Invest1 distinct publisher3 min readPublished
Korea's four biggest venues lost $364m in the first half, and the rebound since is landing unevenly, with Upbit's $1.04bn day beating Bithumb, Coinone and Digital X combined while it has waived fees only on stablecoins.
The Investor · Invest desk
Compiled by The InvestorSomething wrong?How this is made
Add the three challengers and you get $711.5m over that same 24 hours, which leaves Upbit ahead by $328.5m, or 1.46 times the rest of the field [1]; on a four-venue total of $1.75bn, that reads 59.4% Upbit, 36.1% Bithumb, 3.9% Coinone and 0.6% Digital X [2].
The more useful comparison is against the half nobody wants to relitigate. First-half turnover across Korean exchanges came to $366.58bn, down 54.6% year on year [6], which over 181 days is about $2.03bn a day, so a four-venue day of $1.75bn is running roughly 13.5% below the average day of the period that produced the $364m loss [5]. Upbit's own trace says the same thing from the other end: daily turnover sat between 300bn and 600bn won before 20 August and briefly touched 4.61 trillion won on a Saturday afternoon [15], and using the two won-per-dollar rates implicit in the source's own conversions, 1,374 and 1,436 [6], today's $1.04bn is 2.4 to 2.5 times the top of the old band [8] and only 31 to 32% of that peak [7].
The waivers are where the arithmetic gets thinner. Digital X's 36-fold jump in average hourly volume, 270m won to 9.8bn [8], becomes a 5.6-fold jump once you strip the 87% of the increase that was RLUSD handed to large traders, leaving about 1.51bn won an hour [9]; and the $11m 24-hour print implies only 630m to 658m won an hour, some 6 to 7% of that 9.8bn average [10], so either the windows do not line up or the burst has already drained away. Coinone's tripling to 8.2bn won an hour reverted [9], and its snapshot works out to roughly half that peak [12].
Bithumb is the awkward case for anyone who believes fees are the lever: 68 free days last year plus seven more in February never pushed it past 30% share [10], and it is sitting at 36.1% now while charging on most of its book [3][11], with more than 440 listings against Upbit's 180-plus [13], which is $1.44m of volume per listed token against $5.78m [4]. This is probably wrong, but the flow looks like it follows the token rather than the price, since XRP was the most-traded name on both books [16] and Presto Research's Min Jung calls the retail base return-chasers who buy what is already up [17]. The counter-thesis has real money behind it: Mirae Asset is discussing up to 140bn won for a venue that has contracted its trading revenue away until August 2027 [12][7], which is a bid for a licence and a customer file rather than a fee line, and licences do not decay when volume does. If a quarter from now Digital X is still clearing 1.5bn won an hour with no giveaway attached and Coinone has pushed past 8% share, then price beat mix, and I will have called a working subsidy a bad one.
Ranked by verification strength, evidence, and original report placement.
Upbit, operated by Dunamu, handled about $1.04 billion in spot trading volume over a 24-hour window, more than the combined volume of Bithumb, Coinone and Korbit.
Bithumb recorded around $632.6 million in the same window, Coinone about $67.9 million, and Korbit (now called Digital X) close to $11 million.
Coinone saw its trading volume triple to 8.2 billion won per hour after going fee-free, but trading quickly returned to normal levels.
Before the 20 August rally Upbit's daily turnover had been between 300 billion and 600 billion won, but it briefly reached 4.61 trillion won by Saturday afternoon, almost ten times higher than a week earlier; Bithumb also cleared 2 trillion won in the same period.
Korea's top four crypto exchanges lost nearly 500 billion won ($364 million) combined in the first half, which the source dates to the first half of 2026, blaming falling crypto prices that cut the value of exchange-held assets and low trading activity that cut fee income.
Trading fees are the main revenue source for these exchanges and made up 96.91% of Dunamu's first-half revenue, which dropped by half compared with the prior year.
Distinct publishers with included, body-backed reporting in this cluster.
1 article · August 29, 2026
Follow any of these and your For You feed starts watching them — no settings page required.
invest
Upbit and Bithumb lose half their revenue, and the fee-only model shows its cost base1 distinct publisher
invest
ETH's 31% week beat bitcoin by 7.76 points and moved neither altseason gauge1 distinct publisher
invest
MANTRA froze its chain and left the order book open: RWA's risk is plumbing, not story1 distinct publisher
invest
500m SAND minted on a Base contract that held 14.7m, and Upbit posts a caution1 distinct publisher
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
One outlet, quoting itself
Every figure in this story — the $1.04bn day, the 96.91% fee dependence, the 4.61 trillion won Saturday peak, the Mirae price ceiling — comes from a single Cryptopolitan report, which twice cites its own earlier coverage as if it were a second source. The volume ranking is a CoinMarketCap snapshot with no stated hour. And the arithmetic does not close: two won-dollar conversions imply rates 60 won apart, and the $11m day given to Digital X is about 6-7% of the hourly volume the same piece credits it with.
Zero fees are live; loyalty is not
The price war is real and already switched on: Coinone charges nothing on every listed coin, Digital X has committed to free trading until August 2027, Upbit waived stablecoin fees on 26 July. What the waivers actually bought is the weaker half. Coinone's tripling reverted, Bithumb's 68 free days never carried it past 30% share, and 87% of Digital X's 36-fold surge was a token it gave away. Korean flow did come back with bitcoin above $80,000 — but the four venues' combined day still runs below the average day of the half in which they lost $364m.
A floor, sold as a boom
"Crypto trading stays hot" is doing heavy lifting. Take the outlet's own numbers: the four-venue day totals $1.75bn against a first-half average nearer $2.03bn a day — that is coming off the bottom, not running hot. Digital X's 36-fold headline becomes roughly 5.6-fold once the giveaway stablecoin is removed, a caveat the piece supplies and then does not apply. And Upbit's $1.04bn, presented as dominance, is under a third of the peak day the same article describes.
Volume is the number being bought
Nearly everyone in this story has a reason to want a bigger volume print. Two exchanges have set their price to zero, and one of them seeded a stablecoin to large traders while its parent is reported to be negotiating a purchase price of up to 140bn won — the moment when a flattering turnover chart is worth real money. Upbit's own generosity stops at stablecoins, the corner where fee income was thinnest. The reporting itself is a crypto outlet citing its earlier coverage and signing off with a newsletter pitch.
Trust the direction, not the decimals
The corporate numbers are specific enough to be checkable against filings — operating profit down 79.7% and 83.4%, fees at 96.91% of revenue, a combined 500bn won loss — and fee waivers are the kind of commitment an exchange cannot quietly deny. Working against that: one publisher, an untimestamped third-party volume snapshot, an exchange rate that shifts between paragraphs, and a Digital X figure that contradicts itself. The shape of the Korean market here is credible; individual figures should not be quoted without a second look.