The Biggest Open-Source Software Company Just Sold for $34 Billion. How Did Free Become a Business? | QDaily Business History

The word "open source" turns 20 this year.

Last Sunday IBM announced the biggest acquisition in its history — buying Red Hat for $34 billion.

IBM buying companies is no surprise; nearly every big turn the company has made ran through an acquisition. But Red Hat is the world's largest developer of the open-source Linux system, and its operating system does not charge tens of dollars a machine in license fees the way Microsoft Windows does. A business that lives right alongside "free" sold for $34 billion, and IBM paid a 60% premium for it.

How open-source software makes money isn't that complicated. Red Hat's current CEO, James Whitehurst, gave an example in a 2012 interview: the New York Stock Exchange trades stocks on free Red Hat Linux systems. Intel had recently upgraded its processors, and the exchange wanted to use the new chips to work faster — but that meant someone had to deploy new systems and applications on the new hardware. The same thing comes around every few years. NYSE could keep a team in-house to do all the maintenance and development itself, or it could pay Red Hat's engineers to do it. A good number of enterprise customers pick the second option.

What Red Hat actually provides is far more complicated than that example, but the principle is the same.

Among Linux companies of its kind, Red Hat is the biggest. Its annual revenue now tops $2 billion and grows more than 20% a year.

Windows still rules the personal computer, but in the narrower market for enterprise server operating systems, IDC data from 2017 puts Microsoft at 50% and Red Hat at 32.7%. Among Linux distributions, Red Hat holds 70% of the market.

Products that make money off open source are common now, and none is more successful than the Google-led Android Open Source Project (AOSP).

But how open-source software should make money — or whether it should make money at all — was not settled when it was born.

The word "open source" was coined to reject "free of charge"

Open source always makes people think of the internet's spirit of free sharing at no cost, as if profit were its natural opposite.

There is no denying that open source still carries a streak of idealism. On GitHub, the open-source community's home, developers give away their ideas every day.

But commercial thinking has never been far from the open-source world. Twenty years ago, the term "open source" was coined precisely to shake off the impression of "free of charge."

In 1998, in an office in Mountain View, California, Eric Raymond — a 41-year-old hacker and the maintainer of The New Hacker's Dictionary — gathered the most active Linux developers of the day, including Linux's inventor Linus Torvalds, to talk about how to revise and spread the theory of free software.

Only a dozen or so people came, but this was the starting point of "Open Source."

Before that, "Free Software" was the common term, and its history runs back further. In 1969, before the internet existed, Bell Labs built UNIX, a set of operating-system code that could be copied quickly and passed around on disk.

UNIX's creators released the operating system and its source-code license to the public for free, and handed it in particular to many universities for teaching. Later, though, UNIX's owner AT&T made it proprietary, and users had to pay license fees of more than $10,000.

That provoked resistance from the champions of free software. In 1984, MIT computer science professor Richard Stallman started a project called GNU (GNU's Not UNIX!) and founded the Free Software Foundation (FSF), setting freedom and zero cost against the license fees big companies demanded.

The FSF proposed copyleft, the mirror image of copyright: the author of the source code still holds the copyright, but licenses any user to copy, modify and distribute the code, and even to sell it for profit — on the condition that all the code stays public and transparent.

That made it possible to earn money while still giving software away. The FSF floated a few concrete ideas: selling copies of free-software packages on CD or floppy disk; offering installation, training and guidance; customizing software, and so on.

Working from that, in 1989 the first company to sell consulting services around free software, Cygnus Solutions, was born (Red Hat bought it in 2000). Its slogan was "Making free software affordable." Its founder, Michael Tiemann, was also at the 1998 open-source meeting.

Stallman is revered as the touchstone of the free-software movement, but he was not invited to that meeting of a dozen people. The people who were there felt that the word "Free" in the FSF's "Free Software" made people think of free of charge — which was not the core of free software. Programmers were not a charity, and the word could bend the movement off course.

So they decided to replace "Free Software" with "Open Source": plain and direct, saying nothing about commercial choices or whether the software costs money.

The analogy the open-source community now uses to separate open from free is "free as in beer" versus "free as in speech." Free beer costs you nothing, but you get no say in it — Microsoft's Internet Explorer, Adobe's Flash: free to download, free to use, impossible to modify.

Free speech is a right. It is like the beer's secret recipe: once it is public, you can brew and improve the beer yourself. Take Chrome and Firefox — anyone can take the source code and build a different browser out of it. What matters is the freedom, not the software itself.

This became the mainstream thinking in the open-source community, and the larger reason people accepted it was the era's backdrop: big companies had a lock on the software industry.

When Linux was born, Microsoft's Windows already held more than 80% of the market and the company's annual sales had crossed $1 billion for the first time. Four years later, Bill Gates's personal fortune reached $12.9 billion, making him the youngest man ever to become the richest in the world.

