The Thirty Percent
The number that defines Valve Corporation is not a revenue figure, a headcount, or a valuation multiple. It is a percentage: thirty. That is the standard commission Valve extracts from every game sold on Steam, the digital distribution platform that — depending on whom you ask — either represents the most elegant toll booth in the history of software or the most brazen extraction regime in consumer technology. On a platform hosting more than 70,000 games, serving north of 130 million monthly active users, and processing billions of dollars in annual transactions, that thirty percent represents something close to a private tax on the entire PC gaming economy. Valve did not invent video games. It did not invent digital distribution. What it invented was something more durable and more difficult to replicate: the default infrastructure layer through which an industry worth over $180 billion globally conducts the majority of its PC business. And it did so while employing roughly 400 people — a ratio of revenue-per-employee that makes even the most efficient SaaS companies look bloated.
The paradox at the center of Valve is that it is simultaneously one of the most admired and one of the most resented companies in gaming. Developers praise the tools, the community infrastructure, the workshop system, the Steamworks API — then quietly seethe at the margin they surrender for the privilege. Competitors, from the Epic Games Store to GOG to EA's Origin, have spent billions trying to replicate what Valve built, offering lower take rates, exclusive titles, free games as acquisition bait. None have made a meaningful dent. Steam's market share on PC game distribution remains somewhere between 50% and 75% depending on how you count, a dominance so thorough it has attracted antitrust scrutiny in both the United States and the European Union.
What follows is the story of how a company born from a single first-person shooter became the invisible operating system of an industry — and the structural question of whether its dominance is the product of genuine platform superiority or the compounding inertia of a network effect that became, at some point, self-reinforcing beyond challenge.
By the Numbers
Valve Corporation
~$13B+Estimated annual revenue (2024)
~400Employees
130M+Monthly active Steam users
70,000+Games available on Steam
30%Standard platform revenue share
~$30M+Estimated revenue per employee
$0External capital ever raised
A Mod Developer Walks Into Microsoft
Gabe Newell is the kind of figure who resists the standard Silicon Valley founder narrative — no garage-to-greatness arc, no dropout mythology polished for magazine covers. He dropped out of Harvard, yes, but he dropped out to join Microsoft in 1983, at the age of twenty, and spent the next thirteen years there — not in the executive suite but deep in the engineering trenches, working on the first three iterations of Windows. He was employee number 271. By the time he left in 1996, he was a multimillionaire from Microsoft stock options, with a bone-deep understanding of how platforms work: how they create lock-in, how they compound in value, how the real power in a software ecosystem accrues not to the applications but to the substrate on which applications run.
His co-founder, Mike Harrington, came from the same Microsoft cohort — another engineer who understood systems rather than products. They left together, pooling their Microsoft wealth to self-fund a company in Kirkland, Washington, with no venture capital, no board of directors, and no immediate commercial plan beyond a single conviction: that first-person shooters, then a niche genre dominated by id Software's Doom and Quake, could be elevated into a narrative medium. They licensed id's Quake engine and set about building what would become Half-Life.
The decision to self-fund was not merely a lifestyle preference. It was a structural choice that would define Valve's entire trajectory. With no external investors, there was no board to appease, no quarterly earnings cadence to manage, no pressure to IPO, no dilution to fear. Newell retained total control. And total control, in the hands of someone who understood platform economics from the inside of the most successful platform company of the twentieth century, would produce something unusual.
The easiest way to stop piracy is not by putting antipiracy technology to work. It's by giving those people a service that's better than what they're receiving from the pirates.
The Game That Built the Company That Abandoned Games
Half-Life shipped on November 19, 1998, and it was a revelation. Not for its graphics or its gunplay — though both were excellent — but for its refusal to break the first-person perspective. There were no cutscenes, no loading screens between levels, no separation between narrative and gameplay. The player experienced the catastrophe at Black Mesa Research Facility in an unbroken stream of consciousness. It won over fifty Game of the Year awards. It sold roughly nine million copies. It proved that Newell and Harrington's thesis — that shooters could be narrative vehicles — was not just correct but commercially explosive.
