Hi all, happy Thursday!

Crypto's July rebound stalled on Wednesday as renewed US-Iran conflict lifted oil and revived the inflation worry driving this year's selloff. 

Bitcoin steadied near $62,750 this morning, up about 1% and back above $62,000, while gold slid for a fourth day and Treasury yields climbed as the market treats the conflict as a rate event more than a haven trade. 

In today's edition, we dig into why GRASS fell about a third even as revenue matched all of last year in six months, and the accrual gap that keeps its fundamentals from holding a bid.

Bitcoin trades near $62,750 on Thursday morning, up about 1% and back above $62,000 after Wednesday's selloff, holding a weekly gain near 2.6%. ETH sits near $1,745, and SOL near $77.60, down about 2% on the week after erasing its entire July rally. The rebound broke on an external shock, and price has held above $62,000 since.

The shock came from the Strait of Hormuz. The US struck Iranian targets after Iran fired on tankers in the strait, Iran answered with strikes on US installations in Bahrain and Kuwait, and Trump used a NATO summit to call the ceasefire "over" and further talks a "waste of time." Oil jumped about 6%, with WTI near $74 and Brent near $78. Equity futures lost about 1.5% and the dollar firmed. Crypto sold with the risk-off move.

The channel back to crypto runs through inflation. This is the same conflict that pushed oil above $100 in February and delivered a global price shock, and sustained crude above $75 feeds the inflation expectations that set rate policy. That cuts against the easing-inflation read the July bounce leaned on. Today's forced selling hit longs, the mirror of the short squeeze that carried the bounce up, with $350 million of about $450 million in liquidations landing on altcoin pairs. JUP, ETHFI and PUMP fell between 5.5% and 9.3%.

Bitcoin took the hit like a tech stock, down 2.5% at its worst before it steadied. Bitmine added 40,000 ETH, about $71.6 million, on top of 42,000 the prior week. A quieter bid runs the other way, since financial isolation of Iran lifts demand for rails that no government controls, and US authorities have frozen about $500 million in Iranian-linked digital assets. Direction sits with the headlines rather than the chain: whether the strikes escalate and whether oil holds above $75. The PUMP unlock on July 12 is the next scheduled test.

The selloff hit our sector indices as a rotation. Most of the board closed Wednesday red, and Solana Eco led the losses near 10% down, a full retrace of its early-July run and a clean read on where the altcoin liquidations landed. Buyback Leaders sat at the other end up around 5.5%, with Crypto Miners the other green sleeve up close to 4%. Buyers paid up for revenue and buyback names on the same day they dumped their highest-beta ecosystem.

— Nick

Grass sells the news

Grass held its second Token Holder and Network Participants Call on Tuesday, and GRASS closed down about 32% at $0.37. The reaction ran opposite to the disclosure, which is the tension worth holding.

Start with revenue, since it anchors the rest. Grass booked $17M in 2025, split $2.7M in the first half and $14.3M in the second. It then matched that full-year figure in H1 2026, close to six times what it booked a year earlier. The team guided to $70M for the full year, with the second half projected near $52M. Another ~$15M of H1 business slipped into Q3, and management flagged $35M of late-stage opportunities opening in the back half, so the H1 print understates the pace more than it flatters it.

The callback stings for us. Our December report annualized the Q4 2025 projection of $12.8M to a $51.2M run-rate and read it as validation of product-market fit. The full year landed at $17M. Q4 came in near $10M against that projection, and the run-rate framing the market leaned on, ours included, overshot a business lumpier than any single annualized quarter implies.

However, the margin disclosure we asked for in December arrived. Grass reported ~20% net margin in H1 2026 on operating expenses of $2-3M per month, most of it infrastructure rather than headcount, which ran about 20% of first-half spend. The team bought compute and storage ahead of the 2025 price hikes, and says those purchases now cut monthly opex by more than $1M. 

Grass confirmed Stage 2 rewards will be paid in USDC from revenue instead of GRASS tokens, and stressed the payouts do not expand supply. That stops subsidizing supply with dilution and ties payouts to demand the network creates. On the ground it landed as betrayal. Node runners who expected a Season 2 allocation from the ~170M GRASS pool got a final USDC distribution weighted toward actual node usage. 

The call settled one debate and dodged another. Grass confirmed that Wynd Labs is the engineering shop, and the IP, customer contracts, and revenue sit with Grass, and the team has raised no equity and plans none. The token is the sole instrument the business accrues to, and the team said as much. The question it left open is the one a Q&A slide raised: how USDC rewards affect value capture for the token. The team reaffirmed that value accrues to GRASS without naming the mechanism that carries Foundation revenue to the holder. Grass ran a one-time $350K open-market buyback back in November and has not renewed it since, a rounding error against a ~$240M market cap. For a token whose entire case is "the business value lives here," the missing accrual path explains why strong fundamentals keep failing to hold a bid.

The team paused Grasshopper, the hardware device, arguing network optimization already delivers most of what it would have added, so the capital goes elsewhere. An in-app non-custodial wallet ships with Stage 2 and will let contributors swap USDC into GRASS. The most material is Live Context Retrieval (LCR), the inference-time product built on the same crawl stack, with initial release due later this summer. Grass has pitched LCR since it named the product a Stage 2 deliverable in September 2024, and it has stayed "in trials" through both token holder calls since, so the summer window arrives close to two years after the first promise. LCR is where AI usage converts into recurring data demand, and it is the piece most likely to re-rate the token if it ships and monetizes.

GRASS trades at a circulating cap near an FDV around $384M against 2026 guidance of $70M with no equity above the token. On catalysts it is empty until LCR. The market spent Tuesday resolving that contradiction toward the near-term void, and the contributor backlash adds a supply of sellers who no longer want the exposure. Calling current prices an entry depends on your willingness to underwrite a product that is not live, on a token whose accrual mechanism the team still will not name.

— Nick

How Attention Became the New Capital

In this episode of Invest Like The Best, Jeremy Giffon argues that in periods of technological disruption, the scarcest asset is not capital but narrative: investors, founders, and institutions that define a compelling story can attract billions of dollars and shape where capital flows. It explores how AI and algorithmic platforms, particularly X, are reshaping finance, media, and politics by making attention, posting, and "timeline-native" influence increasingly valuable, even eclipsing traditional markers of status like wealth. Looking ahead, Jeremy argues that AI will fundamentally alter software economics, capital allocation, and work itself, favoring businesses with massive scale while creating new opportunities for founders and investors who can adapt quickly to a rapidly changing technological landscape.

Compute Capital Markets: Financializing the AI Supply Chain

This report argues that GPU compute is rapidly becoming a tradable financial asset, with traditional exchanges launching compute futures and ETF issuers racing to build investment products as AI-driven demand transforms compute into a commodity market. While these financial markets solve price discovery and hedging, the biggest long-term opportunities remain in the infrastructure they leave unresolved, including trusted pricing benchmarks, physical delivery, cryptographic verification of compute quality, and GPU financing. We are constructive on the emergence of compute capital markets but remain neutral on liquid compute tokens, arguing that today's token valuations already price in significant adoption while the strongest long-term value is likely to accrue to private index providers and exchange operators that control the market's core data and settlement infrastructure.