Home › Game Piracy Statistics

Game Piracy Statistics 2026: What the Numbers Actually Count, in 75 Sourced Data Points

Here is the specific error, and almost every article on this subject makes it. When a headline reports 216.3 billion piracy visits, it is reporting visits to websites. Not downloads. Not people. Not sales. Game piracy is never measured; it is inferred, and every inference in the chain runs the same way — toward a bigger number. Four silent conversions sit between that number and the losses attributed to it — visit to download, download to person, person to would-be buyer, would-be buyer to full-price sale — and not one of them has ever been demonstrated for games. Worse: the dataset that headline comes from does not have a games category. MUSO, the tracking firm behind the figure, sorts piracy into TV, publishing, film, software and music. Games sit somewhere inside software: 14.88 billion visits, 6.9% of everything tracked. So the honest ceiling for gaming’s share of measured piracy traffic is 6.9%, the true share is lower and unpublished, and everything built on top of it is inference. Strip the inference away and what remains is not a crime statistic. It is a distribution report — and it reads consistently: piracy falls fastest in the categories where legal access got easier, and it does not fall where access stayed broken.

What does the 216 billion piracy figure actually count?

It counts visits to websites that MUSO has classified as piracy domains — 216.3 billion of them in 2024 — and nothing else. A visit is a session. One person opening a site four times in an evening is four visits. A bot is a visit. A page that never finishes loading is a visit. Nobody counted a game changing hands, because nothing in this dataset counts games at all.

Source: MUSO, 2024 Piracy Trends and Insights, published May 2025 — the most recent full-year breakdown MUSO has released publicly as of August 2026. Tier: commercial tracking panel, self-published by an anti-piracy vendor, not independently audited, and distributed through a gated form, so the underlying report cannot be checked without giving MUSO your contact details. Two internal checks. The visit columns sum to the stated totals, which many published piracy tables fail. The share column is MUSO’s own rounding and sums to 99.9%, not 100 — small, and we are leaving it visible rather than silently forcing it, because a page about measurement precision should show its source’s. Games appear nowhere as a category: MUSO’s five report sectors are film, TV, music, software and publishing, so every platform’s game piracy — console, PC and mobile — falls inside software, which is what makes 6.9% a ceiling on MUSO’s own taxonomy. MUSO also reports the split by delivery method — streaming 50.4%, download 49.6% — and by traffic source: direct 59.7%, search 30.4%. Independent corroboration of the headline figures: TorrentFreak, 10 June 2025.

MUSO does not hide this. Its own blog post on protecting gaming content, live in August 2026, quantifies the problem with the same all-media figure everyone else uses — MUSO tracked 216.3 billion visits to piracy websites in 2024 for all media types — and offers no gaming number, because it does not have one to offer. The company markets the vertical as “Gaming & Software” on its sales pages and reports it as “software” in the annual data. Neither version is ever split.

One more thing worth checking before anyone builds a per-person claim on this. MUSO reports average visits per internet user falling from 68.15 to 64.25. Divide 216.3 billion by 64.25 and the implied denominator is about 3.37 billion internet users. The ITU counted 5.5 billion people online in 2024, 68% of the world. On the ITU denominator the same total works out at 39.3 visits per internet user — not 64.25. The two are not on the same basis, so treat MUSO’s per-user figure as internal to its own panel and never as a global rate (16Best analysis of MUSO and ITU Facts and Figures 2024).

Does one pirated game equal one lost sale?

No credible study supports a one-to-one substitution, and the most thorough attempt ever commissioned by a regulator found no robust displacement effect for games at all. This is the claim the entire loss-estimate industry rests on, so it deserves the paired treatment.

Sources by tier. Government audit: GAO-10-423, Observations on Efforts to Quantify the Economic Effects of Counterfeit and Pirated Goods, 12 April 2010. Commissioned primary research: Ecorys for DG GROW, Estimating displacement rates of copyrighted content in the EU, delivered 2015 on 2014 survey data, published 2017. Peer-reviewed academic: W. M. Volckmann II, Entertainment Computing, October 2024. Unretrievable vendor estimate: Tru Optik 2014, cited in the Richmond Journal of Law & Technology, 9 February 2019 — flagged here specifically because it is the number most often reproduced without provenance.

The Ecorys numbers are worth reading in full, because both sides misquote them. The study surveyed nearly 30,000 internet users across Germany, the UK, Spain, France, Poland and Sweden about 2014 consumption, and reported that 51% of EU adults and 72% of minors had illegally downloaded or streamed some form of creative content that year.

One finding from the same survey deserves to be lifted out, because it dismantles the mental model underneath every loss estimate: 99% of illegal users also used legal channels, and 39% of legal users also used illegal ones. The authors give this as their reason for rejecting the older method of estimating displacement as the share of pirates who buy nothing — that share is close to zero, because pirates and customers are overwhelmingly the same people. A loss calculation that treats a copy taken and a copy bought as coming from two different populations is describing a population that does not exist.

