Here at FT Alphaville we love exploring \u201cblack hole\u201d companies: those ultra-private businesses that must largely be observed by the influence they exert on others.<\/p>\n
But even within that category, there aren\u2019t many places quite like Valve Corporation<\/a>, with its bizarre corporate structure, revenue-per-head figures that would make Silicon Valley or Wall Street giants hot and bothered, a product that seems to basically print money, and a seldom-seen leader deified in some corners of the internet \u2014 all without ever taking outside investment.<\/p>\n Few companies can claim a bigger role in video game history than Valve. Founded in 1996 by Gabe Newell and Mike Harrington \u2014 both former Windows developers at Microsoft \u2014 the company\u2019s first title, Half-Life<\/em>, launched in 1998, is widely credited with transforming the first-person shooter genre. Harrington left the company in 2006, leaving Newell (known to some online obsessives as \u201cGaben<\/a>\u201d) as its de facto head.<\/p>\n Valve games have been few in number, but their influence is huge: Half-Life<\/em><\/a> and its sequels for their storytelling; Portal<\/em> for its creativity and humour; the Counter-Strike<\/em><\/a> series for its role in the emergence of esports; Left 4 Dead<\/em><\/a> for its use of artificial intelligence to create dynamic co-operative gameplay. However, its most influential title is arguably Team Fortress 2<\/em>, the game whose cosmetics system brought microtransactions \u2014 where players pay small amounts to unlock in-game content \u2014 into the mainstream. <\/p>\n In doing so, it created a business model that has, for better and for worse, powered the mobile games revolution of the past two decades \u2014 and laid the groundworks for one of the zanier presentation slides<\/a> we\u2019ve seen outside of SoftBank<\/a>:<\/p>\n Valve has even managed to find some success in the notoriously tricky hardware market. Its handheld console \u2014 which looks roughly like a military-grade Nintendo Switch \u2014 has sold \u201cmultiple millions\u201d, the company claimed in late 2023, with market research firm IDC estimating<\/a> it hit 6mn sales by early this year.<\/p>\n But its financial engine, powering almost all its other achievements, is Steam. For a generation of gamers, Steam needs no introduction \u2014 such is its dominance. But for those blissfully unaware, Steam is a storefront, distribution service and social media network for PC gamers that launched in 2003.<\/p>\n And it basically prints money.<\/p>\n Personal computers have<\/strong> been<\/strong> a tricky proposal for game developers. In the early \u201900s, pirating PC games was pretty straightforward. Most titles sold as a physical disc with a unique serial code, but these could easily be faked.<\/p>\n For developers, this was obviously anathema. Digital rights management \u2014 controlling who could use game software \u2014 became an increasingly urgent topic.<\/p>\n In came Valve. After unsuccessfully pitching to companies including Microsoft<\/a>, it developed Steam, which combined game-updating technology, anti-piracy and anti-cheating measures on one platform. It would sell users games and verify that they owned the game \u2014 virtually eliminating the need for discs and serial codes.<\/p>\n Steam landed in the right place at the right time, quickly becoming the default platform for PC gamers to download and manage their games. Safer and more convenient than piracy, it also introduced a social overlay that allowed players to communicate more easily while playing different games.<\/p>\n \u201cThe real value proposition for something like Steam is to have everything in one place,\u201d says Clay Griffin, an analyst at MoffettNathanson. \u201cIt\u2019s a one-stop shop.\u201d<\/p>\n As a result, Steam has established a dominant position in the PC games industry, with analysts assuming it accounts for perhaps 70 per cent of all<\/strong> sales of such games. That equates to a significant chunk of what market research firm Newzoo reckons is a $40bn sector<\/a> (including hardware). The platform has more than 100,000 titles available for download, dwarfing its competitors, and levies a typical 30 per cent commission on sales of games and their (often extensive) add-ons.