But the more important thing Half-Life produced was not revenue. It was a community. The game shipped with modding tools, and those tools generated an ecosystem. Counter-Strike, originally a fan-made modification by Minh "Gooseman" Le and Jess Cliffe, became one of the most popular competitive games in the world — eventually more popular than Half-Life itself. Valve didn't fight the mod community. It hired them. It acquired Counter-Strike, hired Le and Cliffe, and turned a free mod into a commercial franchise that would generate billions over the next two decades.
This pattern — build the tool, let the community create the value, then capture the value by owning the infrastructure — would become Valve's defining strategic motion. It was not quite open source and not quite proprietary. It was something more like platform capitalism in miniature: give away the means of production, own the means of distribution.
Harrington left in 2000, selling his stake back to Newell. From that point forward, Valve was Gabe Newell's company, wholly and irrevocably.
The Infrastructure Insight
The genesis of Steam was not a product vision. It was a logistics problem.
By 2002, Valve was managing an increasingly unwieldy collection of multiplayer games — Half-Life, Counter-Strike, Team Fortress Classic, Day of Defeat — that required constant patching, anti-cheat updates, and server management. The existing distribution model, which relied on physical retailers and third-party publishers like Sierra Entertainment, introduced agonizing delays between a patch being ready and that patch reaching players. Newell, whose formative career had been spent watching Microsoft build the dominant software platform of the 1990s, recognized the problem as architectural: Valve did not control its own distribution channel. It was a content company dependent on intermediaries. That dependency was a vulnerability.
Steam launched on September 12, 2003, initially as a mandatory updater for Valve's own games. The gaming community despised it. The client was slow, buggy, and intrusive. It required an internet connection — a genuine imposition in the broadband-scarce landscape of 2003 — and forced users to create accounts, accept terms of service, and submit to a DRM framework that felt, to many, like an insult. The forums were incandescent with rage.
Newell did not care, or rather, he cared precisely enough to fix the technical problems while refusing to abandon the strategic premise. The premise was this: if you could get every PC gamer to install a single client that served as both a game launcher and a storefront, and if you could make that client's value proposition — automatic updates, centralized game libraries, social features, cloud saves — compelling enough, you would own the distribution layer for an entire platform. You would become, in effect, the app store for PC gaming a half-decade before Apple launched the App Store for mobile.
How Valve bootstrapped its distribution platform
2002Valve announces Steam as a game updater and anti-cheat platform.
2003Steam launches; mandatory for Counter-Strike 1.6 online play.
2004Half-Life 2 requires Steam activation — first major retail game to mandate a digital platform.
2005First third-party games appear on Steam: Rag Doll Kung Fu, Darwinia.
2007Steam opens to third-party publishers; major titles begin arriving.
2010Steam reaches 30 million accounts. PC digital distribution tips from niche to norm.
2013Steam surpasses 65 million active accounts. Greenlight expands the catalog to indie developers.
The critical forcing function was Half-Life 2. When the most anticipated PC game in the world shipped on November 16, 2004, every copy — including those purchased on physical disc at retail — required Steam activation. Millions of gamers who had no interest in digital distribution were compelled to install the client, create an account, and link their purchase to Valve's platform. It was the most aggressive platform bootstrapping move in gaming history, the equivalent of Microsoft bundling Internet Explorer with Windows but executed by a private game studio with no regulatory oversight. Complaints were loud and sustained. Newell absorbed them. Within two years, those same angry users were buying their next game on Steam because, well, they already had the client installed, their friends were there, and the sales were remarkable.
The Seasonal Economics of Digital Desire
Valve did not invent the concept of a sale. But it may have perfected the psychology.
The Steam Summer Sale and Winter Sale, which began in earnest around 2009 and became cultural phenomena by 2011, were not merely discount events. They were behavioral design systems. Valve discovered — and the data apparently confirmed this in extraordinary fashion — that aggressive temporary discounts on digital goods, where marginal cost of distribution is zero, generated total revenue increases that dwarfed the per-unit revenue loss from the discount. A game priced at $60 that sold 100 copies per week might sell 40,000 copies during a 75%-off sale, generating more total revenue for both Valve and the developer than weeks of full-price sales combined. The math was elegant and self-reinforcing: developers began timing their release calendars around Steam's sale events, which attracted more buyers to the platform, which attracted more developers, which made the next sale even larger.
