How to Trade NCLD (Roundhill Neocloud ETF) on Hyperliquid

What NCLD Actually Owns

NCLD is the Roundhill Neocloud ETF, listed on Nasdaq on August 6, 2026. Roundhill describes it as actively managed exposure to "GPU-as-a-Service platforms and AI data centers that support high-performance computing." In plain terms: the companies that buy Nvidia racks and rent the compute out by the hour.

The portfolio is roughly 20 names, and it is not evenly spread. Weights at launch put Nebius Group (NBIS) near 30.8% and CoreWeave (CRWV) near 27.3% — about 58% of the fund in two positions. Behind them: IREN (~7.7%), Hut 8 (~6.3%), TeraWulf (~5.0%), Applied Digital (~4.8%), Cipher Mining (~4.2%), Galaxy Digital (~4.1%) and Core Scientific (~3.9%).

That second tier is the part most people miss. Nearly all of it is former bitcoin miners that converted power contracts and rack space into AI hosting. So NCLD stacks two distinct bets: pure-play GPU clouds, and the miner-to-compute conversion trade. They do not always move together, and the miner cohort carries balance-sheet risk the pure-plays do not.

Structure: 0.65% gross expense ratio, quarterly rebalancing, non-diversified, listed options. Roundhill's own risk language flags rapid technological change, intense competition, obsolescence risk, concentration in information technology and industrials, new fund risk, and significant Asia exposure. Read that list before you decide this is a diversified AI proxy — it isn't, and the issuer says so.

Why the Neocloud Trade Got Its Own Ticker

The thesis is that AI compute demand is outrunning supply, and that neocloud operators capture the shortfall with multi-year contracts that generate steadier revenue than on-demand hyperscaler pricing. Roundhill CEO Dave Mazza framed the launch directly: "DRAM proved that investors want precise exposure to the AI infrastructure trade, not another broad tech fund."

The launch window was fortunate. NCLD started trading days before its two largest holdings reported. CoreWeave posted Q2 2026 revenue of $2.58 billion, up 112% year over year, with backlog of $104.2 billion and raised full-year guidance toward a $12.8 billion midpoint. Nebius reported $582 million in revenue, up 454% year over year, guiding to $3–$3.4 billion for the full year. Both stocks repriced hard on the prints, with CoreWeave up 19% and Nebius up 34%.

NCLD carried that. The fund gained roughly 15% in its first week on about $46 million of cumulative volume, moving off a $22–$24 opening range. It traded between $26.02 and $27.07 on August 14, and the HIP-3 perp is quoting near $26.70.

Two honest counterpoints. First, a lot of the actual buildout is private — Crusoe, Nscale, Lambda and Together AI are not listed, so any public neocloud fund is structurally forced to overweight CoreWeave, Nebius and the miner converts. You are buying the tradeable subset, not the industry. Second, the financing structure invites scrutiny: Nvidia holds stakes in the same neocloud customers it books revenue from, a loop that critics describe as circular. If GPU-hour pricing compresses or utilization slips, the concentrated names take it first and NCLD takes it hardest.

How the HIP-3 Perp on NCLD Works

The NCLD market on Hyperliquid is a HIP-3 builder-deployed perpetual from the xyz deployer, settling in USDC on HyperCore's shared order book. One contract references one share of the ETF, quoted in USD. It never expires and there is no roll — hourly funding does the work that basis decay does in a dated future. When the book trades above the oracle, longs pay shorts; below, shorts pay longs.

HIP-3 markets use a more responsive premium formula than native perps — premium = (0.5 × (impact_bid_px + impact_ask_px) / oracle_px) − 1 — which lets deployers express a wider funding range. Funding is capped at 4% per hour and is charged against position_size × oracle_price × funding_rate, using the oracle rather than the mark. That last detail matters when the book is thin and the mark wanders.

Leverage is capped at 10x. At $26.70 a share, $1,000 of margin carries about $10,000 notional, roughly 375 share-equivalents. There is no KYC, no custody handoff, and no gas on orders.

The real constraint is depth, not access. Twenty-four-hour volume on the NCLD perp is about $136,719. Compare that to the ETF's own $13–$14 million of daily turnover on Nasdaq in its first week. This is a small book on a two-week-old underlying. Market orders will move it.

Key Trading Considerations

Size to the book, not to your conviction. At six figures of daily perp volume, a position that would be routine on BTC is a market-moving order here. Use limits, expect wide spreads, and assume exit liquidity is worse than entry liquidity.

Nasdaq closes; this market does not. Overnight and weekend price discovery happens on the Hyperliquid book alone, with funding as the tether back toward the reference. Expect funding to run hot during closed hours and expect gaps at the equity reopen. That is the feature — round-the-clock access to an instrument that otherwise sits idle from Friday close — but it's also where the risk concentrates.

You are trading two stocks wearing an ETF wrapper. With ~58% in NBIS and CRWV, the catalysts that matter are their earnings, Nvidia's results, and hyperscaler capex commentary — not anything specific to the fund. Both names posted single-day moves of 19% and 34% this month. At 10x, a 10% adverse move against you is the whole margin before fees and funding. Do that math before you set size.

The underlying is new. NCLD launched August 6, 2026 with assets in the $15–$23 million range in its first days. There is no track record, no seasoned premium/discount behavior, and quarterly rebalancing means today's 58% concentration can drift meaningfully before it's corrected. The fund also distributes annually — confirm with the deployer how, or whether, a price-referenced perp handles that, rather than assuming it passes through.

Watch the fundamentals, not just the tape. GPU-hour pricing, fleet utilization, the first neocloud IPO, and whether hyperscalers keep renting capacity instead of building it in-house are the variables that decide this thesis. Those show up in the constituents' filings well before they show up in a two-week-old ETF's chart.