Weekly Intelligence — Week of August 31–September 6, 2026

Revised Sep 14, 2026: every figure re-verified against primary sources; claims are graded by the evidence behind them.

Five questions on the week, answered strictly from verifiable data — and, where the data can't answer a question, that gets said too. Each key claim is graded in the verification scorecard by the evidence behind it.

Q1 — Two unlocks in nine days, both absorbed. What does that actually prove?

What the data says: the market absorbed ~$2B of scheduled supply inside nine days — 14.18M HYPE on Aug 29 (unlock-day close +3.0%) and 9.92M HYPE on Sep 6 (price ran to the $89.60 record on release day). The prior base rate was two declines and a flat print across May–July releases (AInvest). Demand at these levels is deeper than the unlock-fear framework priced.

What it doesn't say: where the tokens settled. Absorption at trade time is verifiable on candles; holding-period behavior is not, on any public API. The documented bid — a 29.3M-HYPE treasury vehicle, a staking ETF, a standing buyback converting ~99% of revenue (third-party estimate), and soon the AQAv2 stream — is consistent with what the candles show, but consistency is not proof of custody.

Q2 — Price at an all-time high while fees fall a third straight week. Which series bends first?

The bull frame: fees were a squeeze artifact normalizing toward baseline, and price is re-rating the machine — two absorbed unlocks, 48.45M HYPE burned to date, a second revenue stream activating Oct 3, a reported on-shore path. The bear frame: churn is decaying toward the pre-spike $8–9M baseline while price makes records — a divergence that has historically resolved in favor of the flow series. The tie-breakers are the other stocks the APIs do print: TVL held $6.41B (no exit), and platform-wide OI held ~$14.4B (no unwind). A market re-rating a machine it still trusts looks exactly like this; so does a market late to a fade. The next two fee prints referee.

Q3 — How much is the U.S. on-shore path actually worth?

The record so far, in dated steps: Aug 19 — the President says the CFTC chair is "working to bring Hyperliquid into the United States in a fully compliant and legal fashion" (AInvest). Sep 4 — reporting describes a framework involving Kraken's parent company and the regulated platform Bitnomial to offer select crypto perps to U.S. users (The Block, via CoinMarketCap). No license, no structure, no timeline has been announced.

The honest valuation: unmeasurable, directionally large. U.S. retail access would open the deepest notional market in the world to a venue that already runs ~$14.4B of open interest without it. But the delta between "priced prospect" and "licensed reality" is exactly where narratives over-run filings. This column's rule stands: the price moved on the comments — $62 → $70 on Aug 19, records through September — and the scorecard grades those moves Confirmed, while the on-shore outcome itself stays Unavailable until a filing exists.

Q4 — What is the second buyback engine worth, on current math?

Mechanism (confirmed, ETHNews): since Aug 26, ~90% of the yield on USDC held across the platform flows to the Assistance Fund; Coinbase manages the balances, Circle the technical side; first payment Oct 3.

On that arithmetic — and it is arithmetic on assumptions — the yield stream would add roughly a quarter of the current fee stream, decorrelated from trading volume. The design's real significance: the fee engine breathes with volatility (down 33% in three weeks); the yield engine breathes with rates and stablecoin float. The first official payment on Oct 3 converts this table from modeled to measured.

Q5 — What triggers matter from here?

• Oct 3: the first AQAv2 payment — the single largest estimate in Hyperliquid's token thesis becomes a line item.
• Sep 29: the third unlock of the month (~14.18M HYPE) — pattern or fatigue, the third data point decides.
• The fee print path: two more weeks above ~$14M would confirm a re-based baseline; a slide toward $8–9M hands the divergence to the bears.
• The $90 level: the record sits at $89.60 — acceptance above the round number opens uncharted discovery; rejection makes a triple-top.

The week, day by day

Prior Sunday closed at $79.99; this one finished at $87.94 with the all-time high printed on the final session. The daily table also shows exactly where the second unlock cleared.

Three structural notes. Monday's +5.2% was the week's real launch — the reversal of the prior Sunday's mechanical −4.1% handoff. Thursday's +6.8% to an $88.20 intraday high was the fee-peak session ($3.47M), the only day above $3M all week. And the record itself, $89.60 intraday, printed Sunday Sep 6 — the same session the second unlock (9.92M HYPE, ~$820M at week prices) cleared, and the market closed +2.9% anyway. The seven daily fees sum to $16.5M, matching the weekly print exactly.

The week in one glance: close $87.94 (+9.9%), range $79.52–$89.60, fees $16.5M (third straight weekly decline), TVL $6.41B, BTC +3.4%, ETH +4.0% — price discovery continuing into a falling fee tape.

The verification scorecard

Every load-bearing claim this week, graded by evidence rather than by whether we agreed with it:

Three confirmed, three partial, one modeled, and the unavailable list stated plainly. That is what a scorecard looks like when it grades evidence instead of predictions.

Frequently asked questions

Related reading

• Hyperliquid Weekly Data Report — Week of August 31–September 6, 2026
• Hyperliquid Weekly News Review — Week of August 31–September 6, 2026
• Hyperliquid's Flywheel, Measured: Fees, Buybacks and the Second Engine
• Weekly Intelligence — Week of September 7–13, 2026