But the deeper effect was psychological. Steam Sales trained an entire generation of PC gamers to accumulate libraries far beyond their actual playing capacity. The average Steam user owns approximately 15 games but has played fewer than half of them. Power users own hundreds. The term "backlog" became a defining concept of PC gaming culture — the vast, half-guilty, half-delighted pile of unplayed purchases. Every unplayed game in that backlog was a unit of switching cost. Every dollar spent on Steam was a dollar anchoring the user to the platform. The library itself became the moat.
We used to think that our customers were the people who bought our games. But actually our customers are the people who buy each other's games.
The Flat World of Cabalistan
The organizational structure inside Valve has been, for two decades, the subject of fascination, skepticism, and mythologizing in roughly equal measure. The company has no formal hierarchy. No job titles. No managers, at least not in the conventional sense. Employees — Valve insists on calling them "contributors" — choose which projects to work on by physically wheeling their desks to the team they want to join. The Valve Handbook for New Employees, which leaked in 2012 and was later officially published, describes a perfectly flat organization where projects are initiated from the bottom up, where hiring decisions are made by peer consensus, and where annual compensation is determined by stack-ranking reviews conducted by colleagues rather than supervisors.
The reality, according to numerous former employees who have spoken publicly, is considerably more complicated. The flat structure, critics contend, produces not egalitarianism but invisible hierarchy — a system where influence accrues to those with the longest tenure and the closest personal relationships with Newell, where accountability is diffuse to the point of nonexistence, and where employees can be terminated without warning in periodic layoffs that feel arbitrary to those affected. Jeri Ellsworth, a hardware engineer laid off in 2013 along with roughly twenty-five colleagues, described the culture as "a lot like high school" — cliquish, socially navigable only by those with certain interpersonal skills, and punishing to those who lacked the political instincts to thrive in an environment with no formal power structures to appeal to.
Whether this culture is a genuine organizational innovation or a romantic self-delusion that masks a personality-driven autocracy is, in some sense, beside the point. What matters strategically is the output: Valve ships products on no discernible schedule, cancels projects with impunity, enters and exits hardware markets without apparent strategic coherence, and has gone thirteen years without releasing a mainline sequel to its most beloved franchise. The flat structure optimizes for one thing — the autonomy and satisfaction of Valve's approximately 400 employees, who are among the highest-compensated workers in the games industry. Whether it optimizes for anything else is an open question.
The Silence of the Lambda
There is a specific kind of pain that only Half-Life fans know. Half-Life 2: Episode Two shipped in October 2007, ending on a devastating cliffhanger — a character death, a narrative thread deliberately and viscerally severed mid-arc. Episode Three was expected within a year or two. It never arrived. Valve never officially cancelled it. They simply stopped talking about it, allowing the silence to calcify into a void that became, over the years, one of gaming's most famous non-events.
In August 2017, Marc Laidlaw — the lead writer of the Half-Life series, who had left Valve the previous year — posted a thinly veiled synopsis of the unproduced Episode Three on his personal website, gender-swapping all character names in what felt like an act of gentle, melancholy defiance. The fan community decoded it immediately. It was beautiful and sad. It was also an implicit admission that the game was never going to be made.
Then, in November 2019, Valve announced Half-Life: Alyx, a VR-exclusive prequel built for the company's own Index headset. It was, by all accounts, extraordinary — the first VR game that justified the platform, a technical and narrative achievement that proved Valve could still make games when sufficiently motivated. But the motivation was revealing: Alyx existed not primarily as a creative endeavor but as a hardware evangelism tool, a system seller for the Index the way Half-Life 2 had been a system installer for Steam. Valve made games when games served the platform. When they didn't, silence.
The silence told you everything about what Valve had become. It was no longer a game studio that happened to run a platform. It was a platform company that occasionally, when strategically convenient, made games.
The Workshop and the Skinners
In 2011, Valve introduced the Steam Workshop — a system that allowed users to create and share modifications, maps, skins, and other user-generated content for games hosted on the platform. It was a formalization of the modding culture that had given Valve Counter-Strike in the first place. But Valve, ever attuned to the possibilities of platform economics, took it further. For certain games — most notably Team Fortress 2, Dota 2, and Counter-Strike: Global Offensive — community creators could submit cosmetic items (hats, weapon skins, character outfits) for potential inclusion in the game. If Valve selected the item, it would be sold in the in-game store, and the creator would receive a revenue share.