Windows itself and Microsoft's applications were both closed commercial products. In 1996 the United States passed the Digital Millennium Copyright Act, which gave Microsoft's digital monopoly the backing of law. Corporate and individual users paid once for the right to use the software; anyone who modified or redistributed it was breaking the law. At bottom this was still the copyright logic of a physical product like a book.

Copyright became the foundation of Microsoft's commercial empire. Bill Gates had come out against free software as early as 1976. In the famous "An Open Letter to Hobbyists" he wrote at 22, he asked: "Who can afford to do professional work for nothing? What hobbyist can put three man-years into programming, finding all bugs, documenting his product and distribute it for free? The fact is, no one besides us has invested a lot of money in hobby software."

Steve Ballmer, who took over as CEO in 2000, called Linux "a cancer."

The open-source community had more than a little of that us-against-Microsoft coloring. Programmers who were sure open source would win once drew Gates as a Nazi, mocking the dictatorship and the closed doors of his desktop operating system.

But black and white are rarely that cleanly divided.

Open-source companies had their own hot moment too, and it was part of the 2000 dot-com bubble

Linux was born against the backdrop of the open-source movement. In 1991, Linus Torvalds, then a sophomore, wrote the first version of Linux and opened the source code, so anyone interested in the operating system could modify it and use it.

Plenty of companies started businesses on the technical support Linux made possible.

Red Hat was one of them. In 1993 Bob Young, a Canadian businessman who had already run a few computer- and printer-rental companies, founded ACC, whose main business was selling Linux and Unix software and earning revenue from phone support.

The next year the developer Marc Ewing released a version of Linux — on Halloween — and called it Red Hat Linux, after a hat his grandfather had given him in college.

"I'm not kidding, in 1994 the Linux community was so small everybody knew everybody." Young later said in an interview that he met Ewing through what he called "online dating."

In 1995 Bob Young bought Red Hat and merged the two companies into Red Hat Software.

The business found its feet fast. According to the IPO prospectus, Red Hat's revenue reached $480,000 in 1995, almost doubled the next year, and doubled again the year after that. By the time it went public in 1999, annual revenue had passed $100 million.

More than 90% of that revenue came from "software and related products" — the Linux distribution Red Hat built on the open-source code. Linux is like a published recipe; Red Hat cooked it into a finished dish and sold it to whoever wanted one. Red Hat also ran its own website and call center: customers who bought the software needed ongoing technical support, and that produced service fees.

Citing IDC data, the prospectus said that as of 1998, 56% of all newly licensed Linux installations came from Red Hat.

On its first day of trading, Red Hat's stock shot up to $54.40, nearly four times the $14 offering price, and two months later climbed past $135 — even though Red Hat reported a $90,000 net loss for 1999, ten times its loss in 1998.

"So far, it's convinced people that this [open source] can become a successful business," a Renaissance Capital analyst said at the time.

Red Hat's listing set off a wave of similar IPOs, and a crop of Linux stocks appeared.

VA Linux went public in December 1999; its main business was selling Linux servers. Its stock rose 733% on the first day. That same day, Red Hat's stock was up more than 1,000%.

An article in Wired said the wild enthusiasm analysts and investors showed for VA Linux came entirely from the Linux story Red Hat had started and from the sky-high valuations across the whole tech sector. Investors were sure Linux had enormous potential — that it might even produce a new Microsoft.

In fact, in the nearly five years after VA Linux was founded, its software and service sales added up to $44 million, against $25 million in losses over the same period. On December 9, 2002, exactly three years after the IPO, VA Linux stock closed at $1.19, down 99.5%.

Or take Andover, which also listed in December 1999 and rose 252% on its first day. As a Linux stock, Andover had nothing to do with Linux until its sixth month as a public company. It had been an aggregator giving developers an overview of site information; in June and August 1999 it bought the sites Slashdot and Freshmeat. Slashdot's core readers included the zealots of Linux and open source.

Jonathan Corbet, editor of Linux Weekly News and a programmer himself, asked at the time whether Andover was simply trying to ride the Linux story for the hype.

When the dot-com bubble burst in 2000, a great many Linux companies' shares fell to the floor. Many of them had given stock to Linux's creator, Linus, when they listed; the bubble made it worthless in short order.

The crash hit every tech company of the day. Amazon came within a step of bankruptcy, and it took Microsoft more than a decade to get its market value back to the peak of the bubble. But unlike the Linux stocks, which collapsed across the board, Microsoft kept producing extraordinary profits on the strength of its absolute lock on desktop systems and office software.

Before the bubble burst, The Economist had already warned that some Linux companies might be good businesses, but investors would have to accept that they would never have the operating profits of a traditional software company, let alone Microsoft's. Because Linux was "open source" — a kind of public property — these companies could not use patents to enforce a monopoly the way Microsoft did with Windows, and could not collect the rich returns that came with one.

But as The Economist also pointed out then, Microsoft's monopoly, which made everyone pay for software licenses and their upgrades, was hugely profitable in the short run for whoever owned such a platform, and yet it ran against the interests of every other company in the industry and dragged down the overall pace of technical progress.

