Hyperliquid's $3.8bn open-interest month came entirely from the side that buys back HYPE

Take the month apart by venue and it reads differently from the headline figure. Total open interest added $3.80bn while builder-deployed HIP-3 markets lost about $200m, according to Cryptopolitan citing hyperscreener data ^[9], which means the core order book added $4.00bn ^[1], or about 105 per cent of the net ^[2]. HIP-3 had been the engine, climbing from 18 per cent of platform open interest to more than 34 per cent and crossing $4bn in August ^[8]; it now sits near $3.8bn, roughly 26 per cent of the total ^[8]. That distinction is worth more to a HYPE holder than to a trader, because core markets route about 99 per cent of fees into buybacks while builder deployers keep up to half of what they generate ^[10]. The same notional, parked on a different venue inside the same protocol, does close to twice the work for the token.

The gradual-climb reading is where the arithmetic pushes back. Open interest bottomed near $4.76bn in February and now reads $14.669bn, about 210 per cent over seven months ^[3][1], which compounds to roughly 17.4 per cent a month ^[5]. The last month did about 35 per cent ^[4] from a base near $10.87bn ^[3], call it double the trend rate ^[6]. So the buildup was gradual right up until the month distribution showed up: on August 19 Coinbase said its Base app would use Hyperliquid-powered perpetual futures across more than 290 markets in crypto, commodities and equities ^[5], and a day later Trump said the CFTC was working on bringing the platform onshore ^[6]. Read the Coinbase side as an allocation decision rather than a partnership announcement: Coinbase keeps the front end and rents Hyperliquid's order book rather than building its own perps matching engine for the Base app.

The September 6 unlock released 9.92m HYPE worth roughly $820m, an implied $82.66 a token ^[7][7], and open interest printed a local high the following day ^[7]. That is one unlock absorbed, a single datapoint, and it is the weakest leg of the durable-demand case: an unlock tests holders, while open interest tests leverage, two separate measures.

Concentration is where the price and the value part company. Artemis put the tracked perp DEX field at $19.2bn on September 8 with Hyperliquid at $14.6bn, Aster at $2.5bn, Lighter at $1.1bn and edgeX at $598.6m ^[11], so Hyperliquid is 76 per cent of the category ^[10][12] and about 3.5 times the next three combined ^[9]. Note also that the two datasets in the same story disagree by $69m, or 0.5 per cent ^[11], which is roughly the precision to assume whenever an on-chain chart offers three decimals.

My read, and the counter is in the same breath: this looks like a distribution story that happens to be measured in leverage, since a routed order flow deal explains a doubling of the growth rate better than sentiment does, and the composition shift means the flow lands where fees convert to buybacks. The counter-thesis is that open interest is notional, directionless, and can be built by longs, shorts or both, and a bigger book simply means more to liquidate when price moves hard, as October showed ^[13]. What would settle it is not another green month but a violent one: if the book holds through a cascade at 76 per cent share, the demand is real, and if it halves, the last seven months were the same leverage in a politer shape.