This was, in effect, the gig-ification of game development. Valve outsourced a substantial portion of its content creation to an unpaid (or, more precisely, speculatively compensated) labor force of community artists, while retaining curatorial control and the majority of the revenue. Some creators earned meaningful income — six figures annually in rare cases. Most earned little. Valve captured the surplus.
The virtual item economy that grew from this system was staggering. Counter-Strike skins, in particular, became a speculative asset class unto themselves, with rare knife skins trading for tens of thousands of dollars on third-party marketplaces. The Steam Community Market, Valve's own trading platform, processed billions in transactions while extracting a fee on every trade. A secondary ecosystem of gambling sites — many of them using Steam's own API to facilitate skin-based wagering — flourished until regulatory pressure and a lawsuit forced Valve to issue cease-and-desist letters in 2016. The gambling sites were shut down (mostly), but the skin economy persisted, a strange hybrid of digital goods market, speculative exchange, and cultural signaling system that generated revenue for Valve with essentially zero marginal cost.
The community is your best programmer. The community is your best QA. The community is your best marketing department.
The Hardware Digressions
Valve's relationship with hardware is the area where the flat organizational structure's weaknesses become most visible. The company has launched — and effectively abandoned — a series of hardware initiatives that, viewed charitably, represent bold experiments in platform expansion and, viewed skeptically, represent the expensive consequences of a company where anyone can start a project but no one is accountable for finishing one.
The Steam Machine, announced with enormous fanfare in 2013, was an attempt to bring Steam into the living room through a standardized PC console running SteamOS, a Linux-based operating system. The concept was strategically sound: Windows was a dependency Valve wanted to escape, the living room was a distribution channel Valve didn't control, and a Linux-based gaming ecosystem would reduce Microsoft's leverage over PC gaming. But execution was scattered. Valve outsourced manufacturing to over a dozen hardware partners, resulting in a bewildering array of devices at wildly different price points, with wildly different specifications, none of which could compete with the PlayStation 4 or Xbox One on price or the simplicity of the console experience. The Steam Controller — an innovative input device with haptic trackpads instead of traditional thumbsticks — was praised by enthusiasts and ignored by the mass market. The Steam Link, a streaming device, worked well but was superseded by software solutions. By 2018, the Steam Machine initiative was quietly dead, the Steam Controller discontinued, the wreckage absorbed into institutional memory.
The Valve Index, released in 2019, was a more focused product — a premium VR headset aimed at enthusiasts rather than the mass market. At $999, it was expensive, technically excellent, and paired with Half-Life: Alyx as its killer app. It sold well enough to remain perpetually backordered through early 2020. But Valve never iterated on it with a successor, content to let the VR market evolve around it while extracting revenue from what amounted to a limited production run.
Then came the Steam Deck, and suddenly the hardware story changed.
Announced in July 2021 and shipped in February 2022, the Steam Deck was a handheld PC running SteamOS — a Linux-based operating system powered by Proton, Valve's compatibility layer that allowed Windows games to run on Linux with startling fidelity. It was, in essence, a Nintendo Switch for the PC gaming ecosystem, and it worked. Not perfectly — battery life was mediocre, the device was large, and not every Steam game ran flawlessly — but well enough to generate genuine enthusiasm and sustained demand. The device was priced aggressively, reportedly at or below cost, starting at $399. Valve wasn't selling hardware for hardware margins. It was selling hardware to deepen Steam lock-in, to prove the viability of SteamOS, and to reduce its existational dependency on Microsoft Windows.
The Steam Deck OLED, an iterative update with a superior screen, improved battery life, and better thermals, followed in November 2023. It was, reviewers agreed, the version the original should have been. By 2024, the Deck had sold an estimated several million units — not iPad numbers, but enough to establish a new product category and, critically, to demonstrate that SteamOS could function as a real gaming operating system outside the confines of Valve's own device.
The Anti-Epic Wars
In December 2018, Epic Games — flush with Fortnite cash, sitting atop a revenue stream that had generated over $5 billion in a single year — launched the Epic Games Store with a direct assault on Steam's business model. The proposition was simple and aggressive: Epic would take only 12% of revenue from game sales, versus Steam's 30%. To seed the platform with users, Epic would give away free games weekly and secure timed exclusives from major publishers, paying guaranteed minimums to developers in exchange for keeping their games off Steam for a year.
