Everyone knows Roblox is huge. But as COVID receded into memory, Roblox shifted from “new” to “familiar,” and gaming experienced its largest contraction in nearly a quarter century, the platform continued to grow. And grow. And grow. As a result, there is now a significant disparity between the perceived size of Roblox and its actual, unprecedented scale.

During the average day, more than 80MM people log onto Roblox. As a historical point of contrast, this means that more people log onto Roblox every 10 or so minutes than used Second Life in a month at its peak. On a monthly basis, Roblox now counts more than 380MM users according to RTrack – 2x as many as PC gaming leader Steam, 3x that of Sony’s PlayStations, 3x the number of unique annual users of the Nintendo Switch in a year, and 5x as many as have bought an Xbox console in the last decade. After accounting for duplication across these platforms, as well as the gap between monthly and annual Switch users, it’s likely Roblox has more monthly users than the entire AAA gaming ecosystem combined. What’s more, NPD/Circana reports that Roblox is typically one of the 3–7 most played games on PlayStation and Xbox (Roblox is not available on Switch or Steam), and SensorTower says that in 2023, Roblox averaged more iOS/Android monthly active users than any other game (including Candy Crush!).

Compared to its most similar competitors—the social virtual world platforms, Minecraft and Fortnite — Roblox has about 5x and 2.25x as many monthly players. For non-gamers, Roblox has about two thirds as many monthly users as Spotify and half as many as Snap (though it probably has a lower share of daily-to-monthly active users) and is roughly as popular as Instagram circa Q4 2015, and Facebook in Q3 2009.

Each month, players spend close to six billion hours using Roblox. This time excludes the viewing of Roblox content on Twitch or YouTube, the largest video platform on earth and which counts non-live gaming content as its second most popular genre, with Roblox one of its five most watched games. Most estimates suggest the average Disney+ account watches no more than 20 hours per month, which would mean about 3.1 billion hours in total monthly watch time—barely half of Roblox’s total.

Not only is Roblox enormously popular, but its growth has been shockingly linear (there is some quarter-to-quarter seasonality, which is common to all media products and particularly significant for Roblox given its over-indexing to children and thus the school calendar). Even at its present scale, it’s difficult to argue that growth won’t continue and that the platform should pass a half billion MAUs and 100 million DAUs. This growth is also resilient. And in contrast to other supposed pandemic darlings such as Zoom, Shopify, and Peloton, Roblox never contracted after the pandemic. Instead, it maintained its overall “pull forward” as well as its accelerated rate of growth.

The composition of Roblox’s growth has also been healthy. The share of Roblox’s total monthly users who use the platform on the average day has grown from 16%–19% pre-pandemic to 22% today. The hours played per DAU and per MAU has also grown, showing that new users are not less engaged users. The share of players over 13 has grown from 40% to 58% (in total there are now 2.5x as many users over 13 as there were users of any age pre-pandemic). Though U.S. and Canadian players are up 250% since the pandemic, their share of the total user base has fallen from 35% to 22% because APAC and ROW users surged 650% and 750%, respectively, taking their share from ~16% and ~21% to 24% and 27%. The global share of players also maps nearly exactly to the global share of playtime, demonstrating that Roblox, like all major social platforms, is a global phenomenon.

Although Roblox has diversified its playerbase away from high-income markets and toward more developing ones, revenues per MAU and DAU are up compared to pre-pandemic levels (spending has come down a bit since the peak of the pandemic, though spending was inflated due to a surge in new users that spent heavily to fill their otherwise empty virtual closets). More impressively, spend per hour is also up since before the pandemic—even though users currently play more per month than at any point during Roblox’s existence, with the sole exception of the peak of the pandemic. In other words, Roblox continues to scale its appeal, usage, and monetization.

