The International Falls From $40M to $3.4M: Dota 2 Loses Its Community-Funding Engine

In 2026, The International paid out $40 million. In 2026, the pool was $18.9 million. In 2026, roughly $3.4 million. In the most recent seasons, the announced figure has settled in the low millions. On the same timeline, the Esports World Cup 2026 announced a $75 million total prize pool spread across dozens of titles, while the Saudi eLeague 2026 recorded more than 4 million SAR across 37 clubs. Two curves, moving in opposite directions, crossing around 2026. I keep a private spreadsheet of every major prize pool in the industry, updated each time an official announcement lands. It has four columns: date, title, organiser, total. None of them records a feeling. Pulling the chart from 2026 to now, the Dota 2 line is down 91.5% from its peak while the multi-title Gulf-backed line climbs. A story with one line is easy to write. Two lines running against each other is the actual data. Before 2026, The International ran as a two-way machine. Players bought in-game items, a share of that revenue went straight into the prize pool, a large prize pool generated media narrative, and the narrative pulled more item purchases. That loop made TI the largest prize pool in esports history, and turned the pool figure itself into a weekly growth metric the whole industry tracked. The key point sits there. The TI prize pool was never a measure of Dota 2 audience interest. It measured in-game item sales multiplied by a publisher allocation decision. When Valve reworked the Battle Pass and cut the link between item sales and prize money, the machine stopped within a single season. From that season onward, prize money became a publisher-determined reward for achievement rather than an income source funded by the community. I was rejected in 2026 over a model. Seven years later, I get paid to write about it. The lesson repeats here: once an index is severed from the mechanism that produced it, it no longer measures what it used to measure. The 91.5% collapse is the arithmetic of a product decision, not the result of a fanbase voting with disappointment. The material I have contains no viewership data, no match counts, no community engagement figures for Dota 2 across this period. With that gap, I do not draw conclusions about audience interest. Anyone reading numbers professionally has to distinguish an empty column from a column of zeros. The second pole of the picture sits in the Gulf. The Esports World Cup 2026 announced $75 million across dozens of titles. The Saudi eLeague 2026 recorded over 4 million SAR and 37 clubs. The money did not leave esports. It moved into a different structure: a handful of enormous multi-title events plus state-backed domestic leagues. The older structure, many mid-tier events funded by community prize pools, is contracting. Calendar density is the most overlooked metric. Falcons entered 18 tournaments within the EWC 2026 framework. Eighteen events a year for one organisation is a staffing problem, not a performance problem. Each event needs a roster, a coaching staff, a practice schedule, a recovery cycle. An organisation maximising title count maximises operating cost before it maximises revenue. Falcons won The International 2026. Shortly after, the organisation withdrew from Dota 2. The only statement in the entire dataset attributed directly to a named source is Falcons' line about long-term sustainable operations. The organisation retains many other titles. Read it precisely: a team that had just won the biggest event in that title decided it would no longer carry that roster. One match is a story. Fifty matches are the truth. For me, four TI seasons are four data points, not four memories. And the last data point in that sequence records the reigning champion walking away, not because it lost, but because its cost structure no longer matched its revenue structure. The third pole sits in Korea. Dplus KIA won the League of Legends title at the Esports World Cup 2026, then still had to search for a new owner after salary payment delays surfaced. Its predecessor, DAMWON Gaming, won Worlds 2026. One organisation holding two of the biggest titles in the game at two different moments, and still hitting a cash-flow wall. Dplus KIA's League of Legends roster cost roughly 3 billion KRW, close to $2 million, for the playing squad alone. Even a billion-dollar contract starts with a small note about minutes played. Here, that note records $2 million in salary for a roster that had just delivered a title, sitting next to a balance sheet that could not fund itself. The paradox: the stronger the roster, the higher the cost, and if commercial value does not rise in step, a strong roster becomes a liability on the books. A roster worth millions but lacking commercial value drags the organisation down, regardless of results. Any buyer of Dplus KIA is not acquiring a losing team. They are acquiring a winning team attached to an unprofitable cost structure. The