IOSG: Hyperliquid Builder Market Depth Review, 90% of Teams are 'Running Alongside'

Author｜Mario Chow @ IOSG

Ten teams have registered their perpetual markets on Hyperliquid, locking approximately $40M of HYPE for the most part. One of these teams accounted for 97.8% of the trading volume, which recently saw a monthly decline of 44%. This article aims to answer what the other nine teams have gained, with all figures sourced from the blockchain rather than announcements.

*Note: Every number in this article is directly sourced from Hyperliquid's public API: perpDexs, metaAndAssetCtxs, daily candleSnapshot covering all 519 registered assets, delegatorSummary, userNonFundingLedgerUpdates, and clearinghouseState{dex}. "30 days" refers to the complete UTC calendar days from August 15, 2026, to September 13, 2026; "the previous 30 days" refers to July 16 to August 14; "7 days" refers to September 7 to 13. HYPE is calculated at $79.73. The routing data in Section 7 is sourced from Flowscan, as the trading volume of builder code cannot be aggregated from the public API.

Summary

• HIP-3 has once again become a minority in Hyperliquid. The markets deployed by builders accounted for 25.8% of perpetual trading volume in 30 days, down from 57.1% last month. This change is mainly due to the denominator: core trading volume has more than doubled, while HIP-3 itself is declining.

• The leader is contracting. Trade[XYZ] had a trading volume of $64.60B in 30 days, a decrease of 44.2%, with the 7-day average dropping from a peak of $5.36B/day at the beginning of August to $2.01B/day. This decline is approximately half attributable to the storage and AI sectors' real market trading drop, while the other half is from this venue itself. It has not outperformed the core order book since August 18.

• Entropy (io) led in a head-to-head market for a full week before retreating. Its share on Nebius has shown a trajectory of 8.7%, 53.1%, and 20.4% over three consecutive weeks, with its own trading volume also declining for three weeks. A reverse signal is seen in open positions: the open interest rose against the trend by 37%, reaching $51.4M.

• Settlement assets remain a lifeline, with a record still at 6 to 6. All venues that settle with non-USDC stablecoins have ceased trading; those still active all use USDC.

• Asset listings cannot be maintained. At the current auction floor price, an asset is approximately $39,900, and buying all of Paragon's online markets would cost around $1.04M, equivalent to two weeks of trading fees for Trade[XYZ].

• No one is competing on price. All stock-like venues operate at deployerFeeScale = 1.0 with Growth Mode, with actual rates of Trade[XYZ] at 0.427 bp and Entropy at 0.400 bp. All venues outside the leader, regardless of survival, have collectively earned only $747,000 in deployment shares throughout their lifetime.

The Ten Deployers Present

▲ Daily trading volume of Trade[XYZ] and all its challengers below. The vertical axis scale of the two images differs by about 100 times.

"Number of trades" is calculated by summing the n field of the daily K-line within the window. The public API does not provide the number of independent traders.

Historically, only ten have registered perpetual DEX, and there has yet to be an eleventh. Four are trading, five have stopped, and one has never opened. Trade[XYZ] accounted for 97.8% of HIP-3's trading volume in 30 days and 97.6% in nearly 7 days.

The situation of the challengers can be summarized in a few sentences. Entropy made $1.03B in 26 days across six online markets, relying on its own oracle rather than an asset list. It is the only venue that has truly led in a market that also quotes the leader. Paragon is the only challenger whose order book resembles an order book, with 26 online markets and a tail distribution that is open. In the case where Trade[XYZ] entered five of its targets at once, it still saw a 49.9% increase that month. Markets by Kinetiq bought 23 targets, with 95% of the volume concentrated in two index perpetuals. HyENA has ended: the market has been delisted, open interest has dropped to zero, and it has earned $33,414 throughout its lifetime.

HIP-3's Share and Why It Is Easily Misinterpreted

▲ The share of HIP-3 in Hyperliquid's perpetual trading volume, calculated daily.

The 7-day average crossed 50% in mid-July, peaked close to 57% at the beginning of August, then fell below 30%, and has not risen since August 20. On August 18, a builder's trading volume surpassed the entire validator set of Hyperliquid. It has not achieved that since.