The gaming community split. Developers, particularly smaller studios and mid-tier publishers, praised the lower take rate. Gamers — the ones who had spent years building Steam libraries, friends lists, achievement histories — were furious at having their purchases fragmented across a second, inferior client. Epic's store launched without basic features that Steam had offered for years: no user reviews, no shopping cart, no community forums, no cloud saves at launch. The technical gap was a canyon.
And yet the strategic threat was real. Epic had the capital — Fortnite alone was generating billions — and the willingness to sustain losses for years. Tim Sweeney, Epic's founder and CEO, framed the battle in explicitly ideological terms: Steam's 30% cut was extractive, he argued, a relic of the physical retail era that bore no relationship to the marginal cost of digital distribution. He was not entirely wrong. The 30% standard — which Apple, Google, and Steam all converged on independently — was less a reflection of cost structure than of market power.
Valve's response was characteristically oblique. No price war. No public statements. Instead, a series of quiet feature investments: Steam Remote Play Together, enhanced Proton compatibility, the Steam Deck, improved developer tools, the Steam Next Fest showcasing indie games. Valve competed not on commission rates but on ecosystem depth — the accumulated infrastructure of community features, workshop tools, trading cards, achievements, friends lists, and broadcast capabilities that made Steam not just a store but a social platform. By 2024, Epic had spent an estimated $2 billion or more subsidizing free games and exclusives, and Steam's market share had barely moved. The Epic Games Store was still not profitable. The 30% endured.
There's a revolution happening in game stores. The store that earns developers' trust by offering them a fair deal will ultimately win the trust of gamers, too.
The Regulatory Shadow
Valve's dominance has not gone unnoticed by governments. In 2021, the European Commission fined Valve €1.6 million as part of a broader €7.8 million penalty against Valve and five game publishers — Bandai Namco, Capcom, Focus Home Interactive, Koch Media, and ZeniMax — for geo-blocking practices that prevented consumers in certain EU countries from purchasing games at lower regional prices. The fine was, for Valve, an accounting rounding error. But it signaled something larger: regulators were beginning to view digital game distribution through the same lens they applied to app stores.
In the United States, Wolfire Games — the developer of Overgrowth — filed an antitrust lawsuit against Valve in April 2021, alleging that Steam's 30% commission and its most-favored-nation clauses (which Wolfire claimed effectively prevented developers from selling games for less on competing platforms) constituted monopolistic behavior. The case was initially dismissed but partially reinstated on appeal in 2023, with the court finding that Wolfire had plausibly alleged that Steam's dominance allowed it to sustain supracompetitive pricing.
The legal and regulatory pressure remained nascent as of early 2025 — no existential rulings, no structural remedies imposed — but the direction of travel was unmistakable. The same antitrust logic that had been applied to Apple's App Store, Google's Play Store, and Amazon's marketplace was being extended to Steam. Valve's counterargument — that developers were free to distribute games on competing platforms, that Steam's 30% reflected genuine value delivered through infrastructure and community features, that the commission structure included substantial volume discounts (dropping to 25% above $10 million in sales and 20% above $50 million) — was defensible but not unassailable. The question was not whether Valve delivered value. It was whether the market structure allowed developers a realistic choice.
The Invisible Empire of Dota and Counter-Strike
Valve's esports operations deserve separate examination because they illuminate the company's broader strategic logic with unusual clarity. Dota 2 and Counter-Strike (now Counter-Strike 2, released in September 2023 as a free upgrade to CS:GO) are two of the most popular competitive games in the world, with combined peak concurrent player counts regularly exceeding two million. They are also, in a precise and revealing sense, infrastructure products — games designed to sustain permanent engagement, drive Steam usage, and generate recurring revenue through cosmetic microtransactions rather than upfront purchases.
The Dota 2 International, Valve's flagship esports tournament, pioneered a model that was both innovative and deeply characteristic of the company. Prize pools were crowdfunded through the Battle Pass — a seasonal content package that players purchased, with 25% of proceeds flowing into the tournament's prize pool. In 2021, The International's prize pool reached $40 million, the largest in esports history. This was not Valve spending money on esports. This was Valve's community funding esports while Valve retained 75% of the Battle Pass revenue. The economics were gorgeous — for Valve. For the competitive ecosystem, the picture was more complicated: outside of The International, Valve invested relatively little in sustaining a year-round professional circuit, leaving third-party tournament organizers to fill the gap with less funding and less stability.
