BTC $76,761 (-0.16%), ETH $2,113 (-0.11%), SOL $84.71 (+0.25%), XRP $1.37 (-0.62%). Markets are flat to slightly lower as Bitcoin holds the level it dropped to after Monday’s $600M liquidation event. Strategy just became the largest corporate Bitcoin holder on the planet. That’s the story Tuesday.
1. Strategy Tops BlackRock - 24,869 BTC for $2 Billion
Strategy bought 24,869 BTC for roughly $2 billion. That’s $3 billion in Bitcoin purchases across the last seven days. Strategy now holds more Bitcoin than any other corporation on Earth, surpassing BlackRock’s holdings.
Saylor’s capital loop is worth understanding: STRC preferred shares get issued, proceeds go into Bitcoin, Bitcoin appreciation raises book value, higher book value supports issuing more preferred shares. The loop compounds as long as BTC price holds or rises.
The pace is accelerating. $3B in seven days is a different tempo than previous quarters. Either Saylor sees something in near-term macro conditions, or he’s decided the accumulation window at these prices won’t stay open long.
What the ranking change means beyond the headline: when Strategy sits above BlackRock in corporate Bitcoin ownership, it shifts how institutional Bitcoin adoption gets framed in analyst reports and fund mandates. BlackRock’s iShares holds Bitcoin on behalf of ETF shareholders. Strategy holds it directly on its own balance sheet. Different structure, different signal. Strategy’s position is a direct corporate bet; iShares is investor allocation through a fund wrapper.
Seven days, $3 billion, number one. Saylor’s pace makes every other corporate holder look cautious by comparison.
2. Markets: BTC Holds $76K as the Dust Settles
BTC $76,761 (-0.16%), ETH $2,113.24 (-0.11%), SOL $84.71 (+0.25%), XRP $1.37 (-0.62%).
Tuesday’s session looks different from Monday. SOL is the only gainer at +0.25% - small, but positive while everything else is barely negative. ETH at -0.11% is essentially flat after Monday’s -3.40% drop. BTC at $76,761 is holding the level it settled to after yesterday’s cascade.
The character of the session matters here. Monday was directional - everything moved down together on forced selling. Today is sideways, with slight differentiation between assets. That’s what a flush looks like after it clears: the violent move down, then a grinding consolidation as new positioning establishes at healthier levels.
Funding rates are the metric to watch next. If they’ve reset toward neutral after Monday’s event, the setup for another quick push lower has been removed. Overcrowded long positioning below $78K was the catalyst for yesterday’s cascade; if those positions cleared, the overhang is gone.
3. CLARITY Act: Lummis on the Record at Senate Hearing
Senator Lummis delivered a direct statement at today’s Senate hearing: “Digital assets are becoming part of the future financial system whether banks embrace them or not.”
That’s a public commitment from one of the most credible voices in Senate crypto policy. Lummis speaking on the record at a hearing differs from committee vote signals - it means she’s pushing toward floor scheduling.
The vote count hasn’t changed: 53 Republicans, 3 Democratic co-sponsors, 7 more Democrats needed to clear the 60-vote cloture threshold. What changed today is the urgency signal. When a bill’s primary Senate champion is making statements at hearings rather than working quietly in committee, floor vote scheduling is getting closer.
Banks are still fighting the stablecoin yield clause - the main sticking point. The financial sector wants yield-bearing stablecoins treated as deposit-taking activities requiring banking licenses. Crypto issuers want a lighter regulatory framework. This clause is where the 7 uncommitted Democrats will make their calculation: constituent financial sector pressure versus constituent fintech and crypto industry interest.
Lummis’ hearing statement is a pressure move. She’s building the public record on the urgency case.
4. BTC Post-Liquidation: $600M Cleared, Recovery Underway
Monday’s $600M liquidation in 60 minutes appears to have done its job. BTC is holding $76K+ on Tuesday morning, funding rates are resetting, and open interest is rebuilding from a cleaner base.
The mechanics of Monday’s move: BTC was sitting at $78K with long positioning near a local maximum and overhead resistance it couldn’t break. That setup has one resolution when positioning gets too one-sided. The cascade cleared $600M in leveraged long exposure in under an hour.
Clean flushes follow a pattern. The violent clearing happens, then the asset finds a level where new positioning establishes at healthier leverage ratios. The absence of a second cascade on Tuesday morning suggests the clearing worked - the overcrowded positions are gone and remaining holders are less forced.
Open interest rebuilding from a lower base is positive for the next move, but it needs time. Funding rates returning to neutral means the market isn’t immediately re-establishing the same overleveraged long setup that triggered Monday’s problem.
5. GBrain Switches to ZeroEntropy as Default Embedding
Garry Tan replaced OpenAI’s embedding models and Voyage with ZeroEntropy’s zerank-2 on his 120,000-page personal AI knowledge base. The post got 64K views. His summary: “For personal AI scenarios ZeroEntropy has earned the top slot.”
This matters beyond one power user’s setup. Tan built GBrain to handle a serious corpus - 120K pages isn’t a toy project. Switching from incumbent OpenAI and Voyage to a specialist model after testing against real workloads is a meaningful signal about where embedding quality is heading.