Sign convention matters here and is routinely dropped: negative values indicate displacement, positive values indicate complementarity. The first four rows come from the study’s methodology 2 (instrumental variables), table S.3; the top-films row uses a different method (quasi-panel, following Rob and Waldfogel) and is the only estimate the authors describe as robust. The study is explicit that a null result does not necessarily mean that piracy has no effect but only that the statistical analysis does not prove with sufficient reliability that there is an effect. Data year: 2014. Six countries, none of them low-income, so the results do not transfer to markets where price is a binding constraint. Source: Ecorys, Estimating displacement rates of copyrighted content in the EU, pp. 13–15.

Now the honest complication, because a page that only stacks evidence one way is advocacy. The strongest recent work runs the other direction. William Volckmann II, publishing in Entertainment Computing in October 2024, exploited the near-randomness of when a protected game gets cracked and found that titles cracked in their first week earned roughly 20% less than the uncracked counterfactual, that the loss shrinks week by week, and that games surviving 12 weeks intact showed close to zero effect. Averaged across the sample, protection preserved a mean of 15% and a median of 20% of total revenue.

Those two findings are not in conflict, and reconciling them is the whole point. Volckmann studied 86 premium PC titles carrying a specific DRM system, released between September 2014 and the end of 2022, at launch. Ecorys studied all game consumption by a general population across six countries in a single year. One measures a launch-window effect on the most expensive slice of the market; the other measures a population-level effect across a market where most games are free-to-play or heavily discounted. Both can be true. What neither supports is the multiplication that turns a visit count into a billion-dollar loss.

There is a further problem with the multiplication that nobody applying it seems to notice: it assumes the displaced sale would have been a full-price one. The median Steam release of 2025 grossed $249 in total, against a mean of $358,900, per our indie game statistics — a distribution so skewed that an average catalogue title has almost no revenue to displace. Loss estimates are built from list prices. Actual game revenue is built from a handful of hits and a very long tail of near-zero.

Is game piracy a price problem or an access problem?

Mostly access — and the evidence that price is not the binding constraint comes from the survey most often cited to argue it is. Ecorys asked pirates what they would pay. For games, 55% said they would pay the current market price or more, the second-highest of the four content types tested. Average stated willingness to pay for a game was €8.40 across the EU28 weighting.

Willingness to pay was asked about the respondent’s last illegal transaction, so it reflects the price of that specific game rather than a full-price AAA release; the study cautions that consumers may value other transactions differently. For scale, the reference market-price band the study used for games in Germany was €5–7, described in the text as roughly the price of one month of gaming rather than a boxed title, against a lowest offered band of €1–2. The country figures, the EU28 average and the final column are the study’s (tables S.5 and S.6); the gap column is 16Best analysis. Currency: euros, 2014 survey year, no conversion applied — do not compare these directly to 2026 dollar prices. The study’s own conclusion: for books, music and games price setting alone cannot explain piracy levels, because most pirates of those categories were willing to pay more than the extremely low bands offered. One correction to the source itself. Ecorys’ summary text states that willingness to pay for games was highest in the UK and Germany; its own Table S.5, reproduced above, puts Sweden highest at €10.10 and the UK fourth at €8.30. We have followed the table, not the prose. Source: Ecorys, tables S.4–S.6, pp. 16–17.

Outside Western Europe the arithmetic changes, and this is where the price argument survives. Most coverage tests it badly, by holding the US list price constant — $69.99 for a major release, with $79.99 now appearing at the top end after Microsoft moved first-party titles there on 1 May 2025 and then pulled one back over market conditions — and comparing it to foreign incomes. That overstates the gap, because regional pricing exists. So we measured the real thing instead.

On 3 August 2026 we pulled Steam’s live store price for one undiscounted premium title, Elden Ring, across thirteen country storefronts, converted each at European Central Bank reference rates, and set it against World Bank 2025 GNI per capita divided by twelve.

16Best calculation, and the method constrains the reading. Prices are Steam store list prices for Elden Ring (Bandai Namco / FromSoftware), read from the public Steam store price endpoint on 3 August 2026, undiscounted at the time of checking and excluding local tax. Conversions use US dollar cross-rates derived from European Central Bank euro reference rates of 31 July 2026, when EUR/USD was 1.1485 — giving USD/PLN 3.7558, USD/BRL 5.0583, USD/MXN 17.3715, USD/IDR 18,052.4, USD/INR 95.3848, USD/PHP 61.2686 and USD/ZAR 16.5750. The ECB publishes euro rates, not dollar rates, so the dollar pairs are ours and anyone can rebuild them from the same daily file. Income is World Bank 2025 GNI per capita, Atlas method, current US dollars, divided by twelve — a national average including non-earners, not a median wage, so read the percentages as a cross-market affordability index rather than a household budget. Regional prices are set per title by the publisher inside Valve’s suggested tiers, so one game is an illustration and not a market average. Turkey, Argentina, Egypt, Nigeria and Pakistan show dollar prices because Steam moved 25 territories to regionalised USD tiers on 20 November 2023. Sources: Steam store; ECB; World Bank.