Annual spending on Roblox is over $3.8 billion and will pass $4 billion by the end of the year, a quarter of which goes to developers. Given these figures, it is no surprise that Roblox users are hyper-productive. In 2022, Roblox’s users designed 170,000 virtual clothing/accessory items and 15,000 virtual worlds each day (current figures are not known, though average daily users have grown 40% and the ease of creation has improved, as has the quality of the output). Over a hundred user-created worlds have been played over a billion times (and one has over 50 billion plays).

So yes, Roblox is unquestionably “working.” Yet Roblox is also unprofitable. Very unprofitable. What’s more, Roblox’s losses continue to swell because its impressive rate of revenue growth has been outpaced by that of its costs. Over the last four quarters, Roblox’s income from operations was ($1.2B) on revenues of $3.2B, representing a -38% profit margin. During the company’s four pre-pandemic quarters, income was ($66MM) on $508MM, for a -13% profit margin. Put another way, revenues are now 6.2x greater, but losses are 18x greater. This is not the typical path of a company, least of all a tech company that is a platform and at Roblox’s scale. So what gives and what’s the significance of the answer?

Growth, But Not Profits

Obviously, Roblox has a costs problem. Over the last twelve months it has averaged $138 in costs for every $100 in revenue.

Unfortunately, many of these costs are outside of Roblox’s control. To start, an average of 23% of revenues are consumed by various App Store/platform fees (this sum is less than 30% because roughly 20% of sales are direct via browser or PC, where Roblox pays credit card processing fees but not 30% store commissions). Another 26% of revenues are paid out to Roblox’s UGC developers. Though the company could theoretically reduce these payments, doing so would harm both brand and developer investment, both of which drive topline. If anything, Roblox wants to grow revenue share, as it would enable developers to invest even more into the platform (many developers are also understandably annoyed they only get a quarter of the revenue they generate, too). In total, Roblox loses 49% of revenue before it even gets to its own costs (let alone the potential of profits).

Infrastructure and Trust & Safety devours another 28% of revenues on average, down from a peak of 37% but only a point below its six-year average. Unlike App Store fee and developer revenue shares, these costs are not strictly marginal ones. There is a fixed cost per hour for Roblox to operate, and it’s not particularly low, but it does not increase if users increase their spending. As such, this category can see declines on a percentage revenue basis. However, this will not be easy. For example, the clearest way to increase spending is usually to increase playtime, which results in incremental services costs (Roblox has three times the users of PlayStation, but Roblox has only 50% more playtime because its players play half as much per month).

In order to attract more players—especially older players with money to spend and the discretion to spend it—more playtime per player, and more spend per player and play hour, Roblox is also investing in experiences that, on average, are more expensive to operate, as is the case with Generative AI. Roblox is increasingly focused on using GenAI to provide real-time player communications (e.g., transcription and translation), as well as power asset and world creation, and operate AI agents and NPCs – none of which is cheap to create or operate. The company is also investing deeply in more AI tools and human moderators in order to combat harassment, predation, and more. In August 2024, Turkey banned Roblox outright for the “protection of our children”; a month earlier, Bloomberg published a brutal report on the platform’s “pedophile problem.”

So after just three of Roblox’s six cost categories, 77% of revenues are eaten. Combined, the next three are about as costly today—though they have comparatively greater room for improvements and are more under Roblox’s control. Both General & Administrative (13%) and Sales & Marketing (5%) costs are down a few points compared to a few years ago and should continue to decline. Yet even if Roblox cuts these categories down by a third each, that’s only six points of the thirty-eight needed just to break even.