<\/p>\n And it\u2019s good to be the king. Amid a sustained period of stagnation for consoles<\/a>, spending on PC games has seen healthy increases. In a presentation<\/a> shared earlier this year, Matthew Ball, at investment adviser Epyllion, identified PC as a \u201cbig bright spot\u201d amid a difficult patch for traditional gaming vectors (high res<\/a>):<\/p>\n Steam\u2019s concurrent player figures hit a record high earlier this year following double-digit annual growth, prompting Goodbody analyst Patrick O\u2019Donnell to tell clients that Steam \u201cnow owns the PC distribution channel\u201d \u2014 if it didn\u2019t already \u2014 with \u201csignificant advantages\u201d in engagement against its rivals. Based on Steam users\u2019 language choices, much of the growth comes from China. <\/p>\n Advisory firm Ampere Analysis estimates that Steam had 170mn global active monthly users in May, up from 153mn the year before. Remarkably, average concurrent users are about three times the number who are actually playing games. <\/p>\n Piers Harding-Rolls, Ampere\u2019s games research director, said this \u201chighlights that many users automatically sign into the platform and also that Steam represents more than a PC game storefront \u2014 it\u2019s a gaming community platform in its own right with millions visiting the app to engage with other users\u201d.<\/p>\n If generative AI does \u2014 as some optimists hope \u2014 remove development barriers to create a supply-side revolution in gaming content production, Valve may find itself selling shovels in a gold rush.<\/p>\n Working out how much money Valve already makes from its dominance of PC gaming is tricky. Newell is presumed to be its biggest shareholder, and the company is intensely private. It makes almost no financial disclosures, and its employees seldom give interviews to the press.<\/p>\n Still, there are some clues.<\/p>\n In 2021, developer Wolfire Games sued Valve,<\/strong> alleging it was distorting the PC games market by imposing \u201cplatform most-favoured nation\u201d clauses on sellers \u2014 forcing them to offer their best price on Steam.<\/p>\n Other individual customers launched a lawsuit against Valve over pricing last summer, and a judge ruled last year that their case could be brought together with Wolfire\u2019s in a class-action suit \u2014 opening the door to other developers joining in. Their allegations are broad, but focus on the notion that without Steam\u2019s dominance, average commission rates across the sector would be lower: improving developers\u2019 margins and\/or lowering consumer prices.<\/p>\n Improperly redacted documents from the initial Wolfire case \u2014 later hidden \u2014 indicated that Valve had only 336 staff in 2021, with just 79 of them working on Steam. We\u2019ve remade it using data reported contemporaneously<\/a>:<\/p>\n The public docket also briefly contained another chart, which purported to show Steam\u2019s commission revenues and, uh, healthy <\/em>margins (high res here<\/a>):<\/p>\n We\u2019re all mature, grown-up blog writers\/readers here, so let\u2019s have a discussion about sources and numbers.<\/p>\n This is a screenshot of a document we can no longer access. Pavel Djundik, creator of SteamDB, a Steam data platform, is cited by contemporaneous reports as having spotted the data underpinning the staff headcount chart above. Djundik subsequently tweeted this:<\/p>\n Said Reddit thread (in r\/fuckepic, \u201cThe premiere source for all Epic Games criticism\u201d) can be viewed here<\/a>.<\/p>\n Does that make the chart credible? Naturally, we reached out to Djundik and the Redditor who posted the charts to see if they still had the original documents. Their responses, respectively: \u201cNo\u201d and (thus far) silence.<\/p>\n Let\u2019s assume, for the sake of our collective sanity, that these figures are basically correct. They would suggest that with an effective commission rate probably somewhat beneath the 30 per cent baseline, and adjusted for costs, Steam handled perhaps $10bn of games sales in 2021.