Poland makes the point without needing an argument attached. A Polish buyer pays 249.00 zł, which is $66.30 — 10.5% more in dollars than an American pays — on 28.7% of US income per head. Regional pricing is not a discount schedule indexed to ability to pay. It is a per-publisher commercial decision that sometimes runs backwards. And the income groups line up against MUSO’s piracy ranking in a way that defeats the simple story from the other end: of MUSO’s top fifteen source countries, the United States, Russia, Canada, the UK, France and Germany are all World Bank high-income economies, and none of the world’s low-income economies appears at all.

Here is where the received wisdom breaks, though, and it breaks against the price theory too. If piracy tracked poverty, the visit ranking would be topped by the poorest large markets. It is not. The United States is number one at 26.68 billion visits (12.33%), Russia — reclassified by the World Bank as high income — is third at 15.40 billion (7.12%), and Canada, the UK, France and Germany all sit in the top twelve. Piracy volume tracks the size of a country’s connected population far more than its income: the ITU puts internet penetration at 93% in high-income countries against 27% in low-income ones. Against UN population estimates the United States, at roughly 4.1% of the world’s people, generates about 3× its population share of tracked piracy visits, while India, at roughly 17.7% of world population, generates about 0.46× its share (16Best analysis of MUSO 2024 country data, UN World Population Prospects 2024 and ITU Facts and Figures 2024).

Population share is the crude version, because it counts people who are not online. Divide instead by the number of people who actually have an internet connection and the ranking sharpens rather than softens.

Visit counts are MUSO’s (2024 report, regional table); internet-user counts are DataReportal’s Digital 2025 country reports, measured at January 2025; the world row uses the ITU’s 5.5 billion for 2024. The per-user column is 16Best analysis, dividing a full-year 2024 visit total by an early-2025 user count, so read it as a ratio of magnitudes. Do not compare these to MUSO’s own published 64.25 visits per internet user — as shown above, that figure implies a denominator of about 3.37 billion and is internal to MUSO’s panel; the whole table here is built on the ITU/DataReportal basis so that every row is comparable to every other row. Two further scope limits: this is all-media traffic, not games, and country attribution is by IP address, which VPN and proxy use distorts in both directions.

On that consistent basis the United States generates 82.9 tracked piracy visits per connected person against India’s 21.8 — a factor of 3.8, in favour of the country with roughly 32× India’s income per head. India has 2.5× the online population of the United States and produced 34% fewer tracked piracy visits in absolute terms. Whatever the top of this ranking is measuring, it is not poverty (16Best analysis of MUSO, DataReportal and World Bank data).

Regional pricing is the mechanism that would test this properly, and it has been moving the wrong way. On 20 November 2023 Valve stopped supporting the Argentine peso and Turkish lira on Steam, converting wallets to US dollars and creating two new regionalised USD price tiers covering 25 territories across Latin America and the Middle East and North Africa. Turkey is MUSO’s sixth-largest piracy source and Brazil and Mexico both sit in the top fifteen. No published study has measured what that pricing change did to piracy in those markets — which is itself a finding about how thin this evidence base is.

Does DRM stop game piracy?

It delays it, that delay is worth real money at launch, and as of 26 April 2026 the delay had gone to zero for every previously released title in the leading system’s catalogue. This is the one part of the subject with a clean, quantified answer, and it is more favourable to publishers than the rest of this page.

The first five rows are from W. M. Volckmann II, Revenue effects of Denuvo digital rights management on PC video games, Entertainment Computing (doi 10.1016/j.entcom.2024.100885, published online October 2024, January 2025 issue) — peer-reviewed, and the strongest evidence in this field. Its central limitation is stated by the author: publishers do not release sales data, so revenue is estimated from new Steam review counts and average active player counts, for single-player titles only. That is a good proxy; it is not sales. The mean and median revenue-protected figures are the paper’s own; the article sits behind a paywall, and the sample size, the week-one effect and the 12-week threshold are independently confirmed in contemporaneous trade coverage. The final two rows are community-compiled crack-status counts reported by games press in April 2026 — unaudited tallies, useful for direction and not for precision, and covering previously released titles rather than the whole catalogue in perpetuity. Irdeto, which owns the DRM system, said in the same reporting that updated security versions were in development.

How much pirated gaming has no legal version to displace?

For the pre-2010 United States catalogue, 86.73% of titles cannot be bought at any price — so at most 13.27% of that catalogue is even capable of producing a displaced sale. This is the part of the argument the loss-estimate method cannot survive, and it is arithmetic rather than opinion.

The Video Game History Foundation and the Software Preservation Network hand-checked a random sample of 1,500 titles between 31 January and 15 April 2023 and found 13.27% still commercially available from their rightsholder, ±2.5 points at 95% confidence. Applied to the study’s own population of 27,627 pre-2010 US titles, that implies roughly 23,961 titles with no purchasable version, in a band of 23,270 to 24,652. The full breakdown is in our video game preservation statistics.