The biggest and most interesting category is R&D. As a platform, R&D is not just the foundation of Roblox’s present-day scale, it’s essential to all future growth - and to that end, the company reinvests much of its current revenues in order to grow future ones (and the company’s history of doing so is precisely why they are so large today). Specifically, improvements in Roblox’s tools and capabilities help attract developers and enable them to produce “better” and “more” experiences, which helps attract more users, retain users as they age up, and encourage more user spend—all of which increases Roblox’s revenue, which in turn funds more R&D and provides more money to developers, who can then invest more in their experiences. As already mentioned, much of Roblox’s R&D investments are in Generative AI. Roblox Founder and CEO David Baszucki has said he believes generative AI will enable users to type or speak entire worlds into existence by 2028, and earlier this year, the company began to demo these tools, too, inclusive of detailed object components and attributes (e.g., when a car is created, it is made up of many parts, rather than just being a “car,” and the car not only moves but has the properties of reflection, the ability to inflict and sustain damage, etc.). Roblox has also said they are working to integrate LLMs into their NPCs and to power real-time translation. By dropping the cost, time, and difficulty involved in creating Roblox experiences while also bolstering their levels of immersion and facilitating more/easier social play, Roblox should be able to increase the platform’s popularity, usage, and revenue.

At the same time, Roblox’s R&D investments are enormous. R&D now takes up 44% of revenues on average, equivalent to $1.5B on a run-rate basis. In comparison, Sony’s PlayStation anticipates spending $2.2B in 2024, with these investments spanning R&D at its twenty-one largely independent studios, virtual reality devices, accessories (including the cloud-based PlayStation Portal), the PlayStation 6 (and, it is rumored, handheld), its network technologies, and so on. In 2024, Unity Technologies will spend “only” $1.1B on R&D, and even the most fantastical (and doubtless egregious) forecasts for Grand Theft Auto VI, anticipated to drop in October 2025, suggest it will cost a billion to develop in total after close to a decade. It is difficult to imagine R&D will continue to grow with sales—let alone in excess of sales, as has been the case over the last few years. For most large technology companies, R&D stabilizes at 6%–12%. Yet even if R&D dropped to 10% (about as much as Spotify and Uber spend), Roblox would still be unprofitable.

Profit Light but Cash Heavy?

Though Roblox isn’t profitable, there are some significant caveats to the situation. Over the last twelve months, operating cash flow—a far more important measure than accounting-defined profits—were $650MM, about 20% of revenue. Roblox has been cash-positive for at least twenty-four quarters. Hard to say that’s not a business that’s “working!”

Part of the cause for this discrepancy is Roblox’s approach to revenue recognition. When a user buys $30 in Robux, the platform’s virtual currency, Roblox recognizes $30 in bookings. An average of $3 of that $30 is spent on a “consumable” (i.e., a single-user or otherwise perishable good), and so Roblox recognizes that $3 as revenue right away. The remaining $27 is spent on “durable” goods such as an avatar. As an avatar can and often will be used over time, Roblox recognizes this revenue over the average lifetime of a Roblox user. This lifetime is currently 27 months, and so the remaining $27 is allocated as $1 for each of the next $27 months, which means that Month One has a total of $4 recognized ($3+$1) and $26 deferred.

Roblox’s accounting practices are generally viewed as conservative. To point, Roblox notes that on average, a user uses all of their newly purchased Robux within only three days. Users do not, as Roblox’s revenue recognition policies seem to suggest, take more than two years to spend their Robux account balance on a miscellany of avatars, outfits, and other items. Roblox also does not allow refunds of Robux or virtual goods, and so the company does not hold a liability to the user (the spend is due to the App Store and developers). Most evidence suggests that the usage of virtual goods, like most cosmetics, is front-loaded. As such, a straightline allocation is probably a misfit versus, say, a double or triple declining balance. Furthermore, Roblox is effectively P&L penalized when it increases the lifetime of a customer—which should be a good thing!—because it extends the revenue recognition period. If Roblox users go from an average of 27 months to 30, for example, then the Month One “loss” above increases from $4.50 to $4.60. Indeed, this has happened since the company went public. In Q1 2021, Roblox users averaged 23 months on the platform, and now they last 27.

