How Internet Cafes Became Esports Centers Around the World

Walk into a PC bang in Seoul at two in the morning and the room is not quiet. Screens are lit, headsets are on and someone is carrying a tray of instant noodles between the rows. Walk down a high street in a mid-sized American town at the same hour and the internet cafe that used to sit there is a phone repair shop.

Same business at the start. Very different endings. The interesting question is not which country loved cafes more. It is what each market did once the reason for the cafe disappeared.

Four turns the business took

The internet cafe has changed what it sells at least four times.

In the middle of the 1990s it sold access. Home computers were not common and home internet even less so. Getting online at all usually meant walking to a cafe. The product was the connection.

From the end of the 1990s into the early 2000s the product became latency. StarCraft, Counter-Strike 1.6 and Quake III rewarded a good connection and punished a bad one. The LAN center grew out of that. Sitting in the same room as your opponents was a competitive advantage as much as a social one.

In the middle of the 2000s the Western version of the business fell over. Cheap and capable home computers arrived at the same time as broadband. Together they removed the reason to leave the house.

After roughly 2012 the same thing happened to access-driven cafes almost everywhere else. 3G networks and smartphones put a connection in every pocket. In markets where the cafe was mainly a way to get online, that was the end of the story.

Then the 2020s brought a fourth turn. The cafe came back in places where it had died. It did not come back as a cafe. It came back as a premium esports room selling high-end machines, a designed space, food and drink, events and tournaments. Kotaku called it the return of the LAN cafe. The description fits. The hardware is the excuse, the room is the product.

South Korea, where the culture outlived the format

Korea is the market people reach for when they want to argue the cafe is immortal. The numbers do not support that.

There were about 7,773 PC bangs in 2023, down from roughly 11,871 in 2020. That is a contraction of about a third in five years. It did not pause at any point along the way. The figures come from the Korea Creative Content Agency's annual game industry report, summarised here.

Operators point at the cost of living, at more people owning capable gaming hardware at home and at habits the pandemic reshaped. All of that is true. The blunter reading is that even in the country most identified with this business, the format is shrinking.

So why does everyone still think of Korea as PC bang country? Because the institution outlived the count. It is still where you go after work, where a team meets before a match, where you sit for three hours because the room beats your bedroom. Cultural weight and business health are different things. Korea is the clearest place where they came apart.

China has the only clean number anybody publishes

China is the one market where somebody actually counted the conversion.

In 2010 there were around 144,000 cafes serving about 163 million users. The peak came later, in 2016, at roughly 152,000 licensed venues. By June 2023 roughly 132,000 were left. The figures come from DFC Intelligence and RADII.

Now look at what sits inside that decline. Between 2019 and 2023, 45,000 cafes closed. The net fall from the peak was about 20,000. Those two numbers only reconcile one way: roughly 25,000 venues opened while 45,000 shut. A headline can take the first number and describe a collapse. The register describes churn.

Here is the part worth stopping on. Of those 132,000, about 100,000 are now esports cafes. Roughly three out of four have changed what they sell. The building stayed put. The sign, the hardware, the seating and the reason a customer walks in did not.

Chinese industry reporting for 2025 describes the count growing again. I am deliberately not putting that figure next to the ones above, because it is compiled on a different basis and the two do not subtract cleanly. That caveat is not a footnote. It is the same measurement problem this article ends on.

What makes the Chinese data usable at all is that cafes are a licensed category there. Closures and conversions both hit a register. No other market counts at that level.

Southeast Asia, where the cafe is still the front door

The six largest gaming markets in the region, meaning Indonesia, the Philippines, Thailand, Vietnam, Malaysia and Singapore, were worth 5.8 billion dollars in 2023, with a forecast of 7.2 billion for 2027, according to Niko Partners.

There is no single regional story. The same map holds countries where the cafe is infrastructure and countries where mobile pushed it into second place.

Vietnam is the outlier. Nvidia counted more than 45,000 gaming internet cafes there in 2018, close to the combined total of Thailand, Indonesia and the Philippines. That density is not nostalgia. It is infrastructure. In that country the cafe is not a side effect of the gaming market. It is a load-bearing column.

Indonesia went the other way. The market is mobile first and, by regional payment-industry estimates, produces close to 30 percent of the region's revenue. The warnet, short for warung internet, still exists but plays a smaller part than it used to.

The Philippines sits between the two. Mobile Legends kept cafes busy and turned them into meeting points for local esports teams. In 2020 the TNC Cebu headquarters opened with more than 400 stations, which is not a cafe in any sense the 1990s would recognize. The cafe there became the physical address of team culture.

The West lost the business then rebuilt it

LAN centers were popular across the United States and Western Europe through the late 1990s and the 2000s. Then cheap home hardware and broadband pulled the floor out from under them in the middle of the 2000s. Most did not adapt. They closed. Most of them closed because what they sold disappeared, not because they were badly run.

What came back is a different business using the same furniture. High-end machines, a room designed to be photographed, food and drink, leagues, watch parties and birthday bookings. The margin is not in the hourly rate anymore. It is in everything sold around the seat.

That difference deserves to be stated plainly, because it separates the two halves of the world. In the West the access cafe died and something new took its place. In East and Southeast Asia the cafe never died. It changed clothes.

So what is the internet cafe to esports ratio in each country?

Nobody publishes it. That is the honest answer and it is worth saying clearly, because you will find pages quoting confident percentages for a dozen countries at once.

China is the exception. Roughly 100,000 out of 132,000 is the only national figure for converted cafes that you can actually cite. Everywhere else you are working with proxies. Korea gives you total hours and venue counts but no split by format. Southeast Asia gives you market revenue and cafe counts but keeps no register separating a gaming cafe from an esports venue. The West does not even have a reliable base number, because the businesses that closed were never licensed as their own category the way they were in China.

Underneath the data problem sits a definition problem. Countries do not classify the cafe the same way. In one place the license is written around internet access, in another it falls under gaming venue rules. The same business shows up in two different categories in two different countries. A cafe in Ho Chi Minh City packed with machines running competitive games is functionally an esports venue and is probably registered as an internet shop. A converted wangba in Chengdu might be counted as esports because it built a stage. The label is doing work the building is not.

So treat any global ratio table with suspicion. The pattern is real. The percentages, outside China, are decoration.

What an operator should take from this

One thing travels across every one of these markets. Selling access is finished everywhere, including the places where cafes are thriving. Korea, Vietnam and China did not keep their cafes by being the only way online. They kept them by being a better room than home. What brings a customer in now is not the connection. It is the room.

Two smaller lessons sit under it. Closure and conversion look identical from outside. China lost 45,000 venues while three quarters of the survivors were turning into something else. A count of businesses tells you almost nothing about the health of the format.

And the surviving model is hybrid everywhere. Hardware plus food plus events, in Chengdu, in Cebu and in a converted warehouse outside a mid-sized American city. The operators who held on to nothing but the hourly rate are the ones who are gone.

If you are opening now, the useful reading of this history is that you are not entering a declining trade. You are entering the fourth version of it. That version has more in common with running a small venue than with the shop your customers' parents remember. This history does not tell you which country you are in. It tells you which turn you are standing on.

We wrote about that shift in you are not an internet cafe, you are a gaming center. The practical side of starting one is covered in how to open an internet cafe.

The HandyCafe feature set was built for the fourth version of this business rather than the first. Whichever software you land on, choose it against the business you are actually running, not the one written on the sign.