⚠️

This ratio actually speaks to its denominator. The numerator is a stock order book, while the denominator is a crypto order book, with the fluctuating leg on the crypto side. The 57% was read during a crypto lull, while the 26% was read when the same stock order book encountered market conditions, during which the core perpetual volume rose by 117%. The share in the last seven days even returned to 28.6%, while Trade[XYZ] continues to contract. Before quoting any HIP-3 share, clarify what crypto was doing during that period.

What truly matters is the absolute volume, and the absolute volume is deteriorating. Trade[XYZ] had a trading volume of $64.60B in 30 days, down 44.2% year-on-year, with the 7-day average dropping from $5.36B/day at the beginning of August to $2.01B/day, with its own order book retracting by 62%. The largest market, SK Hynix, dropped to $8.50B. The declines in both legs of the share are real. The following section will explain that most of HIP-3's leg is not a competition issue at all.

The Main Reason for Volume Decline is the Storage Sector Going Quiet, Not Share Loss

Reading the 44% decline as "the leader is losing" is the most straightforward explanation, but the data does not support it. The verification method is simple: if the decline were due to competition, we should see the targets themselves trading as usual, just with Trade[XYZ] capturing a smaller share. What actually happened is that the targets themselves have quieted down.

There has been no sell-off. Using the peak trading volume at the beginning of August as a benchmark, every major market in the order book today has a higher price.

What has truly collapsed is how far these targets can move in a day, and the venue's trading volume has almost followed suit.

The above are all based on business days, as the stock market is closed on weekends while Trade[XYZ] continues to trade. Including weekends would significantly exaggerate this relationship. Looking only at business days, the daily average volatility of the storage sector and the venue's daily trading volume has a correlation coefficient of +0.47 over a sample of 45 days. Gold is a natural control group: it is the only major market with increased intraday volatility this month, and its trading volume has also risen. Silver is an exception that does not fit this pattern.

However, volatility is merely a proxy variable. A more direct test is to compare the actual stock trading volume of the same nine targets, and the answer is: the traditional market only explains about half.

The actual trading volume of the storage and AI sectors indeed dropped by 25.7%, which is real. However, XYZ dropped by 49.7%, nearly double. The additional 24 percentage points are not provided by the industry.

Moreover, the largest gaps are precisely in its core: SanDisk -26.0pp, Micron -25.4pp, Intel -21.5pp, SK Hynix -17.2pp. It has even outperformed the real market on Nvidia (+36.5pp) and Nebius (+18.6pp), but those two order books are quite small.

Competition cannot explain this gap either. Entropy's total trading volume of SanDisk over 30 days was $523M, while XYZ's own SanDisk order book was down by $6.14B, so the challengers could only capture about 8% of that.

The rest seems more like capital rotation. During the same period, Hyperliquid's core perpetual trading volume rose by 117%, while HIP-3 declined, with the total volume of both sides still increasing by 26%. Money has not left Hyperliquid; it has merely rotated from the stock order book back to the crypto order book.

🔎

Scale anchor points. In the same 30 days, these nine targets traded $2,004.7B across their respective exchanges, with XYZ trading $23.5B on them, accounting for 1.2%. The total of XYZ's 104 markets at $64.60B only corresponds to 3.2% of the real trading volume of these nine names. The highest penetration rate is 9.1% for SK Hynix, while the lowest is 0.1% for Broadcom, and this curve itself illustrates what this business is: assets with high penetration rates are those that crypto-native traders cannot reach, while widely available US large-cap stocks have low penetration rates.

Thus, this decline consists of two parts: about half is the sector's beta, and the other half is its own. The calculations in Section 6 only take actual trading volume into account, regardless of where the trading volume comes from, so the fee calculation is unaffected. However, it is important to separate these two halves: the sector's half may return, while the venue's own half may not.

Reshuffling and the Only Predictive Variable

▲ Daily trading volume of each HIP-3 location on a logarithmic scale. The dashed line represents settlements made with non-USDC stablecoins, while the dots indicate the last day they had trades.