16Best calculation applying a sample-derived rate to a population count, so treat the title counts as an order of magnitude with a stated band rather than a census. Scope: US releases only, unique titles, released before 2010, availability measured to 15 April 2023 — and the market has only moved one way since, so 13.27% reads as a ceiling for 2026. This says nothing about the modern catalogue, where availability is far higher. Source: Video Game History Foundation and Software Preservation Network, Survey of the Video Game Reissue Market in the United States, 10 July 2023.

What has actually moved piracy rates?

Legal availability — and the cleanest evidence comes from the categories where availability changed, not from the categories where enforcement did. Between 2023 and 2024, tracked piracy visits fell 13.14 billion overall. Where that decline came from is the most useful number on this page.

MUSO is an anti-piracy vendor. It sells monitoring, takedowns and site-blocking intelligence to rightsholders. Its 2024 report nevertheless closes with this: Piracy persists not because consumers reject legitimacy, but because legitimate options still fail to meet expectations in price, access, or timing. When the firm whose revenue depends on enforcement writes that its own data reads as a demand signal, that is worth more than a hundred think-pieces.

Set the limits on that before leaning on it. One year of cross-category change is a correlation, not a controlled test: the categories differ in more than access, licensed catalogues expanded over the same period as enforcement did, and the causal reading is MUSO’s own rather than a published model. What keeps it worth something is that no one has offered an enforcement story that predicts this particular pattern — a double-digit fall in the two categories with mature all-you-can-eat licensing, a rise in the one category where official releases arrive late, and near-stasis in the bucket holding games.

Gaming’s own version of this argument is fifteen years old and has never been properly rerun. In 2011 Valve’s Gabe Newell said piracy is almost always a service problem and not a pricing problem, describing being told Russia was not worth entering because everything would be copied, and Russia subsequently becoming one of Valve’s largest European markets. That is a founder’s account rather than a controlled study, and it should be read as such. What can be measured is the storefront that account produced: about 117,000 games, roughly 147 million monthly active users, a 42.3 million concurrent record on 22 March 2026 and about $11.1 billion in gross sales in the first half of 2026, per our Steam statistics. The industry’s most effective anti-piracy measure was not litigation. It was a shop that worked everywhere, all the time, in local currency — which is exactly what Valve withdrew from Argentina and Turkey in November 2023.

There is also a natural experiment running the other way, and it is the closest thing this field has to a controlled test, because the variable that changed was availability and almost nothing else. After February 2022 the major publishers and platform holders suspended sales in Russia, and Russian law exempted domestic companies from paying rightsholders in countries designated unfriendly. Nobody changed Russian prices, Russian enforcement or Russian consumers. The legal shop simply closed. A 2023 survey of Russian gamers by the game-development school XYZ, reported by TorrentFreak, found 69% had played at least one pirated game during 2022 and 51% pirated more than they had in 2021; 27% took more than three titles and 20% more than ten, while 31% pirated nothing at all and 93% still bought at least one game legitimately.

Subscriptions are the obvious next test and the evidence is simply not there. No peer-reviewed study measures whether Game Pass or PlayStation Plus availability reduces piracy of the titles they carry. Our game subscription services statistics show these products behaving as retention tools, with Game Pass near 30 million against a 77 million internal target — a catalogue sized by marketing budget, not by coverage. Anyone claiming subscriptions solved game piracy is asserting, not citing.

Enforcement does produce hard numbers, they are just small ones. Nintendo sued the developer of a Switch emulator in 2024 and settled in March that year for $2.4 million, with the tool discontinued; Nintendo alleged the pre-release build of one title had been downloaded over one million times. Set against that title selling over 10 million units in its first three days (Nintendo, 17 May 2023) and 21.93 million lifetime by June 2025, the alleged pre-release downloads equal about 10% of first-weekend sales and 4.6% of lifetime sales — and the game still set a series record (16Best analysis; the download figure is an unverified litigation allegation, not a measurement).

Why do game piracy statistics disagree?

Because seven different things are being called “piracy” and each one produces a different number, sometimes by an order of magnitude. This is the section to cite. Every figure argued about in public is downstream of one of these choices.

