RWA Perps on Hyperliquid: Not Half the Tape — a Measured Census

Revised Sep 14, 2026: the original version of this report relied on third-party dashboards and repeated an oversized share figure ("half the tape") alongside event details we could not verify. It has been rebuilt on a market-by-market census from Hyperliquid's perpDexs endpoints. The corrected picture: RWA-flavored markets are the most valuable part of a builder book that is 28.5% of platform open interest — not half.

Thesis: real-world-asset perpetuals — the S&P 500, SK Hynix, gold, Brent crude, tradable around the clock in stablecoin margin — are the single most compelling product Hyperliquid has added since launch, and the most misrepresented in size. We measured every builder market directly from the API. The honest numbers: a builder book of $4.12B inside a $14.44B platform (28.5%), led by RWA names that would each rank among the largest markets on most venues, run overwhelmingly by one deployer, and funded by a fee split that redirects roughly a fifth of every fee dollar away from HYPE buybacks. This report sizes the tape market by market, explains the product mechanics that make it work, decomposes the cost side, and states plainly what we cannot verify.

1. The setup — how RWA perps got here

When HIP-3 went live in October 2025, it let approved deployers stake 500,000 HYPE (per the HIP-3 specification) to operate their own perpetual markets on Hyperliquid's order books — and keep up to half the trading fees those markets generate. The design converted Hyperliquid from a single venue into a marketplace of markets: the deployer carries the operational burden — oracles, margin parameters, market creation — in exchange for a share of the economics.

Deployers chased a product category centralized crypto venues had largely avoided for regulatory reasons: tokenized exposure to equities, commodities and indices, settling in stablecoins, with no expiry, trading through weekends when the NYSE is closed. Leveraged gold exposure at 2 a.m. on a Sunday has few other homes. Within ten months the builder book grew from a rounding error to roughly a third of platform open interest at its August peak — and it is dominated by RWA names.

The historical irony still frames the story: the perpetual future was invented by BitMEX, which has announced its own shutdown (effective September 23). The product left the CEX; the assets — stocks, commodities, indices — followed it on-chain.

2. The measured census — how big is the tape, really

Queried directly from the perpDexs registry and each deployer's asset contexts, Sep 14, 2026:

The historical shape matters as much as the level. Reconstructed snapshots put the builder share at ~18% of OI in March, rising to roughly a third in August, and settling back to 28.5% by mid-September — because August's crypto rally re-inflated the core book faster than the builder book grew. So the correct one-line answer to "isn't RWA half the tape?" is: no — the whole builder book is 28.5%, and RWA names are the largest slice of it. A clean RWA-versus-crypto split of every builder market is not published anywhere we can audit, so we do not pretend to one; what we can do is name and size the RWA markets that anchor the book, which is the next section.

3. The anchor markets, sized

The largest RWA-flavored markets on the platform, all measured Sep 14:

Just the three largest named RWA markets — SP500, SKHX and GOLD — hold ~$1.14B of open interest, roughly 8% of the entire platform, inside a segment that did not exist on-chain at scale eighteen months ago. Two design details from the census deserve attention. First, headroom is configured, not aspirational: deployers set explicit OI caps ($1B on CL, $750M on BRENTOIL, $200M on AAPL), and the largest markets run below them — growth in this tape is a cap-raise decision, not a listing decision. Second, the cluster effect is real: SKHX does not trade alone; the memory-chip complex (SK Hynix, SKHY, Micron, SanDisk) forms a genuine sector tape on-chain, the kind of thing that took derivatives exchanges decades to build.

4. The product anatomy — why this tape exists at all

• 24/7 settlement in stablecoins. The contracts never expire and margin in USDC, so weekend gaps and after-hours catalysts — the exact windows where traditional venues are dark — are tradeable. That is the entire retail value proposition, and it cannot be replicated by a CEX with regulated equity exposure.
• Deployer-sourced oracles. Each market prices from feeds the deployer configures — which is also why the equity and commodity names inherit the data quality of their underlying venues. This is the segment's central engineering risk, and it is a deployer-level responsibility, not a protocol-level one.
• Explicit risk budgets. OI caps per market act as circuit breakers on positioning growth; a market that fills its cap must raise it deliberately, which creates a visible, auditable growth trail.
• Counter-cyclical demand. The memory-chip cluster trades on semiconductor catalysts, gold and oil on macro ones — flow that does not correlate with BTC beta. In a crypto risk-off week, this is the part of the book most likely to keep generating fees.