Roblox’s accounting methodology does mean that some costs (i.e., App Store fees and developer payments) are also deferred, which partly offsets the consequence of delayed revenue recognition. But all operating costs (i.e., Infrastructure, Trust & Safety) and personnel costs (R&D, S&M, G&A) are incurred right away, and so the net effect of Roblox’s revenue recognition policy is that it shrinks accounting profits as long as total user spend is growing (and the degree of this compression increases as the rate of user spending increases). On average, Roblox’s bookings per quarter (i.e., the money users are spending on Roblox that quarter) exceed the revenues recognized by Roblox by about 23%. This isn’t enough to overcome Roblox’s 138% in costs per dollar, but as a percentage of bookings, costs shrink down to 114%.

So how does Roblox generate so much cash? Over the last twelve months, Roblox’s total compensation to its employees was $1.8B. However, 53% of this compensation was in the form of stock, a non-cash expense. As such, Roblox is able to “shield” its operating cash flow by a sum equivalent to about 31% of revenues (between 2-4x that of most tech companies).

It would be misleading to ignore the cost of stock-based compensation – especially since companies tend to shift total compensation toward cash and away from stock as they mature, which would decrease profits at the very time a company is supposed to be steadily earning them. If stock-based compensation is normalized to around 10%, then another 21% of revenue would be consumed by cash expenses. This would increase Roblox’s costs from 114% of bookings to 135%—just short of the 138% provided by Roblox’s GAAP methodology.

How Does Roblox Get Profitable?

A detailed picture of Roblox’s P&L paints both a better picture (Roblox is generating lots of operating cash flow, R&D should eventually fall by a quarter of revenue or more) and a worse one (30% of costs are paid in stock, at least half of revenue must be paid out). So how does Roblox become a “great” business?

First, one additional profit caveat is necessary. For users in the United States and Canada, Roblox’s Average Billings per Daily Active User (“ABPDAU,” a Roblox measure) is $37 per quarter, more than four times as high as European ABPDAU, over six times as high as the Asia-Pacific market, and nearly nine times that of all other markets (Rest of Word). This gap is so large that UCAN net ABPDAU (i.e. after 23% App Store fees and 26% developer revenue share) is still a multiple of gross ABPDAU elsewhere.

On a globally blended basis, ABPDAU is $13, and we know that Roblox loses roughly 35% on that spend, meaning costs are about $18. In UCAN, Roblox more than doubles that cost basis.

The more than 50% net income margins on UCAN users is important because not only is Roblox “working” for developers and consumers, it is working as a business . . . for at least 25% of players. Sure, that’s a bit cheap: “Roblox does make money, just not all the time and not enough for the times they don’t!” But Roblox is also unique as a social platform.

Because of Roblox’s high cost to serve per hour, users in lower-income or lower-propensity-to-spend markets typically reduce profitability, rather than just harm overall margins. For Facebook, Twitter, Snap, etc., low monetizing users are still accretive because however little revenue they generate, the marginal costs incurred are near zero. This issue is exacerbated on the revenue side, too. Unlike other social platforms, Roblox’s revenue is nearly all via user spending rather than advertising. As such, Roblox pays 25% of its revenue to Apple and Google (30% of transactions on those platforms) whereas Facebook, Snap, et al pay effectively 0. Note that Facebook, which has structurally lower costs to service users than Roblox and is far more mature, has an operating margin of roughly 40% — if the company had to pay out 25-30% first, it would never have “tech company” profit margins, let alone profit dollars.

So to achieve overall profitability, Roblox needs to either increase its average bookings per DAU from $13 to at least $18 or achieve a level of revenue scale such that all fees other than developer payouts (currently $3.38 per $13) are effectively halved. In practice, both pushes will likely be needed; fortunately, there are a few mechanisms for each.

First is App Store fees. In various markets around the world, Apple’s iOS platform and Google’s Android are being forced to open up their app distribution and monetization models, at least in part. The specifics vary by market, by platform, and sometimes type of transaction. There’s not yet reliable data to suggest the degree to which Roblox might save on commissions (at a minimum, they will substitute a 30% fee with a 4% credit card processing expense, but additional iOS/Android fees might also apply), and the more complicated purchasing flow may require discounts and/or lead to fewer transactions, which would harm revenue. Accordingly, the net opportunity may be a reduction from 25% to 20% in the medium term, with a longer-term possibility of something closer to 10%.