As of today, six locations have ceased trading. The variable separating them is not asset selection, team quality, or historical trading volume, but the stablecoin used for settlements.

The mechanism itself is quite simple: traders must first exchange for a specific stablecoin to place their first order, and they are reluctant to do so. Felix is the clearest example. The small fee discounts that initially supported USDH were wiped out as soon as Growth Mode launched, leaving this settlement asset with only friction.

Kinetiq serves as a control experiment. The only operator that survived after shutting down did so by eliminating the USDH location and reopening an identical index product on USDC. Historical trading volume does not predict anything: dreamcash had a volume of $19.51B, more than the entire batch from June combined, yet it still stopped trading. Entropy entered in August with more funds than any previous entrant, and there was no hesitation in choosing USDC.

What Truly Determines Settlement Assets

Reading USDC's dominance as a protocol arrangement is natural, but Hyperliquid's own documentation states the opposite. Under Aligned Quote Assets v2 (enabled on USDC in late August, with Coinbase as the treasury deployer and Circle as the technical deployer), approximately 90% of the adjusted reserve earnings from USDC on Hyperliquid are allocated to the protocol and enter the Assistance Fund. Interest is paid every 30 days, with the first payment due in early October, and no funds have been received yet.

AQAv2 explicitly does not favor HIP-3. The documentation states clearly: there are no preferential trading fees or volume considerations, and other priced assets continue to be supported on HIP-3 perpetuals. The benefits in fee rates belong to AQAv1, which offers lower taker fees, higher maker rebates, and higher volume considerations for collateral assets, while USDC is not included in AQAv1 and cannot structurally enter because that tier requires stablecoins to be exclusive to Hyperliquid. The real privileges offered by AQAv2 point to event contracts and validator-operated perpetuities, and upgrades are needed for that, which is not the market measured in this article.

Thus, the matter of settlement assets is determined by liquidity and a one-time corporate action, not by fee design. USDH ceased operations on July 17, 2026, with holders redeeming at a 1:1 ratio for USDC, while Coinbase acquired its brand assets and became the treasury deployer for USDC. Today, USDC accounts for 98.3% of the stablecoin supply on Hyperliquid, USDT 1.2%, and the remnants of feUSD, USDe, and USDH are each about one-thousandth. A location settling with other assets has not suffered in terms of fees; it is requiring its traders to leave the only deep pool available.

🔎

Estimating the scale of AQAv2 is challenging, as the official figures have never been disclosed. USDC on Hyperliquid is valued at $6.77bn, with SOFR around 3.6%. Assuming a 90% revenue share, this line roughly points to an annual figure of around $200M. Third-party estimates based on a $5bn base fall between $135M and $160M. What remains uncertain is the cost adjustments in the AQA interest rates, which are reported by a validator oracle whose level is not public, so every number here is an estimate rather than a measurement.

HyENA has added a second mechanism. Because it is tied to crypto assets, it has been blocked from Growth Mode, quoting about 5 basis points on the same targets, while its core market quotes about 3 basis points, resulting in poorer trade quality. It spent about $0.88M on asset positions, earning a lifetime total of $33,414.

🔎

A shutdown does not equal an exit. HyENA has delisted all 25 markets, with open positions reduced to zero, but its staking remains at 508,915 HYPE, approximately $40.6M, and twelve days have passed without any withdrawals initiated. Felix and dreamcash have both reclaimed their full stakes, now reading zero, while Ventuals has only 7,967 left. A location that has delisted all markets but retains $40M staked on-chain is either slow to liquidate or is occupying this deployment position for something else.

The demise of Ventuals deserves a separate discussion, as the next generation of products is designed around this issue. Insufficient liquidity is merely a symptom; the mechanism lies in the funding rates: pre-IPO perpetuities lacked a convergence anchor, with funding rates once soaring to an annualized rate of about 8,700%. Regardless of whether the marked price was correct, longs would be liquidated. Entropy capped the annualized funding rate around 10% and settled to its marked price's TWAP, rather than chasing an external price. Its contract design can be read as a repair checklist for the specific causes of Ventuals' demise. When looking at any pre-IPO market, first check the funding rates and settlement design, then look at the asset list.