<\/p>\n Taken by itself, the chart indicates Steam had an operating margin of about 60 per cent and made $2bn of commission revenues, for about $1.3bn in profit just from Steam commissions in 2021.<\/p>\n If we assume that:<\/p>\n 1) All Valve\u2019s admin staff work on Steam (obviously untrue, and therefore conservative for per-head estimation purposes) It suggests Steam perhaps, maybe, possibly makes about $11.4mn per head in profit, per year, from commissions alone. Which is pretty spicy. For context, Wall Street money machine Jane Street can only boast a profit-per-head of $4.3mn. Even if you assume that the rest of Valve only breaks even, Steam\u2019s cut from sales alone would mean the company as a whole makes nearly $3.9mn of profit per employee (higher than Citadel Securities, FWIW).<\/p>\n One thing that does perhaps support the chart\u2019s credibility is how rapidly Valve\u2019s lawyers moved to get the documents removed from the public docket.<\/p>\n Filed in the court docket<\/strong> for the active class action is this chart, which projects that Steam\u2019s overall revenues will pass $10bn next year (high-res<\/a>):<\/p>\n No source is given in the filing, but the chart is taken from the Global PC Games Market Report 2024<\/a> by Video Game Insights, a market research group that has subsequently been bought by Sensor Tower.<\/p>\n In its methodology<\/a>, VGI explains that those figures were derived from a mix of factors, including the number of reviews, how often certain games appear on user profiles, bestseller lists and \u201cproprietary research\u201d. Logically, we would expect this to include sales of Valve\u2019s own games via Steam.<\/p>\n Like with most market research, it ought to be taken with a bucket of salt, although it\u2019s at least in the same rough ballpark as an estimate by Microsoft<\/a> that Valve as a whole made $6.5bn in revenue in 2021. Again, assuming these figures are even roughly <\/em>true, what can they tell us?<\/p>\n Chiefly, it would indicate that commissions \u2014 despite being extremely lucrative \u2014 aren\u2019t responsible for the majority of Steam\u2019s overall revenues (if both disclosed charts, or Microsoft, are correct, we can compare $2bn in Steam commission income in 2021 to $6.5bn to $7.4bn of overall sales).<\/p>\n It\u2019s tricky to believe. Sure, Valve likely makes chunky direct revenues for its own games, and associated microtransactions for things such as Counter-Strike<\/em> \u201cskins\u201d \u2014 on which margins are likely to be excellent \u2014 but we\u2019d suspect things are more tilted towards commissions. <\/p>\n Or, of course, that VGI\u2019s figures \u2014 or any of the others we\u2019ve seen \u2014 are just wrong. \u00af\\_(\u30c4)_\/\u00af<\/p>\n Hard profit figures remain elusive, but on overall revenues, at least, a court document<\/a> from 2024 \u2014 shared by The GameDiscoverCo Newsletter<\/a> \u2014 offers a teasing look inside the black box. In it, a heavily redacted email chain shows Valve employees discussing the company\u2019s revenues per hour.<\/p>\n Kristian Miller, one of Valve\u2019s data scientists, describes the company in the emails as a revenue \u201coutlier\u201d, sharing two charts that seem to put it above Silicon Valley\u2019s top companies \u2014 possibly well above (high-res one<\/a> and two<\/a>):<\/p>\n In this, as with so many things, we invite readers to believe whatever it is they want to believe. If you work for Valve and want to tell us more, please get in touch. The company did not respond to our requests for comment on its financial performance or any other part of this article. <\/p>\n The prospect of a drawn-out court battle may mean more of the platform\u2019s secrets spill out in the coming years. For now, Steam\u2019s position \u2014 of an ultra-lucrative, efficient presence of a relatively small but healthy sector \u2014 seems secure. Yet as the games industry continues to grow, the temptation for a major tech company to take a serious swipe at it will only grow.<\/p>\n \u201cAnybody who wants to put Valve out of business could do so, but nobody cares,\u201d argues Michael Pachter, a gaming industry analyst at Wedbush Securities.