Another option is to increase ABPDAU so that profitable users are more directly profitable and unprofitable users are less directly unprofitable. This would mean more revenue, which reduces the mostly-fixed costs of R&D, G&A, and S&M. In the average month, fewer than 6% of users and 17% of average DAUs purchase Robux. There are many natural reasons to skip buying Robux in a given month—some MAUs are just trying Roblox for the first time, others don’t use the platform enough to routine spenders, a large portion of users are likely too young and thus purchasing depends on parental permissions, there are financial discounts for bulk purchasing (which naturally results in a delay to the next purchase), and so on. But there is clearly upside here. As a point of example, PlayStation, which has 120MM MAUs to Roblox’s 375MM and 50MM DAUs to Roblox’s 80MM, generated over $31 billion in revenue over the last twelve months compared to Roblox’s $3.8 billion in bookings.

These increases can come from further aging up Roblox’s userbase (older users can typically spend more—mainly because they’re no longer reliant on their parents for money), higher-end experiences (PlayStation is evidence of the premium users will pay for “premium” games), growing cultural cachet (which should enable higher prices for Roblox’s virtual goods), as well as more playtime per user (PlayStation has about twice the hours per user per month).

Advertising solves a few goals for Roblox. First, it turns all users into revenue-generating users (no matter how miniscule) while also increasing the revenue for high-spending customers (which helps cover all other costs and/or subsidize the money-losing users). Second, it means that marginal consumption will continuously result in marginal revenues rather than just marginal cost (primarily via Infrastructure and Trust & Safety expenses). Third, neither the Apple App Store nor Google’s Google Play collect a portion of ad revenues. Accordingly, all advertising would necessarily decrease app store fees as a percentage of revenue. For instance, if Roblox currently earns $10 per month from each paying user, they pay $2.40 to app stores, which is 24% of their revenue. However, if they generate an additional $3 from advertising on top of that $10, the $2.40 payment to stores would now represent just 18% of their total revenue. Moreover, as already stated, that additional revenue from non-paying users would go directly to Roblox (i.e. 0% would go to app stores). There are marginal IT&S costs to support advertising on a per-hour basis, as well as R&D to continuously improve Roblox’s ad platform, plus incremental G&A for ad sales, ad ops, and ad service, etc. However, IT&S should be relatively modest (and far cheaper than the other way to increase ABPDAU—more playtime), and the latter cost categories should be mostly fixed costs and thus offer greater operating leverage.

Advertisers have been investing in Roblox for years through the production of their own branded worlds. As with all of Roblox’s worlds, Roblox doesn’t charge for their creation (though certainly, some brands do pay agencies to construct them). Instead, Roblox is “compensated” through the engagement and purchases that world might inspire (to this end, Roblox might even be paying the advertiser through its developer royalty payments). At the same time, Roblox did have ways to generate advertising revenue from these world-builders, as well as from those whose primary business was the creation and operation of Roblox worlds. Specifically, Roblox would sell banner ads inside its launcher/application (think the Netflix homescreen), as well as sponsored search results. However, these ads were limited to the promotion of virtual worlds or items (e.g., Starbucks could promote Starbucks World or virtual Starbucks aprons, but not real-world Starbucks drinks or coupons, etc.). These restrictions, though “good” for users, had pretty severe constraints on advertising revenue as it meant that advertiser’s who lacked Roblox worlds or integrations couldn’t buy any ads (lots of Facebook ad spend is for local businesses, but your local hair stylist can’t build a Roblox world, nor can a local doctor’s office, etc., and even if they could, they probably shouldn’t). And those companies or organizations that did have a world or various items had the business cases from platform advertising effectively limited constrained to the efficacy of those worlds or various items (yes, in theory Starbucks advertising a sellable apron also aided Starbuck’s overall brand awareness, but that’s a pretty inefficient way to advertise online).