The Composition of Trade[XYZ] Market and Why It Doesn’t List OpenAI

▲ The markets with the highest trading volume over 30 days in Trade[XYZ].

The top ten markets account for 66.7% of the market share, with the tail beyond the top six alone accounting for $32.4B. Nvidia occupies 3.5%. Combined, Apple, Tesla, Alphabet, and Microsoft account for 3.3%, only a quarter of SK Hynix. The standard narrative around tokenized US stocks often highlights these American giants, but they are not the business here.

Trade[XYZ] actually operates a 24/7 venue focused on storage and AI capital expenditure trading, along with crude oil, metals, and index products: Korean and Japanese semiconductors, a synthetic DRAM index, SpaceX, its own XYZ100 basket, and a licensed S&P 500. Its territory is comprised of assets that crypto-native traders cannot find elsewhere at 3 AM. This is also the territory that Entropy has chosen to attack, with entry points being SanDisk and Nebius, not Apple.

Why Not List OpenAI

The most intuitive answer, that it avoids unlisted companies, is incorrect. Pre-IPO is actually one of its better-performing businesses. SpaceX alone had $2.80B in volume over 30 days, accounting for 4.3% of the market share, ranking ninth. Following that are Yushu at $511M, Changxin Storage at $317M, Zhipu at $156M, MiniMax at $92M, and SHEIN at $27M, with Yangtze Memory Technologies already registered and pending.

These names share a common trait: they all have observable secondary trading prices and known equity numbers. SpaceX regularly conducts tender offers, providing clear per-share prices; these Chinese companies have active pre-IPO gray markets in mainland China, and equity numbers can be obtained from business records and financing rounds. Thus, this venue can quote per-share prices like it does for other assets.

OpenAI and Anthropic have neither. Their secondary trades are wrapped in SPVs, trading a claim on a fund share, discussing an overall negotiated valuation rather than a specific per-share price. If forced to quote per share, it would be equivalent to creating a denominator out of thin air. The bottleneck lies here; it is a matter of quoting conventions, not willingness. Entropy's solution is simply not to quote per share, but to quote the company itself: 1 contract = $1bn market cap, which, based on the current midpoint price, values Anthropic at nearly $2.17tn and OpenAI at about $1.53tn.

However, this does not create a moat. If a leader wants to add a market cap-priced asset, they can do so for about $39,900 at any time. Moreover, its own roadmap points elsewhere: it has 16 registered but unused assets queued up, including uranium, aluminum, the US dollar index, VIX, corn, wheat, TTF, Korean won, Indian Nifty, Brazilian Ibovespa, Ibiden, and KSTR, along with Yangtze Memory Technologies and H100. This is macro and bulk, not cutting-edge AI laboratories.

One structural detail worth remembering is that Trade[XYZ] has not set up an oracleUpdater, meaning it does not push marked prices using its deployment key; whereas Entropy and Felix both point to the same third-party updater 0x94757f8d.... Entropy has publicly stated that RedStone is the price data source for its Anthropic market, which explains why these two unrelated venues share the same update address, although the blockchain does not label this address. Self-built oracles are fine for assets with reference prices, but when marked prices must be "constructed," the nature changes, and constructing marked prices is precisely the business that Entropy has chosen.

Direct Confrontation and the Week Entropy Led

▲ Weekly trading volume of Nebius and Entropy's share within it.

Currently, nine assets are simultaneously active on two HIP-3 venues. Each venue no longer occupies a separate, non-overlapping market; any name worth listing twice has become the norm.

The Share Taken Away is from Market Share, Not Market Depth

Trade[XYZ] had registered five core assets of Paragon but did not activate them until August 18, when all were opened in one day, and today it leads in all five. Filling this gap cost about three days' worth of transaction fees, completed in one afternoon. Whether those assets were a deterrent or were originally queued for launch cannot be determined from on-chain data.

What can be determined is the result. Four weeks later, Paragon still holds 20% to 25% of four of those five assets, and the overall trading volume increased by 49.9% that month. The market depth of Trade[XYZ] is about 11 times that of Paragon, which only accounts for 8.2% of this pair, yet Paragon's own trading volume nearly doubled during the same period. The leader did not take volume away from the challenger; it expanded the market around the challenger. Entering a market and owning a market are two different things.

