Happy Monday!

Markets diverged sharply last week as BTC gained 1.4% while equities, gold, and the broader AI trade sold off. The move offered an early indication that BTC may be regaining relative strength, although modest ETF inflows suggest institutional demand has yet to confirm the shift. Within crypto, liquidity rotated back toward Solana, led by PUMP and a broader rebound across ecosystem tokens.

We then examine Hyperliquid’s plan to make HIP-4 deployment permissionless. The key detail is the initial allocation of 100 outcomes per deployer, which materially lowers the marginal cost of launching markets after the initial stake and could accelerate growth from HIP-4’s currently muted base.

Last week saw a sharp divergence across major asset classes. BTC was the only major benchmark to finish the week in the green, rising 1.4%, while the Nasdaq, S&P 500 and Gold fell -3.3%, -1.2% and -2.1% respectively. If sustained, this would mark a notable shift after BTC broadly underperformed traditional risk assets over the past year.

The weakness in equities was concentrated in the AI trade. Hot names such as Marvell (-20%), IBM (-26.9%), Super Micro Computer (-13.8%), Sandisk (-24.5%) and KOSPI (-12%) all sold off sharply. Investors have begun questioning whether the massive AI infrastructure spending can generate adequate returns, while concerns around cheaper Chinese models and delayed product launches triggered profit-taking that was amplified by leveraged positioning.

BTC, meanwhile, showed signs of decoupling. Part of the strength may reflect capital rotating out of crowded AI positions into sectors that have significantly underperformed, with BTC still down more than 45% over the past year. That said, institutional demand remains modest. Spot BTC ETFs recorded just $22M of net inflows last week, making ETF flows the key metric to watch if this relative strength is to continue.

Within crypto, the launchpad sector was the best performer, gaining 6.8% on the week. The rally was almost entirely driven by PUMP, which surged 37% after Ansem revealed a $100K purchase. 

However, the timing of several high-profile bullish posts has drawn scrutiny as they coincided with investor and team token unlocks. Fundamentally, activity has yet to meaningfully accelerate, with weekly memecoin volumes sitting at roughly $1.8B, only around 30% above recent lows.

We also saw early signs that liquidity may be rotating back toward the Solana ecosystem. After highlighting last week how capital had shifted toward Robinhood Chain assets, Solana ecosystem tokens rebounded 3.9% while the Ethereum ecosystem fell -6.8%. Tokens such as JTO, JUP, META and CLOUD all posted strong gains alongside the PUMP rally.

The key question now is whether this marks the beginning of a sustained rotation back into Solana or simply a short-lived bounce. Alongside ETF flows, where liquidity chooses to concentrate over the coming weeks will be one of the clearest signals for the next leg of the market.

— Kunal

Hyperliquid Opens HIP-4

Hyperliquid announced that HIP-4 will adopt a permissionless deployer model, extending the HIP-3 framework to outcome markets. Under a future upgrade, deployers staking 500,000 HYPE will be able to define and settle markets using validator-approved templates, receive initial capacity for 100 outcomes, recycle that capacity after settlement, and eventually retain up to 50% of trading fees. By comparison, HIP-3 deployers receive only three auction-free initial markets.

We believe the 100-outcome allocation is underappreciated. HIP-3 deployment has become increasingly concentrated as auction costs have risen, raising questions about whether lower-volume outcome markets could support attractive deployer economics. HIP-4 materially lowers the marginal cost of market creation by allowing each deployer to launch up to 100 outcomes without additional bids, although the process for acquiring further capacity remains unclear.

This could be a meaningful catalyst for a product whose traction remains limited. HIP-4 generated $17.8 million in weekly volume, or approximately $2.5 million per day, down 8.9% week over week, while average daily users fell 23.3% to 1.2K despite the high-profile World Cup. Delegating market creation should broaden the available market set and give builders that have historically favored Polymarket or Kalshi a stronger reason to integrate. The potential flywheel is straightforward: more markets attract more builders and order flow, improving deployer economics and incentivizing further listings. 

In terms of deployers to watch, we highlight Outcome. The team has spent nearly two years building prediction-market infrastructure on Hyperliquid, has staked 500,000 HYPE, and reports more than $30 million in HIP-4 volume. During the World Cup, Outcome says its markets generated close to $20 million in volume, accounted for 40% of Hyperliquid’s World Cup activity on its busiest day, and maintained median pre-kickoff spreads of 0.2 cents, roughly five times tighter than comparable markets on Polymarket and Kalshi. It also reports that approximately one-third of its active traders were new to Hyperliquid, highlighting the symbiotic relationship deployers have with Hyperliquid.

— Shaunda

Trust or Verify: The Case for Private AI

IOSG Ventures argues that private AI will become increasingly important as enterprises and individuals grow more concerned about intellectual property leakage, surveillance, and data privacy when using frontier AI models. The report compares policy-based privacy solutions with cryptographically verifiable approaches such as TEEs, end-to-end encryption, and local inference, arguing that advances in confidential computing are making private AI both practical and affordable. It concludes that the biggest long-term opportunity lies not in basic private inference, but in solving the remaining challenges around private training, agent workflows, and encrypted tool calls, where defensible infrastructure has yet to emerge. 

The Path Forward for HIP-4

Outcome argues that HIP-4’s current validator-led operating model is limiting market creation, liquidity, and adoption despite Hyperliquid’s strong underlying infrastructure. The report proposes extending the HIP-3 framework to outcome markets by allowing qualified deployers to stake HYPE, list and resolve markets, earn fees, and face slashing for misconduct, arguing that prediction markets require dedicated operators to manage distribution, liquidity incentives, contract design, and settlement. It concludes that permissionless deployment could unlock a broader set of prediction and structured markets on Hyperliquid, with Outcome positioning itself as an early operator after generating more than $30 million in HIP-4 volume and bringing new users onto the platform.