<\/p>\n That said, it\u2019s not like nobody has tried. Steam\u2019s biggest rival is probably the Epic Games Store, which has been powered largely by the success of battle royale sensation Fortnite<\/em>. Yet despite this huge draw, Epic \u201chas failed to really impact Steam in any meaningful way\u201d, according to Harding-Rolls:<\/p>\n This strategy has involved offering a better share of sales to developers and publishers and giving away lots of free games. This has prompted Steam to change its own revenue share policies. However, Steam continues to grow and is as dominant in terms of premium PC game sales as it has ever been. I think this illustrates that displacing Steam would be very challenging.<\/p>\n<\/blockquote>\n There are also several smaller distributors, such as Good Old Games \u2014 which made a sliver of profit on sales equivalent to about $55mn last year, and is owned by Poland\u2019s CD Projekt \u2014 and indie platform Itch. Microsoft (which, in a different timeline, could surely have dominated this area) runs a somewhat unambitious store, while gaming heavyweight EA shut down Origin, its own launcher, earlier this year. <\/p>\n Visibility on market share across the sector is poor, but none, currently, seem to be approaching the reach and revenues of Steam. <\/p>\n Valve is a very weird company, <\/strong>and cheerfully admits it. Its 56-page Valve Handbook for New Employees<\/em><\/a> lays out its unusual structure, and right at the start concedes that \u201cit can take some time getting used to\u201d.<\/p>\n And, pretty quickly, it\u2019s made clear that Valve is indeed not like the other businesses:<\/p>\n Hierarchy is great for maintaining predictability and repeatability. It simplifies planning and makes it easier to control a large group of people from the top down, which is why military organizations rely on it so heavily.<\/p>\n But when you\u2019re an entertainment company that\u2019s spent the last decade going out of its way to recruit the most intelligent, innovative, talented people on Earth, telling them to sit at a desk and do what they\u2019re told obliterates 99 percent of their value. We want innovators, and that means maintaining an environment where they\u2019ll flourish. <\/p>\n That\u2019s why Valve is flat. It\u2019s our shorthand way of saying that we don\u2019t have any management, and nobody \u201creports to\u201d anybody else. We do have a founder\/president, but even he isn\u2019t your manager. This company is yours to steer \u2014 toward opportunities and away from risks. You have the power to green-light projects. You have the power to ship products.<\/p>\n<\/blockquote>\n The message is that Valve employees basically self-organise in what the company describes as a \u201cflatland\u201d, with even Newell (the \u201cfounder\/president\u201d referred to above) freed from hierarchical duties (the reality, according to one ex-insider<\/a>, may be slightly less radical).<\/p>\n Flatness involves some interesting decisions. On a practical level, all desks at Valve have wheels, with staff physically shifting themselves around the office to be near the people they\u2019re working with.<\/p>\n The handbook even includes a little illustration of how this should work:<\/p>\n Does this system produce results? On one level, yes: Valve seemingly makes loadsamoney. On another very visible level, absolutely not: Valve releases major new games at a glacial pace. The much-anticipated third title in the Half-Life<\/em> series became the video game equivalent of Samuel Beckett\u2019s Godot, until it was quietly cancelled<\/a>.<\/p>\n \u201cI\u2019ve not met anybody who had a career there,\u201d says Wedbush\u2019s Pachter. \u201cThey all leave after, you know, 10 years. They last long enough to make money. But the developers I\u2019ve known there have left for greener pastures because the company doesn\u2019t publish games.\u201d<\/p>\n Another oddity is the company\u2019s use of \u201cstack ranking\u201d. Once a year, staff are asked to rank colleagues within their division based on skill level, productivity, group contribution and product contribution. These rankings are aggregated, and then used to determine compensation. As the handbook says:<\/p>\n By choosing these categories and basing the stack ranking on them, the company is explicitly stating, \u2018This is what is valuable.