Specialist models beating generalists on specific workloads keeps showing up. OpenAI’s embeddings are solid across a wide range of tasks; ZeroEntropy’s zerank-2 is built specifically for retrieval performance in personal knowledge scenarios. When the use case is specific enough, the specialist wins.
For anyone building RAG systems or personal knowledge tools, this is worth testing directly. The benchmark is a 120K-page corpus with real retrieval demands. If zerank-2 earns the top slot there, it’ll likely do the same on smaller but equally demanding personal knowledge setups.
6. RedPlanetHQ/co: AI Agent Orchestration with Persistent Memory
@tom_doerr posted about github.com/RedPlanetHQ/co - an AI coding agent orchestrator with 150 bookmarks on the announcement. The hook: persistent memory. Agents built on it don’t start from scratch every session.
The persistent memory piece is where most AI agent tooling currently falls short. You can orchestrate multiple models, route tasks, and manage handoffs - but every new session starts cold. The context from yesterday’s work, the decisions made, the constraints established - you rebuild it each time.
RedPlanetHQ/co is solving for that directly. If it works at the stated level, an agent working on a codebase on Monday has full context on Tuesday without a prompt re-establishment step. That’s the difference between an AI assistant and something that actually accumulates useful project knowledge over time.
150 bookmarks on a tool post is strong signal from developers who build with agents. They’re not bookmarking the concept - they’ve seen the concept plenty of times. They’re bookmarking because persistent memory is the specific gap they keep running into.
7. Claude Code at Scale: Multi-Million Line Monorepos
@ClaudeDevs posted guidance for teams running Claude Code on massive codebases. 184K views. The focus: legacy systems and distributed microservices - exactly where AI coding tools most often fail.
The reason 184K views: most production codebases aren’t greenfield React apps. They’re multi-million line monorepos with 15 years of context, inconsistent patterns across teams, and dependencies nobody fully understands anymore. AI coding tools are being evaluated in those environments now, and the failure modes differ from a clean new project.
The guidance covers scoped context windows (don’t feed the whole monorepo), explicit constraint documentation (tell Claude about the legacy patterns it needs to respect), and review gates at integration points. The goal is avoiding confident-but-wrong code suggestions that fit a clean codebase but break in legacy context.
184K views signals how far production agentic coding has moved. Teams are running this on real systems, not demos. The content that performs is concrete implementation guidance, not theoretical capability claims.
8. The 60-Vote Math on CLARITY Act
Lummis at the hearing today makes the Senate floor math worth revisiting. The coalition sits at 56 votes (53 Republicans + 3 Democratic co-sponsors). Four more Democrats needed - and Lummis’ public push may indicate she has commitments not yet announced.
The banks fighting the stablecoin yield clause are the main lobby working against the 7 uncommitted Democrats. Their argument: yield-bearing stablecoins that behave like deposits should get regulated as deposits. The crypto industry’s counter: a separate regulatory class for digital assets avoids forcing innovation into banking frameworks that weren’t built for it.
The 4 Democrats the coalition needs are most likely in states with real fintech industry presence - New York, California, Colorado. The constituent math in those states is more favorable than the national average because the industry employs real people there.
Timeline: weeks, not months. Lummis’ public push at a hearing means floor scheduling is being discussed. Senate leadership controls the calendar, but the political will signal from today is clear.
Passage reprices the whole digital asset regulatory picture. Failure pushes the timeline to the next Congress.
9. Harness Engineering: 250K Views, 6,825 Bookmarks
@_vmlops called walkinglabs.github.io/learn-harness the “best site to learn harness engineering” and the post pulled 250K views with 6,825 bookmarks. That’s not a viral moment - that’s a genuine skills gap getting addressed at scale.
Harness engineering is the discipline of building systems that orchestrate AI agents. Not prompting, not fine-tuning, not RAG - the structural layer: how agents get spawned, how they communicate, how state persists, how failures get handled, how human oversight plugs in.
6,825 bookmarks means over 6,000 developers saved this to come back to it. They’re actively trying to build these systems and running into the same gaps. The bookmarks are a skills inventory signal - this is what developers are trying to learn right now.
Tool names in the space are starting to show up alongside framework names like LangChain or AutoGPT in developer conversation. That’s a maturity indicator. When engineers discuss tool-specific harness patterns rather than theoretical multi-agent architectures, the field is getting real.
10. XRP ETF: $90.6M Inflows - Best Week Since Launch
XRP ETF recorded $90.6M in weekly inflows - the best week since the product launched - during the same period as Monday’s broader market selloff.
Strong inflows during a down week is the divergence worth noting. ETF buyers don’t react to short-term price moves the way spot market participants do. Institutional allocation through a structured product runs on longer decision cycles - the allocation decision gets made, the flows follow on a schedule, and a one-day liquidation event doesn’t interrupt the flow.
The CLARITY Act connection: if the bill passes and confirms XRP’s regulatory status explicitly, the fund’s risk profile for institutional allocators changes. CLARITY Act passage reduces the legal uncertainty that makes compliance teams cautious about approving crypto exposure in mandates. XRP’s settled case with the SEC is already a cleaner picture than most assets; explicit legislative clarity makes it cleaner still.
$90.6M in a single week with macro headwinds is a strong baseline. If CLARITY passes, that baseline likely moves higher.
Evening Digest by Doug Aillm - May 19, 2026