Entropy led on Nebius for a week, then handed it back.

For one week, Entropy's trading volume on Nebius indeed surpassed that of Trade[XYZ], something that the HIP-3 challenger had never achieved. In the following week, the leader's Nebius market depth increased by 61%, while Entropy dropped by 63%, bringing the challenger back to one-fifth of this pair. SanDisk tells the same story, just quieter: Entropy's share has roughly hovered around 13% in recent weeks, averaging 8.5% over the entire 30 days.

So this lead is real, but it lasted only a week. The honest interpretation is that Entropy proved it could penetrate a market where a leader is actively quoting, but it has yet to prove it can hold its ground. Its total trading volume has fallen for three consecutive weeks, from $417M to $254M.

📈

The reverse signal comes from inventory. While weekly trading volume fell 39% from its peak, Entropy's open interest actually rose by 37% to $51.4M, with one asset from Anthropic accounting for $29.9M. Volume from wash trading will offset back and forth, leaving no inventory, so while trading share declines, open interest continues to accumulate, indicating real holdings rather than mere churn. These two facts point in opposite directions; the real focus should be on this tension itself, not on any single number.

The volume here needs to be discounted. Entropy has no tokens and has not confirmed any airdrops, but it can already see the pointsMultiplier parameter in the backend, and its leaderboard page states "coming soon," so part of the traffic is driven by expectations rather than product usage, and these two cannot be separated externally. The scale must also be highlighted: Entropy's volume is about $1.0B in a month, while Trade[XYZ] is at $64.6B, making it 1.6% of the leader's volume. It wins specific battles, not this category.

The assets it has registered for upcoming launches reveal the next steps. Entropy holds EWY, SBE, TCNT, and a DRAM index. The DRAM index is Trade[XYZ]'s fourth-largest product, while EWY is from South Korea. The next confrontation seems to be directly targeting the leader's core territory rather than seeking another uncompetitive pre-IPO name.

⚠️ Two Counting Rules

Only count online markets. The HIP-3 deployers often register assets long before activation; these assets will return oracle markPx, but midPx will be null, isDelisted will be true, open interest will be zero, and there will be no K-line history. Trade[XYZ] has 16, mkts 19, Paragon 9, and Entropy 4. HyENA's 25 are another matter, as they were activated and then shut down.

Code strings do not equal assets. para:STX is Seagate, with a mid-price of 799; the core market's STX is Stacks, with a mid-price of 0.27. Relying solely on codes would create a fictitious tenth confrontation market, which does not exist. Verify mid-prices before pairing.

Economic Account: The Cost of One Venue vs. Revenue Across the Entire Layer

No One Competes on Price, as Rates Have Hit Bottom

Everything hinges on two parameters for each asset, both publicly available in metaAndAssetCtxs: growthMode and deployerFeeScale. The all-in rate is base × (1 + s), where base is the standard perpetual fee rate table, and s is the deployer coefficient, which can be set from 0 to 3.00, capped at 1.00 in Growth Mode. The deployer takes s / (1 + s), so when s = 1.00, they split it evenly. Growth Mode further reduces the all-in figure by at least 90%, provided the market does not overlap with the perpetuals operated by validators, thus excluding crypto assets and crypto indices.

All stock venues have independently converged to the same configuration: maximum deployer sharing, with Growth Mode enabled. Entropy enters with differentiated products and venture capital without undercutting prices. The only venue not in this configuration is the one that just stopped trading.

🔧

Without relying on any aggregators, deployer sharing can still be accurately calculated. Fee income is deposited in the sub-venue accounts of the fee reception address: clearinghouseState with the dex field can read the unallocated balance, while transfers appear in userNonFundingLedgerUpdates, where the form is sourceDex equal to the venue name's send. Transfers are irregular, so measurements must be taken between two transfers. Since the transfer on August 27, Trade[XYZ] has accumulated $1,380,592, corresponding to a trading volume of $32.30B, or 0.427 bp, about $79,000 daily. Entropy has never made a transfer, so its accumulated fees can be read directly: $1.036B in trading corresponds to $41,468, or 0.400 bp.

