Markets are becoming more selective. AI spending is being rewarded only where it produces growth without crushing cash flow, while crypto continues to face ETF outflows and rising yields.

Within crypto, leadership rotated toward Solana and DEXs. We also examine the growing debate around TradeXYZ, Hyperliquid, and how value from HIP-3 is ultimately shared.

Over the last week, performance across the major benchmarks was mixed. The S&P 500 and Gold posted modest gains of 0.74% and 0.65% respectively, while the Nasdaq finished broadly flat. BTC was the weakest performer, ending the week down -3.0%.

Last week also brought a wave of earnings from the AI heavyweights. Amazon surged more than 15% after delivering its strongest quarterly revenue growth in over four years, with AWS revenue rising 37% year over year to $42.2B. Microsoft also gained more than 15% after reporting stronger than expected cloud growth while guiding to lower capital expenditures than Wall Street had anticipated. In contrast, Meta fell 10% after second quarter free cash flow dropped 91% as higher AI related capital expenditure weighed on profitability. One theme has become increasingly clear this earnings season. Markets are no longer rewarding AI spending alone. They are rewarding companies that can demonstrate returns on those investments without sacrificing cash flow.

Crypto, meanwhile, continued to face headwinds. ETF flows turned negative once again, with BTC and ETH ETFs recording net outflows of $255M and $69M respectively. Risk sentiment has also been pressured by the bond market, with the US 30 year Treasury yield climbing to 5.23%, its highest level since June 2007.

Within crypto, leadership rotated once again. The Solana ecosystem was the best performing sector, gaining 8.5% on the week, while the Ethereum ecosystem fell -8.8% after several weeks of strong performance following Robinhood Chain's launch. DEXs were the second strongest sector, posting gains of 5.2%.

The Solana ecosystem was driven by META, which rallied 36% following its Upbit listing, and PUMP, which gained 3% and accounts for roughly one third of the index. More importantly, both Pump.fun volumes and revenue continue to recover from their June lows, providing early signs that activity in the trenches may be returning.

The DEX sector was led by Uniswap, which gained 6.5% on the week. UNI benefited from the expansion of the fee switch across Robinhood Chain and the rollout of protocol fees on selected v4 deployments. 

A lot of the recent interesting onchain innovation, including FWA which we covered last week, is being built through Uniswap v4 hooks. Uniswap and its broader ecosystem should definitely be on your radar. 

— Kunal

The TradeXYZ Debate

With TradeXYZ's dominance of crypto volumes continuing, and RWAs now doing over 50% of Hyperliquid's volume, discourse over its alignment and its concentration on Hyperliquid has taken over. The concerns range from valid (Hyperliquid's long-term plans to monetize HIP-3) to much shakier (TradeXYZ leaving Hyperliquid), so the current relationship is worth laying out.

To be clear: TradeXYZ is an independent team building on Hyperliquid. It is programmatically required to share 50% of its HIP-3 revenue with Hyperliquid, and beyond that it can use its own half however it wants. We find that TradeXYZ (the same team as Unit) has used, and continues to use, its HIP-1 spot revenue to buy back HYPE, but does not do the same with its HIP-3 revenue.

The first concern, and in our view the weakest, is the idea that TradeXYZ will leave Hyperliquid, on the argument that the 50% share is too high and that it could capture more value on its own. Back in April, I argued the exact opposite: that Hyperliquid had outsourced too much of the value in its biggest markets to deployers. Consider what each side supplies. Hyperliquid provides the infrastructure layer, the collateral, and above all the user base behind the majority of TradeXYZ's volume. To leave, TradeXYZ would need to rebuild the exchange layer, the hardest piece of the stack, forfeit nearly its entire trader base, and commit reputational suicide in the process. For Hyperliquid, in-housing RWAs would be a similar form of reputational suicide: undermining the dominant deployer behind so much of its success would signal to every future HIP-3 deployer, and every builder on Hyperliquid, that any team successful enough will be replaced. This is one of the most symbiotic relationships in crypto, and it makes no sense reputationally, economically, or architecturally for either side to leave the other.

The second concern, over monetization, is the fair one, but it needs nuance. First, it is disingenuous to ignore that Hyperliquid's RWA markets would not have reached this scale without TradeXYZ's execution; 100% ownership of a much smaller pool would be worth less than half of this one. Second, the 50% fee share is not Hyperliquid's only avenue: it also monetizes through write priority fees and the read fees market makers pay.

It also monetizes through second-order effects such as increased USDC supply, where it retains 90% of the cost-adjusted revenue on balances sitting on the chain. Open interest is the proxy: since HIP-3 went live, its OI has grown $3.68B against $1.38B of trailing-year USDC supply growth, with crypto OI ending the year lower. As more traders bring USDC on to long RWAs, the revenue on that supply accrues to Hyperliquid as well. That revenue, projected at around $30M a month, is already larger than the entire HIP-3 perp fee pool that Hyperliquid and TradeXYZ split between them.

To us, the most interesting question is less the 50/50 split between TradeXYZ and Hyperliquid than the path out of growth mode, and how both teams eventually transition from a low take rate to a steadier, higher fee base.

— Shaunda

Spark Q2 Token Holder Report

The report finds that Spark continued gaining share despite one of the weakest quarters for DeFi lending. The report highlights Spark more than doubling its lending market share while remaining profitable, even as revenue across the sector declined. It also points to the Spark Liquidity Layer becoming a key distribution engine through integrations with Uniswap, BitGo, and Robinhood Earn, while the protocol's risk management attracted more than $1B of inflows following the April market disruption. Looking ahead, the key questions are whether deposit growth persists, distribution rewards continue scaling, and lending demand recovers enough for Spark's larger market share to translate into stronger earnings. 

The Rise and Quiet Death of the Neobank

The report finds that the neobank sector has reached massive scale, with 368 active companies serving roughly 1.46 billion users, but much of that growth rests on fragile infrastructure. Only about one third hold full banking licenses, leaving most dependent on sponsor banks, BaaS providers, and card issuers that can become single points of failure. The report also finds that shutdowns are far more common than public coverage suggests and that real AI adoption is limited, with emerging market lenders leading because automated underwriting solves a genuine need. The core takeaway is that neobanks have transformed access to finance, but the next phase will be shaped by licensing, infrastructure resilience, and banking rails built for AI agents.