• Visits are not downloads, users or titles. MUSO counts sessions on classified domains. A person browsing four pages in one evening can register as multiple visits; a bot registers as a visit; a failed page load registers as a visit. There is no published conversion rate from visits to completed downloads for games, which means every “X million games pirated” headline built on visit data has inserted a number nobody has measured.
• There is no games category at all. Games live inside software (14.88bn visits, 6.9% of tracked traffic in 2024) alongside operating systems, creative suites and business tools. MUSO’s 2023 edition put over 35% of that sector’s traffic on unlicensed mobile application sites, which would pull the ceiling for everything else down to roughly 9.7bn visits, about 4.5% — but that report is now gated, so we flag the tightening rather than lead with it. Anyone quoting a games-specific share of either figure has apportioned it themselves and usually has not said so.
• Country attribution is unreliable by construction. Visits are attributed by observed IP. VPN and proxy use moves the apparent origin of traffic, and it is heaviest in exactly the markets where content is geo-restricted. So country rankings measure where traffic appears to come from. Treat every national figure as directional.
• Torrent-only measurement misses most of the field. MUSO’s own split shows 50.4% streaming against 49.6% download. Any study built on public torrent swarm counts — still a common method in academic work — sees a shrinking minority of activity and none of the private or direct-download channels.
• “Lost sales” requires a substitution rate that nobody has measured. The GAO said this plainly in 2010: the assumed rate at which consumers would substitute can have enormous impacts on the resulting estimates. MUSO’s own film case study demonstrates the sensitivity perfectly — it reports $440 million in opportunity at full conversion, then concedes realistic conversion may be 10–30%. Run their own band and the same case study is worth $44 million to $132 million: the assumption alone moves the answer by up to 10× (16Best analysis of MUSO 2024 figures).
• Commercial value is not revenue. The BSA’s widely cited $46.3 billion for unlicensed PC software (37% unlicensed installation rate, 2017 data, ~23,000 responses across 110+ countries) is explicitly the commercial value of unlicensed installs — what it would have cost to license them all — not money anyone lost. It is quoted as a loss figure constantly. It is a different quantity. The GAO reviewed the BSA’s method directly and reported that it uses assumptions that have raised concerns among experts we interviewed, including the assumption of a one-to-one rate of substitution — the same assumption this whole page is about, named by a government auditor sixteen years ago and still embedded in the figure being quoted today.
• Available titles and unavailable titles get counted together. A copied 2026 release competes with a purchase. A copied 2003 release with no seller competes with nothing. Since 86.73% of the pre-2010 US catalogue has no legal seller, any estimate spanning the historical catalogue is averaging two incompatible situations and calling the result a loss.

Our house basis for this page: MUSO for volume, always described as tracked visits and never converted; Ecorys for population-level displacement, always with its own robustness caveat attached; Volckmann for launch-window displacement on premium PC, always with its proxy stated; VGHF and SPN for availability. Where those sources point in different directions we have said so rather than picking the convenient one.

Who funds the numbers in this field, and which way do they bend?

Almost every piracy statistic in circulation is produced by an organisation whose revenue depends on piracy being large, and the one major exception was suppressed for two years. That sentence is uncomfortable and it is also just the provenance record.

MUSO sells anti-piracy monitoring and site-blocking intelligence; its report ends with a sales contact. The BSA is a software-industry trade body. Denuvo is owned by a security vendor selling the protection its research supports. Tru Optik was an audience-measurement company selling data on unmet demand. None of that makes their numbers wrong — MUSO’s tracking is the best global volume data anyone has, and Volckmann’s peer-reviewed finding is genuinely inconvenient for the anti-DRM position. It does mean the field has a systematic direction of error, and it runs one way.

Against that sits the Ecorys study: commissioned by the European Commission, delivered in May 2015, and not published at the time. It reached the public in September 2017 after a document request by Julia Reda, then a Member of the European Parliament, and the contract value was widely reported in that coverage as around €360,000 — a press figure rather than a published procurement record, so treat it as approximate. Reda’s account, and the reporting that followed it, notes that the study’s film displacement finding had already been cited in the interim while its null results were not. The Commission has never given a full public account of the delay. Whatever the reason, the sequence is on the record: the largest independent displacement study ever commissioned in Europe found nothing usable against games, and it took two years and a document request to reach anyone.

So the honest description of this evidence base is not “studies disagree.” It is: several well-funded interested parties publish large numbers with unstated substitution assumptions; one government audit says those numbers cannot be substantiated; one independent commissioned study found no detectable effect for games and was withheld; and one peer-reviewed academic paper found a real but narrow launch-window effect using a proxy for sales because the industry will not release the sales data that would settle it. Cite the numbers with that on the record.