Counter-Strike's economy was, if anything, even more remarkable. The skin market — driven by randomized loot boxes (called cases) that players could purchase keys to open — generated an estimated several billion dollars in cumulative revenue, with Valve taking its cut on every key sale, every Community Market transaction, and every in-game purchase. The September 2023 release of Counter-Strike 2, which ported all existing CS:GO skins into the new engine, was a masterclass in platform continuity: not a single dollar of players' cosmetic investments was stranded. The upgrade was free. The skins transferred seamlessly. The economic moat deepened.
The Privately Held Paradox
Valve Corporation has never reported quarterly earnings, never held an investor call, never filed an S-1, never issued a public share. Gabe Newell, whose personal net worth is estimated at approximately $10 billion, owns the company outright — or very nearly so, with some equity distributed to long-tenured employees. There are no external shareholders to satisfy, no board composition to manage, no activist investors to fend off, no proxy fights to survive.
This structure is Valve's ultimate competitive advantage and its most profound limitation, and the tension between these two realities is the central paradox of the company.
The advantage is obvious. Valve can think in decades. It can absorb years of hardware losses to seed a platform. It can decline to ship Half-Life 3 because no one with authority can compel the decision. It can ignore quarterly revenue targets because there are no quarterly revenue targets. It can maintain a 400-person headcount while generating revenue that would, in a public company, demand thousands of employees servicing investor expectations for growth, diversification, and strategic narrative. The absence of external governance is the absence of friction.
The limitation is subtler. Without external accountability, Valve's strategic direction is functionally one man's intuition. If Newell's instincts are correct — and the track record, from Steam to the Steam Deck to the decision not to engage in a price war with Epic, suggests they overwhelmingly have been — the structure is optimal. If his instincts falter, or if his attention shifts, or if he is simply no longer there, there is no mechanism for course correction. There is no succession plan that has been publicly articulated. There is no institutional process for strategic decision-making that exists independent of Newell's judgment. The flat organization, which distributes tactical autonomy to individual employees, concentrates strategic authority in a single person who happens to own everything.
Valve has been valued, in various secondary-market estimates and press reports, at anywhere from $10 billion to $15 billion — numbers that likely understate reality given estimated revenue levels. But the valuation is academic. There is no liquidity event on the horizon. There are no shares to buy. The company exists in a peculiar state of financial superposition: enormously profitable, completely opaque, answerable to no one.
The Weight of the Library
Consider the position of a PC gamer in 2025. They have a Steam account they created in 2009. It contains 347 games, purchased over sixteen years of impulse buys, seasonal sales, and bundle deals. Their total investment is perhaps $4,000 — not a fortune, but not trivial. Their friends list contains 83 people. Their achievement history spans a decade. Their cloud saves are synchronized across three devices. Their Steam Workshop subscriptions keep their modded games updated automatically. Their Steam Deck connects to the same library seamlessly.
Now consider the proposition from a competing platform: download this other client, create a new account, rebuild your friends list, repurchase the games you want to play, lose your achievement history, manage a second set of cloud saves, maintain a second social graph. The competing platform might offer a lower commission to developers. It might give you free games. It might even, technically, have a superior storefront experience.
The user does not switch. The library is the moat.
This is the deepest truth about Valve's competitive position, and it is the detail on which the entire analysis turns. Steam is not dominant because it is the best store, though it may be. It is not dominant because it offers the most features, though it does. It is dominant because 130 million users have sunk costs — financial, social, emotional — that make switching irrational at the individual level, even if it might be rational at the collective level. The network effect is not merely that friends are on Steam. It is that your entire gaming identity is on Steam. Every purchase, every achievement, every screenshot, every review you wrote, every mod you subscribed to, every hour logged. You are, in a meaningful sense, in Steam. And Valve takes its thirty percent.
In Bellevue, Washington, in a nondescript office that most visitors describe as surprisingly quiet for a company of its influence, roughly 400 people maintain this machine. No quarterly earnings calls. No investor presentations. No roadmap. Just the hum of the platform, the daily transaction flow, and the compounding weight of 130 million libraries that grow heavier — and harder to leave — with every sale.
