had to be won first for the speed effect to become noticeable to users at all. In this way USR took the position of the brand for the professional core of the market long before it even began competing for the mass buyer. The second move, delayed by seven years, finished the job: in 1993 the same DSP architecture developed for the expensive Courier became the basis of the budget Sportster, not a new development but the same asset sold down-market to the mass market, and that is what made USR the largest modem supplier in North America by the mid-1990s.

The paired story (Hayes + U.S. Robotics)

Hayes and U.S. Robotics grew up on the same wave: the FCC allowed equipment from companies other than AT&T to be connected to the telephone network, and both companies started out with modems for enthusiasts. After that their paths diverged. Hayes won the battle for the interface: its AT commands became the language that every modem understood, including USR's modems. U.S. Robotics won the battle for speed: its own DSP architecture instead of the bought-in chips Hayes relied on, and in 1993 the cheap Sportster brought that advantage to the mass market. The endings differed too: Hayes went through bankruptcy twice, and in 1999 its modem assets went at auction for about $5 million (see the Hayes dossier), while U.S. Robotics sold itself to 3Com in 1997 in a deal announced at $6.6 billion and worth $8.5 billion on the closing date.

Parallels today (projects from the catalog)

Armature (armature): the same move of a tool for the technical elite on a standard that has not settled yet. Armature makes observability for companies whose product is an MCP server or a Claude Connector, meaning it builds on the early, still-forming infrastructure of AI agents, roughly the way HST in 1986 bet on BBS sysops before the modem market consolidated around Hayes compatibility. The difference between the eras: MCP as a protocol does not belong to Armature (unlike USR's closed HST), so the bet here is not on incompatibility as a barrier but on speed of getting to market on top of someone else's standard. view this project's dossier →
Breakcold (breakcold): the mechanics of laying out one and the same technology asset across a product line for different segments. Breakcold builds a CRM for salespeople on a shared base (collecting conversations from different channels + AI agents), which it then adapts for agencies and for solo founders as well, much as one DSP chip developed for the expensive Courier became the basis of the budget Sportster and opened the mass market to USR. The difference between the eras: at Breakcold the segmentation runs through the pricing plans of a single cloud product rather than through physically different lines of hardware. view this project's dossier →

What a builder can take from this in 2026

A closed but objectively faster product can be as much of a growth lever as an open standard: just address it first to the audience for whom speed matters more than compatibility, and subsidize that audience's entry in particular.
Do not be afraid to sell the same technology asset twice at different prices: a premium version for professionals and a budget one for the mass market on a single engineering base pay back the development faster than building two separate products.
A good price for selling the company is not the same thing as a successful deal. USR sold at the peak of its share price, but the combined company could not hold its position even three years later, so it is worth asking separately not how much you will be paid now, but what the buyer will do with the asset next.
The most valuable asset in a deal is not necessarily the one the deal was made for. 3Com was buying USR's modem business, not Palm, but it was Palm that turned out to be the asset worth more at its IPO than the entire deal.
A technology victory in one cycle does not protect you from the next. USR won the modem speed race, but neither it nor 3Com built a position in broadband internet, the market that wiped out that very race.

Discrepancies and what we could not verify

The closing date of the 3Com deal: FundingUniverse (a secondary source) gives July 1997, but the primary S-4/A filing directly names June 12, 1997 as the date the deal was completed; we preferred the primary document.
North American modem market share by 1996: FundingUniverse puts it at more than a quarter (~25%+), while a Crain's Chicago Business article (August 1996), as retold, gives 22% of a $2.5 billion market; the full text of the article is unavailable, and the exact figure is not confirmed estimate.
The exact retail prices of the Courier and the Sportster in 1993 were not found; only the general architecture (a single DSP chip) and the fact of an aggressive pricing strategy are confirmed, not specific dollar figures unverified.
The first office on Lincoln Avenue in Chicago is known only from a secondary summary estimate.
The reason Cowell stepped down as president of USR in January 1997, a month before merger talks began: the fact itself is confirmed by a primary document, but its connection (if there is one) to the subsequent sale of the company is neither explained nor checked anywhere unverified.
The full list of the five co-founders: Cowell and Collard as co-founders in 1976 are confirmed by two independent sources (Wikipedia, independently of FundingUniverse/the aggregator; Cowell also by the primary S-4/A); the names of the other three (Steve Muka, Stan Metcalfe, Tom Rossen) rest on only one aggregated summary without a primary source estimate.

Sources (primary first)

Secondary (context, cross-check):

Wikipedia — USRobotics — a pointer and independent confirmation of the year of HST (1986)
Encyclopedia.com — U.S. Robotics
Crain's Chicago Business, "Modem War," August 17, 1996 — read only through a search summary, the full text is behind a paywall
ManualsLib / legacy.hylafax.org — technical documentation for the Courier HST (aggregated through search)
mergr.com — the USR/Palm deal card (the $44 million amount; direct access returned 403)