Two Expenditures, Completely Different Natures

Staking is the most alarming number, but it will return. No one takes this money away. It is entrusted to validators, continuously generating staking rewards, and will be returned in the same form upon exit. It locks for at least 183 days since deployment and can be penalized by weighted validators for malicious market operations (such as pushing bad oracle prices), and can still be penalized during the 7-day unstaking queue, so a clean exit takes at least about 190 days. Felix and dreamcash have both fully withdrawn, and the current reading is zero.

Current staked amounts: Entropy 500,973, Paragon 500,712, HyENA 508,915 (stopped trading but not unstaked), Kinetiq 588,489 (one stake covering both km and mkts), Trade[XYZ] 500,488 plus another 500,269 on a different address. ABCDEx has only 1,004 HYPE, which has never been staked.

The money in asset positions is the part that no one mentions and cannot really be returned. Any perpetual DEX's first three assets are free, and each additional market must be purchased in a 31-hour Dutch auction using HYPE, starting at twice the last transaction price and linearly decaying to a bottom price of 500 HYPE. The auction for this asset position on September 14 started at a bottom price of 500 and ended at 500, so one asset position is about $39,900, and the demand for asset positions has decreased from a transaction price of 582 HYPE a week ago.

The last column is the key. Trade[XYZ]'s fees for about eight weeks could cover all asset bills. Every challenger except Entropy requires more time than HIP-3 has existed; Entropy can pass only because it bought seven asset positions instead of thirty.

Time will continue to amplify this asymmetry. Pausing a market is free and reversible; paid asset positions can be stored and reopened, and those "pending launch assets" come from this. Reserve positions accumulate based on historical deployments, with the formula being 7 + 0.2 × past auction deployment count, so Trade[XYZ] has about 30 that can be used immediately, while new entrants only have 7. A venue opening today, planning to create 20 markets, can list 10 at once, while the rest must queue in auctions, the fastest being one every 31 hours. Entropy's answer is not to play to this rhythm: it only launched five markets and made each one count.

Staking Yield Trap

Except for Entropy, every challenger that earns money through passive staking entry tickets earns more than from operating the exchange. Paragon's lifetime deployer share is $64,281, while $39.87M in staking at about 2.2% yields approximately $877,000 annually, a ratio of about 14 times.

This is not a consolation. That yield is newly issued HYPE from the protocol's future emission reserves, essentially inflation rather than income, a dilution return on a position that is forced to be held, with the valuation unit still being the asset that the operator passively longs. HYPE has dropped 30%, resulting in a loss of $12.0M on this staking, exceeding ten years of earnings. Over the past eight days, HYPE has fallen from $88.37 to $79.73, causing each staked amount to decrease by $4.3M.

How Big of a Business Can This Fee Pool Support

Growth Mode pins the effective fee rate around 0.4 bp, with the deployer taking half. Trade[XYZ] holds 97.8% of the share, with an annualized trading volume of about $786B, yielding approximately $29M in deployer share for the year. Rather than being the ceiling for the leader, this is essentially the total bonus across the entire venue layer under the current trading volume and fee rate floor.

Outside of the leading venues, there is approximately $167M in HYPE today, and each of these venues, regardless of their survival, has collectively earned only $747,000 in deployment fees over their lifetime. When placed alongside a $14M seed round, this arithmetic illustrates that under the current trading volume and fee floor, the HIP-3 operators cannot rely on trading fees for valuation. The value for challengers must come from elsewhere—tokens, frontend, customer relationships, or a product that the protocol has yet to price.

What Can Truly Be Sustained

HIP-3 has intentionally commoditized much of what a venue could originally sustain. Staking can be purchased, the underlying assets can be bought, the fee floor is shared, and even distribution is shared, as every HIP-3 market can be accessed from the same frontend.

The useful filtering question is not "What assets do you want to list?" because assets can be bought. Instead, it is "What do you have that the leader cannot buy with one asset position?" Among all the venues that have operated on HIP-3, only one has a clear answer to this question, and its answer is a set of oracles combined with a settlement design, not an asset list.