5. The cost side — what the tape pays and what it takes

HIP-3's bargain: the deployer keeps up to 50% of the fees on its own markets; on core markets, essentially all net fees (~97%) fund HYPE buybacks. The aggregate cost of the builder book shows up in the weekly reconciliation — on recent weeks (~$13–16M gross), roughly 18% of fee dollars pass through to builders, market makers and the HLP vault: about $2.3–2.9M a week, on the order of $120–150M annualized redirected from the Assistance Fund to market operators. We label the ~18% an estimate reconciled from weekly prints; the transparency proposal awaiting its final vote would turn it into disclosed, per-deployer data.

Whether that is overpayment depends on what the tape delivers: category-defining markets (SP500, SKHX, GOLD at real scale), a fee base that trades when crypto doesn't, and configured headroom for growth — versus a capture leak that scales with the builder share. That trade-off is the same one examined in our HIP-3 deep dive; this report's contribution is the corrected denominator: 28.5%, not half.

6. Concentration and the first exit

Two census facts reframe the risk discussion. First: Trade.xyz runs 98% of builder open interest — $4.05B across 104 markets. Every RWA anchor market above is a Trade.xyz market; its oracle choices are, in practice, the platform's equity-and-commodity price feed. A deployer-level dislocation there would stress more than a quarter of the entire platform book at once — larger in absolute dollars than the "half the tape" narrative ever implied, concentrated in one operational domain.

Second: the deployer set is not permanent. HyENA — one of the ten registered builder DEXs — returned zero active markets at our measurement, a complete exit corroborated by dated press. Nothing locks a deployer in. That is self-cleaning for the ecosystem (weak operators leave rather than zombie on) and a live reminder that builder OI can vanish without a protocol decision. Watch whether other small deployers follow, and whether HyENA's niche is filled by migration or simply disappears.

7. What it means for HYPE

HYPE's demand machine is mechanical: roughly 97% of net protocol fees are programmed into Assistance Fund buybacks. The RWA tape affects that machine twice — negatively through the pass-through (section 5), positively through a fee base that survives crypto drawdowns. The measured context as of this revision: August delivered $67.3M of protocol fees (annualizing to ~$0.81B), the two following weeks faded to $16.5M and $12.9M, and roughly $2.0B of unlock supply (24.1M HYPE across Aug 29 and Sep 6) was absorbed without breaking price — the strongest demand environment in the platform's history, now cooling. HYPE printed an all-time high of $89.60 on Sep 6 and closed the week of Sep 7–13 down 12.1%.

We deliberately do not publish a price-to-fees multiple for August: a verified historical circulating-supply series is unavailable to us, and the current-snapshot multiple (~26x on circulating supply, per our Sep 7–13 report) moves with the price daily. What we will say: the RWA tape is the strongest long-run argument that Hyperliquid's fee base is more than a crypto-beta trade — and the pass-through is its ongoing cost.

8. What to watch

• The builder share trajectory — 28.5% and falling into a crypto rally, or cap-raises pushing it back toward a third? The census is repeatable; we re-run it for the weekly data report.
• Transparency proposal final vote — per-deployer fee disclosure converts the ~18% estimate into data and makes concentration priceable.
• Cap raises on the anchor markets — SP500, GOLD, CL and BRENTOIL running at their configured ceilings would be the clearest demand signal in the segment.
• Deployer exits — HyENA set the precedent; further exits shrink the breadth argument the whole HIP-3 bargain rests on.
• AQAv2's first print (Oct 3) — the reserve-yield stream (~$527K/day modeled) is the demand-side offset to the pass-through; its first execution is the quarter's key datum for the buyback math.

9. Conclusion

RWA perps are the best product story on Hyperliquid and the worst-marketed number in its ecosystem. Measured market by market, the tape is enormous by any historical standard for on-chain finance — three markets worth over $300M each, a real sector cluster in semiconductors, configured headroom for more — and simultaneously not the "half the tape" that the commentary cycle repeats. The builder book is 28.5% of open interest, run 98% by one deployer, funded by a fifth of the fee stream, and already carrying its first full deployer exit. None of that makes the segment less compelling; it makes it measurable. Our advice for reading this market: trust the census, watch the caps, and treat every share figure — including in our own archive — as a hypothesis the API can test.

FAQ

Related reading

• Hyperliquid Weekly Data Report — Week of August 17–23, 2026
• HYPE Market Intelligence — Week of August 17–23, 2026
• Hyperliquid Weekly News Review — Week of August 17–23, 2026
• Half the Tape? What HIP-3 Builder Markets Really Take and Give