Key takeaways

• The headline number counts sessions, not games. 216.3 billion tracked visits in 2024, down 5.72%. Games have no category; the software bucket that contains them is 14.88 billion visits and 6.9% of tracked traffic. That 6.9% is a ceiling, not a measurement, and MUSO’s own gaming blog post quotes the all-media total because there is no gaming total to quote (16Best analysis).
• Nobody has demonstrated one-to-one substitution for games, in either direction. Ecorys found no robust displacement across 11,226 game respondents and a point estimate of +24% pointing the other way; the GAO warned in 2010 that the substitution rate drives everything and is assumed rather than measured. The correct statement is that the effect could not be detected, not that it is zero.
• DRM buys about twelve weeks, and in 2026 it stopped buying them. A week-one crack costs roughly 20% of revenue on 86 premium PC titles; 12 weeks of protection costs close to nothing. By 26 April 2026 the uncracked list had gone from around 52 titles to zero in a month. That is the strongest pro-industry finding on this page and it has a short shelf life.
• Price is not the constraint in high-income markets and regional pricing barely dents it outside them. 55% of European game pirates said they would pay market price or more, and the study concluded a price cut would not move games piracy in those six countries. Measured live on 3 August 2026, the same Steam title costs 0.81% of monthly income per head in the United States and 52.93% in Nigeria — incomes across thirteen markets vary 65× and the dollar price varies 2.0×. Poland pays 10.5% more in dollars than the US (16Best analysis).
• Piracy volume tracks connectivity, not poverty. The USA is the largest single source at 12.33% of visits, about 3× its share of world population; India generates about 0.46× its population share. Per internet user on a consistent basis it is starker still: Russia 115.8, Vietnam 93.3, the USA 82.9, Indonesia 57.1, India 21.8, against a world rate of 39.3. Any argument treating piracy as a poor-country phenomenon is contradicted by the ranking it cites (16Best analysis).
• Availability is what actually moves the number. Music and film are 17.7% of 2024 visits and 64.8% of the year’s entire decline — the two categories where licensed all-you-can-eat access became default. Publishing, the only category growing, is 70% manga, which MUSO attributes to fan translations outpacing official releases. Software moved 2.1%. And the reverse test agrees: publishers withdrew from Russia in 2022, and 69% of surveyed Russian gamers played a pirated game that year with 51% pirating more than in 2021.
• For most of the historical catalogue there is no sale to lose. About 23,961 of 27,627 pre-2010 US titles cannot be bought at any price, capping the displaceable share of that catalogue at 13.27%.
• So the spine, landed: game piracy is measured as a crime and behaves as a distribution report. Every stage of the standard measurement — visit to download, download to person, person to buyer, buyer to full price — inserts an assumption, and the two datasets that move most cleanly move with access: down 18.5% and 18.0% where licensed access matured, up 4.3% where official releases arrive late. The most effective anti-piracy product the games industry ever shipped was a storefront that worked everywhere in local currency — and in November 2023 it stopped doing that in two of its highest-piracy markets.

Frequently asked questions

How many games are pirated each year?

Nobody publishes that number, and the figures usually quoted for it do not measure it. MUSO recorded 216.3 billion visits to piracy sites in 2024, down 5.72% from 229.4 billion in 2023, but those are website visits rather than downloads, users or titles, and MUSO does not break games out as a category. Games sit inside the software category, which recorded 14.88 billion visits, 6.9% of all tracked traffic, and fell 2.1% year on year. That 6.9% is therefore a ceiling for gaming's share of tracked piracy traffic, since software also includes operating systems, creative tools and business applications. Any article stating a specific number of games pirated per year has applied a conversion rate that has never been published.

Does game piracy cause lost sales?

The evidence is split by scope, and both halves are real. At population level, the European Commission's commissioned study by Ecorys, using 2014 survey data from nearly 30,000 respondents across six EU countries, found no robust statistical evidence of displacement for games; its point estimate was +24% on 11,226 game respondents, pointing towards complementarity rather than displacement, though the study is explicit that this result is not statistically robust either. At launch-window level, a peer-reviewed 2024 study by William Volckmann II in Entertainment Computing found that a Denuvo-protected PC title cracked in its first week earned about 20% less than its uncracked counterfactual, with the effect falling to near zero if protection held 12 weeks. That study covered 86 premium PC titles released between 2014 and 2022 and estimated revenue from Steam review counts and player numbers because sales data is not public. Both can be true: one measures all game consumption in six European countries, the other measures the first weeks of expensive PC releases.

Did the EU really find that piracy does not harm sales?

Not quite, and the study is misreported in both directions. The Ecorys report delivered to the European Commission in 2015 and published in 2017 concluded that the results do not show robust statistical evidence of displacement of sales by online copyright infringements. It explicitly added that this does not necessarily mean piracy has no effect, only that the analysis could not prove one with sufficient reliability. It did find one robust displacement result: 40% for recent top films, equivalent to about 5% of sales volumes. For games the coefficient was +24%, in the direction of illegal consumption increasing legal consumption, but that estimate is not robust either. The accurate summary is that no effect could be detected for games in any direction, which is not the same as proving there is none.

Does DRM like Denuvo actually work?

It works for a measurable but short window, and that window closed in 2026. The peer-reviewed study of 86 protected PC titles found protection preserved a mean of 15% and a median of 20% of total revenue, that a week-one crack cost around 20% of revenue, and that surviving 12 weeks uncracked left essentially no measurable piracy effect. The same study found no traits that predicted which games would be cracked quickly: not price, popularity or genre. As of 26 April 2026 games press reported that no previously released title in the system's catalogue remained uncracked, down from roughly 52 a month earlier. Those crack-status counts are community-compiled and unaudited. Irdeto, which owns the technology, said in the same reporting that it was already working on updated security versions for the games affected.

Which country pirates the most games?

No dataset answers that for games specifically, because games are not tracked as a category. For piracy overall, MUSO's 2024 data ranks the United States first at 26.68 billion visits, 12.33% of the global total, followed by India at 17.56 billion, Russia at 15.40 billion, Indonesia at 12.11 billion and Vietnam at 7.44 billion. Two cautions. Visits are attributed by observed IP address, so VPN and proxy use misassigns an unknown share of traffic, and heavier VPN use in geo-restricted markets biases the ranking. And the ranking tracks the size of a country's online population more than any national tendency: the ITU puts internet penetration at 93% in high-income countries against 27% in low-income ones, and Russia is now classified by the World Bank as a high-income economy.