\u2019<\/p>\n<\/blockquote>\n This is potentially problematic, and not just from a financial planning perspective. <\/p>\n Whatever a company\u2019s culture,<\/strong> it is bound to work for some people better than others. An investigation<\/a> by YouTube channel People Make Games offers some interesting anecdata from people for whom it didn\u2019t work, and explores the limitations of stack ranking (which, it suggests, rewards short-term, high-visibility work over less-glamorous, longer-term projects):<\/p>\n![\"\"](\"https:\/\/www.ft.com\/__origami\/service\/image\/v2\/images\/raw\/https%3A%2F%2Fd1e00ek4ebabms.cloudfront.net%2Fproduction%2F7e70dc4f-6be1-4a3d-8472-523d15bc41bb.png?source=next-article&fit=scale-down&quality=highest&width=700&dpr=1\")
<\/picture><\/figure>\n![\"\"](\"https:\/\/www.ft.com\/__origami\/service\/image\/v2\/images\/raw\/https%3A%2F%2Fd1e00ek4ebabms.cloudfront.net%2Fproduction%2F79595385-5ba1-4439-8bd9-63d38d5cff61.png?source=next-article&fit=scale-down&quality=highest&width=700&dpr=1\")
<\/picture><\/figure>\n![\"\"](\"https:\/\/www.ft.com\/__origami\/service\/image\/v2\/images\/raw\/https%3A%2F%2Fd1e00ek4ebabms.cloudfront.net%2Fproduction%2F0a1ea86c-6929-4c0a-be50-52fbb3db8cc7.png?source=next-article&fit=scale-down&quality=highest&width=700&dpr=1\")
<\/picture><\/figure>\n<\/picture><\/a><\/figure>\n<\/div>\n
![\"\"](\"https:\/\/www.ft.com\/__origami\/service\/image\/v2\/images\/raw\/https%3A%2F%2Fd1e00ek4ebabms.cloudfront.net%2Fproduction%2Fb19c24a3-b34a-4f90-a3d3-e20dac65428b.png?source=next-article&fit=scale-down&quality=highest&width=700&dpr=1\")
<\/picture><\/figure>\n![\"\"](\"https:\/\/www.ft.com\/__origami\/service\/image\/v2\/images\/raw\/https%3A%2F%2Fd1e00ek4ebabms.cloudfront.net%2Fproduction%2F82d0e690-449a-400c-a731-6115fbec70de.png?source=next-article&fit=scale-down&quality=highest&width=700&dpr=1\")
<\/picture><\/figure>\n
2) The numbers we\u2019ve cited are roughly correct (not unfathomable)<\/p>\n![\"\"](\"https:\/\/www.ft.com\/__origami\/service\/image\/v2\/images\/raw\/https%3A%2F%2Fd1e00ek4ebabms.cloudfront.net%2Fproduction%2Fdb113efa-e25e-42e9-bf4c-235e808213e3.png?source=next-article&fit=scale-down&quality=highest&width=700&dpr=1\")
<\/picture><\/figure>\n![\"\"](\"https:\/\/www.ft.com\/__origami\/service\/image\/v2\/images\/raw\/https%3A%2F%2Fd1e00ek4ebabms.cloudfront.net%2Fproduction%2F7f79775e-3af9-453f-b927-da0466995068.png?source=next-article&fit=scale-down&quality=highest&width=700&dpr=1\")
<\/picture>![\"\"](\"https:\/\/www.ft.com\/__origami\/service\/image\/v2\/images\/raw\/https%3A%2F%2Fd1e00ek4ebabms.cloudfront.net%2Fproduction%2F1d29a9d2-5388-4133-bf4b-b6a27e209bd3.png?source=next-article&fit=scale-down&quality=highest&width=700&dpr=1\")
<\/picture>![\"\"](\"https:\/\/www.ft.com\/__origami\/service\/image\/v2\/images\/raw\/https%3A%2F%2Fd1e00ek4ebabms.cloudfront.net%2Fproduction%2F3f37855c-4296-4914-ad42-debcae273ea5.png?source=next-article&fit=scale-down&quality=highest&width=700&dpr=1\")
<\/picture><\/figure>\n<\/div>\n<\/div>\n<\/div>\n\n
![\"\"](\"https:\/\/www.ft.com\/__origami\/service\/image\/v2\/images\/raw\/https%3A%2F%2Fd1e00ek4ebabms.cloudfront.net%2Fproduction%2F6c8ced7d-9a77-4db9-baf5-b7cac79d8607.png?source=next-article&fit=scale-down&quality=highest&width=700&dpr=1\")
<\/picture><\/figure>\n\n
![\"\"](\"https:\/\/www.ft.com\/__origami\/service\/image\/v2\/images\/raw\/https%3A%2F%2Fd1e00ek4ebabms.cloudfront.net%2Fproduction%2F8ae49187-7499-4753-967e-18b9af38d175.png?source=next-article&fit=scale-down&quality=highest&width=700&dpr=1\")
<\/picture>\n
![\"\"](\"https:\/\/www.ft.com\/__origami\/service\/image\/v2\/images\/raw\/https%3A%2F%2Fd1e00ek4ebabms.cloudfront.net%2Fproduction%2F979ab04c-7304-43e0-be01-42f615d5939a.png?source=next-article&fit=scale-down&quality=highest&width=700&dpr=1\")
<\/picture><\/figure>\n
![one<\/a> and](\"https:\/\/d1e00ek4ebabms.cloudfront.net\/production\/uploaded-files\/table1-e5909749-184b-4fb4-a320-0c6d5ac7c0b2.png\"){kind=link}