Today's software industry is a compromise between the two sides of that old opposition.

Open source or not, "selling software" is no longer a software company's first choice

At the 2014 All Things Open conference, Red Hat founder Bob Young, by then gone from the company, recalled that his biggest inspiration for building his own company was Louis Gerstner, the man who once turned IBM around and the author of Who Says Elephants Can't Dance?:

"Gerstner came to IBM and took it from losses to profit in three years, which really was a miracle… Gerstner's insight came from talking with a huge group of IBM's customers and finding out that customers didn't actually like his products. They thought the products were fine, but whenever he sat down with a customer, someone would say there were plenty of products better than IBM's. He asked, 'So why do you buy from IBM?' And the customer said, 'Because IBM is the only company that can set up an office anywhere we do business.' So Gerstner understood that what he sold was not a product but a service.

"Gerstner talked about this publicly, and at Red Hat we'd say the same thing: 'OK, we have no product to sell, because everything we make is open source and anyone can innovate as fast as we can, so we're not really selling a product.' IBM's Gerstner taught us that customers don't buy products, they buy services — they buy the thing that makes them more successful. That was our early insight into what we were really doing: I was in the services business, and even back when we were selling Linux disks in a box, we treated that as a temporary stage until we were strong enough to sign service contracts with customers."

IBM had no great product. Red Hat had no product anyone was forced to buy. Both earned their profits by providing services.

Android, the most successful business ever built on open-source software, doesn't sell software either.

The first Android phone came out in 2008, entirely thanks to Linux. The first iPhone had been out a year by then, but the system kernel it used traces straight back to Steve Jobs's work at NeXT in the 1980s and 1990s — more than 20 years of accumulation.

Leaning on Linux, Android's inventor Andy Rubin led a team far behind Apple's in both development experience and size, and got Android phones to market in only five years.

Google used what the open-source community had built, and opened its own system for others to use. That also helped it win over the handset makers: traditional phone companies were losing ground to Apple, while Microsoft's phone system cost money and was slow to improve.

But Android's success is no open-source fairy tale. The Android Open Source Project (AOSP) has always existed, yet the services that shape how an Android phone actually feels are Google's own — Gmail, Search, Maps, the Play Store. Google bundled them together, phone makers had to install the whole set, and those apps let Google make its money from advertising.

It isn't only open source. The other big companies that used to sell software have largely stopped selling software too.

Apple used to put out a new version of OS X every other year and sell it on a disc or a USB stick. Starting in 2013, Apple gave major system upgrades free to anyone who bought a Mac. Users were still paying for those upgrades; the cost was simply folded into the price of the Mac.

Adobe used to make users buy expensive software once, use it for life, and pay again for each major new version. With the Creative Cloud (CC) service in 2012, users could subscribe by the month and keep using the whole suite as long as they kept paying.

The sharpest reversal, though, is Microsoft's.

Ballmer, the Microsoft CEO who called open-source software a cancer, stepped down in 2013. His successor, Satya Nadella, has been chipping away at the Windows division's standing ever since and has put the company's whole center of gravity on cloud computing. Microsoft's cloud was once built on Windows; it was even called Windows Azure.

Under Nadella, Microsoft is willing to give up the absolute control it had in the Windows era in exchange for more enterprise customers, whether they run Windows or Linux. At a 2014 event he declared "Microsoft loves Linux," and sent engineers to support open-source projects directly.

That same year, Microsoft's cloud platform began supporting Red Hat's Enterprise Linux, a direct competitor to Windows Server.

In 2016, according to figures GitHub published, Microsoft contributed to 16,419 open-source projects, passing Facebook to become the largest contributor in that community.

Once Microsoft no longer needed — or could no longer ask — users to keep buying Windows, and turned to making money on services instead, its fight to the death with the open-source community stopped being necessary.

Just a week ago, Microsoft's $7.5 billion purchase of GitHub closed. GitHub promised to keep supporting open source.

IBM's purchase of Red Hat is the same kind of move. IBM is still not a company especially friendly to open source: it will write into a contract that an employee must stop contributing code to an open-source project, even one that doesn't conflict with IBM's business. That hasn't stopped IBM from wanting Red Hat's strong cash revenue and the groundwork it has laid for moving enterprise customers to the cloud.

After the dot-com bubble, Red Hat became one of the few survivors of that IPO wave, and it still holds 70% of the global Linux market.

Red Hat went on to shore up its position by buying company after company — more than 30 of them. Its results have grown for more than 60 straight quarters. Its OpenShift technology lets a company deploy its systems on several cloud platforms at once and spread the risk. All of this appeals to an IBM in a hurry to catch Amazon and Microsoft.

"Every dollar spent on a Red Hat product takes $10 of market share away from proprietary software — software that wears a halo of patents and is really a pile of junk," Red Hat CEO Whitehurst said in that 2012 interview.

Now all of it is history.

Cover image from Pixabay