league-level response has already arrived. The LCK imposed a salary cap with a luxury tax mechanism. Read as data, this is a redistribution tool at league level: the highest spenders contribute to a shared pool, and the league uses it to keep competitive distance inside a safe band. During the growth phase, player prices rose faster than revenue generation. The cap is the arithmetic consequence of that gap, not an administrative punishment. Four data sets, placed side by side, produce a coherent picture. TI prize pool: $40M (2026), $18.9M (2026), roughly $3.4M (2026), low millions (most recent). Esports World Cup 2026 total: $75M. Saudi eLeague 2026: over 4 million SAR, 37 clubs. Dplus KIA LoL roster cost: roughly 3 billion KRW, close to $2M. These four figures do not describe the same thing. They describe four points on the same capital curve. My way of reading transfer data comes from a model that was rejected. In 2026 I built an xG model from 26 rounds of V-League data and flagged one club as a severe relegation risk. The editors at the time said football is not mathematics. At the end of the season that club was relegated exactly as the model projected. I kept the spreadsheet. Since then my working rule has been simple: a validated model stays, even when the crowd disagrees. Applying that rule here, I do not read 91.5% as proof Dota 2 is dying. I read it as proof a funding channel has been closed. The correlation between a falling prize pool and a game's health is weak, because a third variable sits between them: the publisher's product decision. Dropping the third variable from the model is the fastest route to a wrong conclusion. I do not trust intuition. I trust the version of intuition that has survived seven seasons of verification. Intuition says an event losing 91.5% of its prize pool is in decline. The data says that in the same window, one champion organisation still chose to leave that event, and another champion organisation still had to sell itself. Those two events are the real signals, because they measure the behaviour of decision-makers, not the feelings of viewers. The second contrarian angle: the esports winter framing is built on the wrong frame. Money still exists inside the system; it simply no longer flows evenly through it. It concentrates on three points: major tournaments, commercially viable titles, and organisations with sustainable operating structures. This is a distribution problem, not a supply problem. Confusing the two leads to bad investment decisions on both sides. The third: the LCK salary cap looks like tightening, but it is the only thing keeping the league's shape intact over the next three seasons. If player prices keep outrunning revenue, every transfer window produces another insolvent organisation. The luxury tax turns overspend into a contribution to the rest of the league, shifting systemic risk back onto the party creating it. The industry's biggest implicit assumption has just been falsified by data. The old model assumed winning saves you: win a title, get sponsors, get money, get a future. Dplus KIA won a major title in the biggest title and still needed a new owner. Falcons won The International and still left the game. Those two facts together are enough to remove that assumption from every esports financial model for the next two years. Between the transfer sheet and the pitch, I stand in the middle and measure both sides. The pitch tells you who wins. The transfer sheet tells you who can still pay wages next season. Over the last three seasons those two sides have converged less and less, and that is the largest structural change in professional esports, larger than any gameplay balance change in the same period. The signals I will track over the next two transfer windows are specific. First, how many tier-one Dota 2 organisations renew a full five-man roster. Second, whether the LCK extends its salary cap to other leagues, because if it does not, Korea will lose stars to uncapped leagues. Third, the share of appearance fees in mid-tier organisations' total revenue, since that money depends on invitations from event organisers rather than on competitive results. A verifiable prediction: within two transfer windows, at least one tier-one single-title organisation will shrink or exit entirely, and the stated reason will be portfolio restructuring rather than results. The verification method is simple: place the results table and the staffing table of the same organisation in the same season side by side. If results hold while headcount falls, the cause sits in the accounts. Dota 2 fans will keep watching The International because it is their tournament, and no spreadsheet is needed to confirm that. My job is to record that its prize pool fell 91.5% in four seasons, that the most recent champion left the game, and that new capital is flowing into a different structure. This industry does not lack money. It lacks a distribution mechanism that stops winners from having to sell themselves after they win.