Since late 2022, Roblox has begun to prioritize new ad units and technologies to support its burgeoning ad business. In 2024, for example, Roblox launched in-world billboards that, unlike other ad units, could advertise whatever (e.g. an upcoming music, real world sneakers, etc.), and could be programmatically bid upon by advertisers and served across any worlds that chose to support billboard ads. This sort of advertising has existed “unofficially” for years (a developer would just build a billboard and either sell the ad slot directly or, more commonly, hook into an outside ad network to have the unit filled). In 2023, however, Roblox began to block developers from pinging a third-party server for ad units. Developers are still permitted to sell custom ads inside their world, but programmatic insertion, dynamic pricing, and/or targeting all require the use of Roblox’s ad network.

Another new unit are “Immersive Ads,” which are portals that enable a user to travel directly from one Roblox experience into another. In this case, Pepsi or Nike might (programmatically) purchase a “portal” inside hit sports-themed worlds (e.g. a Golden State Warriors world, a FIFA world, etc.) that make it easier to directly acquire these world’s users compared to just throwing up a digital billboard. This format, too, was originally “hacked” by independent Roblox developers whom would cut their own deals with other world developers, but Roblox has since built it natively into their platforms and requires developers to use their system. Developers can still try to bypass Roblox through custom deals, but only Roblox's platform provides advanced analytics, such as tracking how many users entered the portal, how long they stayed in the destination, and what activities they engaged in, and advertisers can only programmatically purchase immersive ads through Roblox.

The last new ad unit lets advertisers sponsor items that users can claim for free. For example, Nike might do a deal through Roblox whereby any users who visit Developer World A on a given weekend would receive a free virtual pair of Air Jordan Sneakers. Again, this could be achieved through a direct deal between Nike and Developer A, but Roblox’s own platform offers programmatic placement, better analytics, a greater number of partnering worlds, related tools, etc.

It’s not clear how big Roblox’s advertising business might become. In November 2023, Baszucki told Ben Thomspon “Yeah, I don’t think we’ve ever said advertising is going to be the meat of our business” – but he also argued “the size of the potential immersive advertising experience is uncharted right now. It’s immersive, there’s so much time spent there, the memories are really profound when I go with a friend to do that. So I do think it’s uncharted and very big.” Baszucki also mentioned “someday on Roblox, when we go to Nike World, we will not just be trying on virtual Nike shoes. We may be buying the physical versions as well from that place because it’s been such a connection for us with that brand.” Regardless of any physical tie-ins, it’s obvious that with over 350 million young, and otherwise hard-to-reach players, Roblox has the potential to generate meaningful revenue even if most of its business is based on user subscription.

Master Builder(man)

Smaller App Store payments, more ARPU, and a bigger ad platform are the linear opportunities for Roblox.

At 380MM monthly users and nearly 6B hours of monthly usage, Roblox is already one of the biggest media properties globally – not just the most played video game. Very few products ever get to Roblox’s size, let alone continue growing. And at this present size, altogether “new” opportunities start to emerge.

In 2023, for example, Roblox launched new features that enabled “basic” communications capabilities between connected players that were not actively using a Roblox word, including messaging and voice calling, as well as avatar-to-avatar "video calling” that uses a phone or tablet’s front-facing camera to power real-time motion capture. This functionality puts Roblox in more direct competition with social networks and platforms such as Snap and WhatsApp, which boast 800MM and 2.8B users each, but underindex to younger Gen Zs and especially Gen Alphas. Moreover, such functionality is not just “bigger” than immersive socializing, it’s also cheaper to operate and easier to advertise through. Roblox Founder/CEO David Baszucki seems to believe instant messaging is just one of many 2D use cases that the company will soon be able to conquer. At the Roblox Developer Conference in 2023, he teased a forthcoming dating product for verified users over the age of 17, corporate recruiting integrations, and educational programs, among others.

Matthew Ball (@ballmatthew)