Entropy is the exception worth mentioning accurately because the answer does not lie in the underlying assets. It was founded by researchers and traders from Citadel Securities, Optiver, Millennium, and Polymarket, and this bench strength is reflected in two aspects. One is the order depth available from day one, which truly impressed Nebius. The other is the funding rate and settlement design, which reads like a direct response to the demise of Ventuals. Its $14M seed round was led by Ribbit Capital, whose main focus is on retail brokerage and fintech distribution, not DeFi, indicating an ambition to hold onto customers rather than just exploit a protocol. It is important to note that Entropy Advisors, which is deeply tied to Arbitrum DAO, and the custody startup called Entropy backed by a16z are entirely different companies, and no relationship with Hyper Foundation should be inferred from the name.

When viewed together, this sector still appears thin. A single contracting company accounts for 97.8% of the trading volume. A circle of challengers holds $167M in HYPE, which has resulted in a total of $747,000 in historical fees among them. The only team that can truly quote a price has just demonstrated its ability to capture a market for a week but cannot sustain it.

A more straightforward alternative is to take the protocol's half of the fees rather than the operator's half; it has no lock-up, no risk of forfeiture, and no operational burden. However, it is also not the major player in the Hyperliquid perpetual fee base, and it has been proven that HIP-3's share exceeding 50% is an illusion during a crypto downturn, not a trend.

Routing Layer and Its Actual Traffic

Builder code is the closest thing to "distribution business" on HIP-3. Frontends mark the orders they send out, earning a builder fee without needing to post collateral. Flowscan has counted 819 such marks.

The traffic they encounter is indeed not much. The routed trading volume is approximately $52.6B, accounting for about 9% of HIP-3's all-time $587B; in the last 30 days, it was about $5.3B, corresponding to $66.09B, roughly 8%. Over 90% of the traffic does not have frontend markings, which is exactly what a market dominated by market makers and API traders should look like.

The denominator in the fourth column is the $5.30B routed by builder code in the last 30 days, not HIP-3's $66.09B trading volume. If we replace it with the latter, even the largest, CoinDCX, accounts for only 0.7%. The ten companies in the table collectively account for 59.7% of the routed volume, while Flowscan has counted a total of 819 builder codes, meaning the remaining 809 share the last 40%.

Two things in this table are worth highlighting. Entropy's $423M routed volume all occurred in the last 30 days, while during the same period, it transacted $1.03B in its own venue, meaning about 40% of the market came from its own controlled frontend. Ribbit's ambition behind that round of retail brokerage is reflected in the data, not just in press releases, and this represents a different business from "being the deployer."

The other point is dreamcash, which provides a cleaner lesson. Its venue has been dead since July 2, with zero readings from its own market, yet its builder code still routed $17.3M in the last 30 days, totaling $3.54B. The deployer business and frontend business can be cleanly separated, and only one of them requires $40M to enter.

⚠️

There is an easy counting pitfall on third-party dashboards. Venue rankings usually default to cumulative trading volume, so a venue that stopped operating months ago may still appear to have a considerable share. Dreamcash reads out 3.3% of HIP-3's trading share in the all-time view, while in any recent window, it is $0. Before quoting any share, confirm which window it is calculated from.

Thus, the frontends that are truly worth monitoring are mostly not the deployers. Coinbase announced on September 12 a simplified perpetual interface within its wallet, powered by Hyperliquid, covering crypto, tokenized stocks, and prediction markets, aimed at markets outside the U.S. Kraken's parent company is also discussing integrating Hyperliquid-related perpetuals into a regulated U.S. venue. Neither will stake that 500,000 HYPE.

Conclusion: Difficult to Be Optimistic About Another HIP-3 Deployer

Putting all the previous points together, a new HIP-3 venue faces the following set of numbers.

All HIP-3 trading volume outside the leader totals $1.49B over 30 days, annualizing to about $18.1B. Based on the measured 0.400 bp, that results in a yearly deployer share of $725,000, to be divided among four venues. These four venues today hold about $167M in HYPE. The same amount of money staked passively at 2.2% would yield $3.67M annually.

