All figures cover 1–31 July 2026.
HIP-3, live since October 2025, allows anyone who stakes 500,000 HYPE to deploy a perpetual market on Hyperliquid's order book. In practice it has delivered continuous synthetic exposure to real-world assets alongside the venue's native crypto book (Core). Seven builders have listed markets, and Trade[XYZ] accounts for 99.75% of the volume.
HIP-3 volume rose from $85.5 billion in June to $114.75 billion in July. Korean markets — the two SK Hynix contracts, Samsung, and the MSCI South Korea ETF — accounted for 27% of the total and 88% of the growth. Because SK Hynix and Samsung together also represent more than half of South Korea’s main index, stress in these two names was enough to trigger circuit breakers and drag the entire market lower.
Those names sit inside a wider memory trade. Adding Micron, SanDisk, the Roundhill memory ETF and ChangXin Memory takes the category to half of all HIP-3 volume, up from 2% in early April, when commodities were the venue's largest. That is a narrow slice of the AI complex, and concentrated in the listings hardest to reach from outside Asia. SK Hynix alone traded more than the S&P 500, Nvidia, Apple, Microsoft, Meta, Amazon, Alphabet and Tesla markets combined.
On weekdays, when the underlying assets trade on the main markets, HIP-3 captured 53.8% of perpetual volume against Core; on weekends it retained only 13% of its weekday activity, while Core retained 47%.
Funding rates reveal how the market is positioned. In July, longs paid heavily to hold the memory names, while positioning in the S&P 500 stayed balanced and longs were paid 3% annualized.
SK Hynix is the clearest illustration. The company listed American Depositary Receipts on Nasdaq on 10 July. Those receipts have since traded at a roughly 30% premium to the ordinary shares in Seoul. Normally that gap would close through arbitrage — buy the cheaper Seoul shares, convert them into receipts, and sell them in New York — but the Korea Securities Depository capped conversion at 2.5% of outstanding shares and the quota was exhausted on day one.
Hyperliquid lists both lines and lets anyone trade either contract without KRX access or a U.S. brokerage account. Traders who want the premium to narrow short the ADR perpetual, regardless of what they hold in spot. Traders with no route to Korean shares treat the Seoul contract as their only way to own the cheaper line, creating one-directional long pressure. The funding rates reflect exactly that flow, most sharply in the days after the listing: −4% on the ADR, +37% on Seoul as monthly medians. On 14 July, when the gap was widest, 293 addresses held the long-Seoul / short-ADR pair against only 44 on the other side, while 9,341 held just one of the two contracts.
Holding the long carries a cost. A quarter spent long SK Hynix cost about 9% of notional at July’s funding rates, while the same quarter in the S&P 500 earned roughly 1%. Trading itself remains relatively cheap at approximately 2.4 basis points round-trip including the spread.
On a normal minute in July, how many dollars could you buy without moving the price more than 0.10%? Median executable depth (the value of sell orders resting within 0.10% of the midpoint price) across eighteen markets falls into three clear bands:
SK Hynix generated more turnover than both index markets combined, yet typically offered only $300,000 within the 0.10% band. Continuous refilling (roughly $700 million per day) supports high turnover but not large instantaneous size. As a result, the Korean and single-stock contracts that give the venue its unique appeal remain too thin to absorb large orders, while real capacity exists mainly in the S&P 500, a market any institution can already access elsewhere.
This thinness is a property of this specific market. On ten days in July the SK Hynix book held under $10,000 for at least one minute, and on 3 July it held nothing at all; by contrast, the S&P 500 book never fell below $307,000 in any minute of the month. Stress made the distribution wider rather than lower: during the volatility of 27–31 July median depth in SK Hynix actually rose to $377k–$446k and spreads stayed tight at 0.91 basis points through both circuit breakers, while the worst 5% of minutes lost more than half their depth. Tight prices and healthy average size can coexist with minimal size available at the moment an order arrives.
100,669 addresses traded HIP-3 as takers in July. The 105 that each did more than $100 million accounted for 65.2% of volume. Those addresses turned over $128 billion gross and finished the month with only $478 million of net directional exposure, 0.37% of the notional that passed through them. Their average ticket was $2,245; almost 90% of them posted liquidity as well as taking it; and they were active in 22.5 markets on 25 of 31 days. Finally, only 1.55% of their flow used the venue's TWAP tool. These are the mechanics of inventory management rather than position building, which suggests institutional allocators are not yet a meaningful presence.
HIP-3 has established continuous markets for exposures that were previously hard to reach, most notably Korean semiconductors. That is genuinely new, and July demonstrated it under stress.
Those exposures come with clear constraints: limited depth in the single names, elevated holding costs, and a tendency for available size to collapse without warning even as average depth holds. These frictions may help explain why institutional allocators have not yet arrived in size, with current institutional flow consistent with market-making rather than directional positioning.
For a mandate that needs Asian semiconductor exposure and has no clean alternative, the venue is usable in small clips. For anything requiring capacity, it is not there yet. Greater depth in the single-name contracts, not just higher volume, is the variable that would change the picture.
Looking to use Dune for your company?