Is piracy caused by high game prices?

Not in wealthy markets, on the best available survey evidence. The Ecorys study found 55% of people who had illegally downloaded a game said they would pay the current market price or more for it, with average stated willingness to pay of 8.40 euros across the EU28 weighting, ranging from 6.20 euros in Poland to 10.10 euros in Sweden. The study concluded that for books, music and games, price setting alone cannot explain piracy levels, and that a price cut would not change piracy rates in the six countries surveyed. Outside high-income markets the arithmetic is different, and regional pricing only partly answers it. Checking Steam list prices for one undiscounted premium title across thirteen markets on 3 August 2026, the price ranged from 0.81% of one month of GNI per capita in the United States to 52.93% in Nigeria, a 65-fold gap in burden, while the dollar price of the identical download varied only 2.0-fold, from 33.18 in Indonesia to 66.30 in Poland. Poland in fact pays 10.5% more in dollars than the United States on 28.7% of the income per head. Valve also removed local-currency pricing for Argentina and Turkey on 20 November 2023, moving 25 territories to regionalised US dollar tiers.

Why do game piracy statistics disagree so much?

Seven reasons, and each one moves the answer. Tracked website visits are counted and then reported as downloads, users or titles, which are four different quantities with no published conversion between them. Games have no category in the main dataset and are folded into software. Country attribution relies on observed IP addresses, which VPN and proxy traffic distorts. Torrent-only measurement now sees under half the activity, since MUSO's split is 50.4% streaming to 49.6% download. Lost-sales figures require an assumed substitution rate that has never been measured, and the US Government Accountability Office warned in 2010 that this assumption can have enormous impacts on the resulting estimates. Commercial value figures, such as the BSA's $46.3 billion for unlicensed PC software, measure what licensing everything would have cost, not money lost. And estimates covering the historical catalogue mix titles that can be bought with the 86.73% of pre-2010 US titles that cannot be bought at all.

Sources

These do not carry equal weight, so here is the tiering. Government audit: the GAO report — the only source here whose job was to check other people’s methods. Commissioned independent research: the Ecorys study for the European Commission — large sample, published methodology, stated null results, and a two-year publication delay that is itself part of the record. Peer-reviewed academic: Volckmann in Entertainment Computing — the strongest causal evidence in this field, using a proxy for sales because sales data is withheld by publishers. Primary research with published margins of error: the VGHF and SPN reissue survey. Official statistics: ITU, UN Population Division, World Bank classifications. Vendor tracking panels, self-published and not independently audited: MUSO and the BSA — both produced by organisations with a commercial interest in the size of the problem, and flagged as such wherever quoted. Unretrievable: the Tru Optik $74 billion estimate, whose underlying report can no longer be checked by anyone.

