Linux: Underpins ~60% of all web servers
Red Hat: $34bn behemoth that proved open source could be profitable
Linus’s Law: “Given enough eyeballs, all bugs are shallow”
AI Enterprise Adoption: Will need guarantees of reliability and safety
TL;DR: Red Hat revolutionized the software industry by proving that a business model built on "free" code could achieve massive scale (and revenue) by prioritizing enterprise-grade reliability over proprietary licenses. By shifting the value proposition from the software itself to a subscription-based guarantee of security and support, they made open source software accessible to high paying enterprise customers. Today, as the AI landscape mirrors the "Wild West" era of early Linux, the industry faces a familiar gap between rapid innovation and enterprise readiness. The next multi-billion dollar opportunity lies with a new generation of companies that can provide the same trust and stability for open-source AI that Red Hat once established for the operating system.
Open Source Beginnings and the Foundation of Red Hat
Red Hat is the poster child of financially successful companies built on top of open source software. Before Red Hat, it was not clear whether it could be done; but looking back, they truly were ludicrously successful (eventually being acquired by IBM for $34bn). This week we will explore the early beginnings of Red Hat, the open source environment at the time, how they spearheaded a new open source business model, and consider the open source opportunities ahead.
Red Hat started as a merger between Bob Young’s retail software catalog business and Marc Ewing’s (who used to wear a red hat) technical Linux distribution. Bob brought marketing and distribution expertise, while Marc had created a Linux distribution with superior package management.
Prior to Linux, there was no widespread successful open source operating system. In the early days of computing, the concept of proprietary software did not exist. Companies like IBM and DEC made money by building and selling hardware systems, which simply came with software bundled. While at universities, researchers and academics would share code freely with each other. In the 1970s and 80s, companies like Microsoft pioneered the incredibly high margin software license model. In response, Richard Stallman, a programmer at MIT’s AI lab, launched the GNU Project and created the GNU General Public License (GPL), a “copyleft” license that required derivative works to also remain free.
In 1991, Linus Torvalds, a Finnish university student, released Linux version 0.01 as a free kernel (the core component of an operating system that handles most fundamental tasks). By 1992 it was released under the GPL, which allowed anyone to use, modify, or redistribute it (Linux Evolution), and enabled its rapid adoption and massive decentralized collaborative development.
Side Note: The Linux Bazaar
The scale of Linux’s communal development was incredibly novel at the time and inspired a notable 1997 software essay (The Cathedral and the Bazaar) that depicted two versions of open source software:
The Cathedral: where code is freely available with each new release, but code written between releases is held internally and is proprietary.
The Bazaar: where code is shared publicly over the internet as it is written.
The author, Eric Raymond, makes the argument that the bazaar model is superior because “given enough eyeballs, all bugs are shallow” (termed Linus’s Law). If open source software is developed in the public, there will be more developers to scrutinize and improve it, which will speed up development and make for more robust code. Interestingly, Linus Torvalds also went on to create Git, which as a distributed, clone-everything, style version control system was perfectly aligned with the bazaar model of open source software development.
Red Hat’s Initial Success
Red Hat, along with others, initially packaged Linux in boxed sets of CD-ROMs and printed manuals that sold in retail stores. Red Hat scaled from inception in 1995 to $100m+ annual revenue within 7 years, even through the dot com bust. They achieved a ~$3.5bn first-day valuation through their IPO in 1999, which included a novel “community share” program (and nod to the open source community), wherein a portion of shares were reserved for core open source contributors to purchase. Many of these developers saw significant immediate returns (SonarSource).
The Pivot to Enterprise and the Guarantee Subscription
Even though Red Hat achieved initial success with the boxed set model, it was fundamentally flawed as the code could be copied and distributed for free - and leadership knew it. In late 1999, Red Hat acquired Cygnus Solutions, a company that was monetizing free software through support contracts rather than boxed set sales. This acquisition set the stage for pivoting the business away from mass market retail sales to the enterprise grade services and assurance subscription model it is known for today. During the early 2000s Red Hat left the boxed set model, but also separated itself from the “open core” model leveraged by other vendors, where proprietary features were sold on top of an underlying open source core. Rather, Red Hat provided all of its software open source, but sold the guarantee of reliability, security, and updates that were so important for enterprise grade customers.
Open Source: Core Underlying Infrastructure
Over the next two decades Red Hat cemented itself as the dominant enterprise grade Linux server OS provider and achieved “Switzerland” status as a neutral platform running across all major clouds (AWS, Azure, GCP). Linux as a whole commands close to 60% of the web-server market. By 2019, when IBM acquired the business, Red Hat was doing $3.2bn in annual revenue. Today the Red Hat subsidiary is on a nearly $8bn run rate.
Much of the world runs on core open source technology. Even state of the art technology is built on top of public research as well as open source projects. Major LLMs, for example, train on public and open source data. Anthropic, one of the best coding agents today, leverages The Pile and CommonCrawl for its training data, which both include significant amounts of open source code from GitHub repositories.
While closed models dominate the Western markets, 2025 saw the rise of open source Chinese LLMs, most notably with DeepSeek-R1. Many Western companies have an aversion to leveraging Chinese software and proprietary services due to security and snooping concerns. By publishing open source models, Chinese firms that are innovating in the AI space can still gain market influence. A positive side effect for the industry as a whole is that it forces further competition in an already highly competitive LLM landscape, fostering further innovation from both open source and closed source providers.
We are in a “Wild West” era of AI, with immense value creation (and disruption), but somewhat unclear value capture. We see new models getting pushed monthly with incredible gains, but also instability. As we wrote about last week, deep AI adoption is actually fairly limited. The future is here, but it is unevenly distributed. For true enterprise grade adoption we will need more mature and robust systems, but we will also need reliability. A new class of Red Hat style companies will need to provide the guarantees and long-term support of open source AI systems that large enterprises can rely on. While certainly a hard problem to tackle, it is a worthy endeavor.
Takeaways: Red Hat revolutionized the software industry by proving that a business model built on "free" code could achieve massive scale (and revenue) by prioritizing enterprise-grade reliability over proprietary licenses. By shifting the value proposition from the software itself to a subscription-based guarantee of security and support, they made open source software accessible to high paying enterprise customers. Today, as the AI landscape mirrors the "Wild West" era of early Linux, the industry faces a familiar gap between rapid innovation and enterprise readiness. The next multi-billion dollar opportunity lies with a new generation of companies that can provide the same trust and stability for open-source AI that Red Hat once established for the operating system.
Have a great weekend,
Josh