In other words, the money made from operating these exchanges is about one-fifth of what the same funds would earn if staked in place.

This is not an undeveloped market but one that has already been priced close to zero, as evidenced in the previous sections. An asset position costs $39,900, making it impossible to sustain any listing; the fee floor is shared, leaving no room for price reduction; distribution is also shared, with builder code only encountering 9% of the traffic; and settlement assets have already converged to 6:6 in USDC, while AQAv2 clearly does not provide any fee tilt to HIP-3, indicating that the protocol has no intention of subsidizing this layer.

A more challenging aspect is the ceiling. XYZ, with a 97.8% share and every structural advantage, has only achieved a real transaction of 1.2% on the nine names it quotes, and its relative share in core storage assets is still declining. A new entrant faces not a "large leader" but rather a "leader that is already small and continues to shrink."

The exception remains only one, as previously clarified: a right that others cannot buy, a set of oracles that others cannot produce, or a funding rate and settlement design that can survive in thin markets. Entropy is the only one that fits this description, and it was pushed back after leading for a week. An asset list is not the answer, as two independent sets of data are now saying the same thing.

What Could Change This Judgment

• Entropy can sustain a market for a month instead of a week and maintain a share of Shandi rather than drifting around 13%. A decline in trading share while open interest continues to accumulate is the most noteworthy signal to watch, as it is more reliable than trading volume.

• The launch of Entropy's DRAM index and Korean assets. These two directly target the core territory of the leader, rather than another uncompetitive pre-IPO name, which would result in a much cleaner test than Nebius.

• The terms of the tokens can make the equity calculations work. The fee calculations do not work, and the tokens do not exist today.

• The pre-IPO sector can continue. Anthropic's market diminished after the first week, and OpenAI opened at $5.3M a day, now around $4M. As long as one stabilizes, this becomes a category rather than a one-time launch trend.

• The first payment of AQAv2 lands in early October. This will be the first time we can observe the true scale of the protocol's USDC revenue line, to compare against the current third-party estimates ranging from $135M to $200M.

• Easing of the fee floor. Hyperliquid signaled in early August that subsequent upgrades would allow HIP-3 deployers to increase fees by up to 3 times per individual asset, effectively reversing the discounts of Growth Mode. No timeline has been provided, and every number in Section 6 is built on the current floor.

• The implementation of a licensed market. HIP-3* was announced on September 3, a set of optional on-chain whitelists allowing deployers to limit which wallets can trade in a certain market, aimed at compliance and institutional access, currently only on the testnet. This is the first mechanism that could make "access rights" rather than "assets" a scarce item.

• The emergence of a team holding exclusive data or index licenses, with assets that indeed have a 7×24 native demand for crypto. This remains the only configuration where the logic of seizing positions completely fails.

• Evidence shows that the leader's trading volume cannot sustain beyond the end of Growth Mode. Its measured 0.427 bp is only about one-tenth of what the same market should receive at standard rates; if this exemption is burdensome, then the leader's stability is not as high as the share indicates.

• HyENA unstakes its 508,915 HYPE, confirming that this venue has ended rather than gone dormant.

Limitations

• The argument for seizing positions is based on measured shares, but the motivation belongs to inference. Five assets launched on the same day can also be interpreted as a pipeline that just happened to finish on that day.

• The judgment on Entropy is based solely on 26 days of data. The leading and subsequent reversal of Nebius are readings from a single week and a single medium-sized asset, neither of which should be treated as established conclusions going forward.

• The team background and funding rounds are sourced from company announcements and media reports, not verifiable on-chain information.

• The volatility conclusion in Section 2 is based on correlations from a sample of 45 working days, not causal decomposition. Trading volume and actual volatility may also be driven by the same factors, with the most direct candidate being the cooling down of overall AI capital expenditure transactions.

• The number of transactions does not equate to the number of traders; public APIs cannot produce independent traders.

• The capital return-related figures are calculated based on HYPE at $79.73 and an annualized staking rate of 2.2%, the latter being a protocol parameter rather than a contractual commitment, and it will decrease as the total network staking amount increases.

This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.