• MUSO — 2024 Piracy Trends and Insights (published May 2025; 216.3bn visits in 2024 against 229.4bn in 2023, category table, 50.4/49.6 streaming-download split, country rankings, visits per internet user, film ROI case study)
• MUSO — Global Piracy by Industry Report 2023 (March 2024; 229.4bn visits in 2023, sector breakdown with software at 15.2bn, and the finding that over 35% of software piracy traffic goes to unlicensed mobile software sites. The report itself is now distributed only through a gated form, so this figure could not be re-verified from a public page in August 2026 and is flagged as such wherever it is used)
• MUSO — Protecting Gaming Content in a Complex Piracy Landscape (accessed 3 August 2026; MUSO’s own gaming article, which quantifies gaming piracy with the all-media figure of 216.3 billion 2024 visits and publishes no gaming-only number — the clearest confirmation available that the vendor does not hold one)
• TorrentFreak — Pirate Site Visits Dip to 216 Billion a Year, But Manga Piracy is Booming (10 June 2025; independent reporting of the MUSO figures)
• DataReportal — Digital 2025 country reports (Kepios / We Are Social / Meltwater, January 2025; 322m internet users in the United States at 93.1% penetration, 806m in India at 55.3%, 133m in Russia at 92.2% and down 711,000 year on year, 212m in Indonesia at 74.6%, 79.8m in Vietnam at 78.8% — the denominator for the visits-per-internet-user table)
• World Bank — GNI per capita, Atlas method (current US$), 2025 (classifications released July 2026; United States $88,810, Poland $25,520, Turkey $16,300, Argentina $14,650, Mexico $13,730, Brazil $10,550, South Africa $6,270, Indonesia $5,120, Philippines $4,850, Egypt $3,260, India $2,760, Pakistan $1,500, Nigeria $1,360)
• European Central Bank — Euro foreign exchange reference rates (rates of 31 July 2026, used for every currency conversion in the price-to-income table)
• Steam — store list prices for Elden Ring (Bandai Namco Entertainment / FromSoftware), read from the public Steam store price endpoint on 3 August 2026 across thirteen country storefronts; undiscounted list prices excluding local tax
• TechSpot — Almost 70% of Russian gamers are pirating in the wake of sanctions (24 July 2023, reporting a survey by the game-development school XYZ first covered by TorrentFreak; 69% played at least one pirated game in 2022, 51% pirated more than in 2021, 27% more than three titles, 20% more than ten, 31% pirated nothing, 93% still bought at least one game legitimately — self-reported, sample size not disclosed, not independently audited)
• Ecorys for the European Commission, DG GROW — Estimating displacement rates of copyrighted content in the EU (Final Report, van der Ende, Poort, Haffner, de Bas, Yagafarova, Rohlfs and van Til, dated May 2015 on 2014 survey data, published 2017, 307 pages; nearly 30,000 respondents across DE, UK, ES, FR, PL, SE; table S.3 displacement coefficients, tables S.4–S.6 willingness to pay, 51% adult and 72% minor piracy rates, 40% top-film displacement)
• US Government Accountability Office — Intellectual Property: Observations on Efforts to Quantify the Economic Effects of Counterfeit and Pirated Goods (GAO-10-423, 12 April 2010, 41 pages; the substitution-rate assumption and its effect on estimates, the finding that three widely cited US government loss estimates cannot be substantiated for want of underlying studies, the non-disclosure of industry association methods, and the specific criticism of the BSA survey’s one-to-one substitution assumption at p. 21)
• W. M. Volckmann II — Revenue effects of Denuvo digital rights management on PC video games (Entertainment Computing, published online October 2024; 86 titles, September 2014 to end 2022, ~20% week-one crack effect, 12-week threshold, 15% mean and 20% median revenue protected, Steam review and player-count proxy)
• KitGuru — Study shows Denuvo is only effective up to 12 weeks after release (summary of the Volckmann study including its sample size and proxy-measurement caveat)
• TheGamer — There Are No More Remaining Uncracked Denuvo Games (26 April 2026; roughly 52 titles uncracked in late March 2026 falling to none, hypervisor bypass technique, countermeasures in development — based on community crack trackers, unaudited)
• Video Game History Foundation & Software Preservation Network — Survey of the Video Game Reissue Market in the United States (Phil Salvador, 10 July 2023; 13.27% of pre-2010 US titles still in release, n = 1,500, ±2.5% at 95% confidence, 27,627-title population)
• World Bank Data Help Desk — World Bank Country and Lending Groups (accessed July 2026; FY2027 income thresholds of $1,175, $4,635 and $14,375 GNI per capita, Atlas method, based on 2025 data; classifications for India, Vietnam, Indonesia, Brazil, Turkey and the Russian Federation)
• International Telecommunication Union — Facts and Figures 2024: Internet use (November 2024; 5.5 billion people online, 68% of world population, 93% penetration in high-income countries against 27% in low-income countries)
• United Nations Population Division — World Population Prospects 2024 (July 2024; world population 8.2 billion, used as the denominator for population-share comparisons)
• BSA — 2018 Global Software Survey: Software Management, Security Imperative, Business Opportunity (May 2018 on 2017 data; 37% unlicensed installation rate, $46.3 billion commercial value of unlicensed software, nearly 23,000 responses across 110+ countries — commercial value, explicitly not lost revenue)
• SteamDB — Steam switched Argentina and Turkey to USD pricing beginning November 20th (October–November 2023; wallet conversion, LATAM-USD and MENA-USD tiers covering 25 territories)
• PC Gamer — Valve is dropping local currency support for Turkey and Argentina (October 2023; exchange-rate volatility rationale, 25-country scope)
• Richmond Journal of Law & Technology — The Unwinnable War on Video Game Piracy (Kirk Kaczmarek, 9 February 2019; source for the Tru Optik $74 billion 2014 estimate as originally reported by CNBC, 14 January 2016)
• Game Developer — Switch emulator reaches $2.4 million settlement with Nintendo (March 2024; Nintendo v. Tropic Haze settlement terms and the allegation of over one million pre-release downloads of a single title)
• Nintendo — The Legend of Zelda: Tears of the Kingdom sells over 10 million worldwide in first three days (17 May 2023; publisher-reported sell-through used as the denominator for the litigation-allegation comparison)
• GamesRadar — Gabe Newell: Piracy is an issue of service, not price (2011; the service-problem quotation and the account of Steam entering the Russian market — a founder’s account, not a study)
• PC Gamer — Gabe Newell on piracy and Steam’s success in Russia (corroborating report of the same remarks)
• TechPowerUp — EA Sits Out $80 AAA Game Price Race for Now (coverage of the 1 May 2025 move of Xbox first-party titles to $79.99, subsequently partially reversed; $69.99 remains the prevailing standard-edition list price for premium releases)
• Gizmodo — The EU Suppressed a 300-Page Study That Found Piracy Doesn’t Harm Sales (September 2017; the publication history of the Ecorys study and the reported contract value — press reporting, not a procurement record)