Weekly Roundup 06/12/26 (Strategy survives, Zcash Orchard bug, the thin model hypothesis) (EP.725)

Strategy did not sell Bitcoin as many had speculated, instead purchasing 1,500 BTC and topping up reserves through common stock sales exceeding 100 million dollars. Michael Saylor paid off 1.5 billion dollars in convertible debt three years early, cutting cash runway from roughly two years to about six months for servicing preferred shares. The argument is that Strategy is protecting STRC and keeping its dividend alive at the expense of common shareholders, a conclusion reportedly reached independently by Claude when fed the available information. Bitcoin's price decline over the period was attributed to multiple factors including Strategy's buying pause, quantum computing fears, macro uncertainty, and IPO liquidity drain, with the primary driver identified as Strategy halting large purchases rather than any actual selling.

A security researcher using Claude discovered a critical inflation bug in Zcash's Orchard shielded pool. The bug is considered especially dangerous in a privacy coin because auditability is traded for privacy, making it impossible to confirm with certainty whether exploitation occurred, though no one believes it was. Zcash plans to use a turnstile approach, pulling funds out of the shielded pool and confirming the correct amount exits. A similar inflation bug affected Zcash around 2018 or 2019 and was resolved the same way, and Monero has had a comparable prior incident. The bug had a material negative impact on Zcash's price and was cited as evidence that AI tools are lowering the barrier for regular researchers to find serious vulnerabilities.

A La Liga club believed to be Osasuna used an intermediary to place a bet against themselves on Kalshi, effectively buying insurance against relegation. Susquehanna was on the other side of the trade. The club was not relegated, so the insurance went unused. The episode illustrates that prediction markets need corporate hedgers, not just retail speculators, to grow meaningfully, and firms like Susquehanna are expected to increasingly offer over-the-counter prediction market products directly to brokerages without settling back to Kalshi or Polymarket.

The CFTC proposed rules that would allow regulators to reject prediction market contracts vulnerable to manipulation, potentially banning contracts on war, terrorism, assassination, and certain sports. One speaker opposed banning war contracts, arguing they are among the highest signal contracts available and that insider trading by military personnel should be addressed through existing enforcement rather than contract bans. Sports betting prediction market rules are expected to face legal challenges potentially reaching the Supreme Court. Polymarket has sued Minnesota, New Mexico has sued Kalshi, and Illinois wants to impose a per-wager tax on prediction markets. Separately, Polymarket believes Kalshi is spying on it and maintains a dossier of roughly a dozen incidents showing suspicious similarity between the two firms. Kalshi called the allegations sad and borderline delusional.

Japan's three largest banks, MUFG, SMBC, and Mizuho, are forming a council for joint stablecoin issuance by March, expected to be yen-denominated. Non-US banks could issue interest-bearing dollar stablecoins offshore outside the scope of US legislation like GENIUS, and offshore interest-bearing stablecoins from reputable institutions would almost certainly gain adoption. Janus Henderson's subsidiary took a stake in Athena's ENA token and will use staked USD for cash management, and Coinbase recently took a similar stake. Morgan Stanley announced it will refer wealth clients to Galaxy to lend crypto in exchange for shares of spot crypto ETPs including Morgan Stanley's own Bitcoin trust. A JP Morgan representative stated tokenized deposits have not seen significant demand and framed this as an indictment of blockchain, despite JP Morgan previously championing tokenized deposits. One speaker argued stablecoins are more fungible and secure than tokenized deposits, and that critics are arguing against a hypothetical world rather than engaging with existing regulated products.

Humanity Protocol, a palm scanning identity project, was exploited for 32 million dollars via compromised private keys, and its token fell 80 percent on the news. The DAT sector is trading at significant discounts, with some receiving exchange notifications for failing minimum listing requirements, some missing SEC filing deadlines, some already undergoing reverse splits, and delistings expected. In a world of abundant ETFs for any coin, the rationale for DATs is considered weakened.

Tokenized inference compute was described as non-fungible because different providers vary in uptime, latency, location, networking, GPUs, and models, making it unclear who the natural seller would be in such a market. Cheaper per-unit inference cost is expected to trigger the Jevons effect, increasing aggregate compute consumption. Commoditization of AI models could dent the credit quality of large labs and cause capital market panic around data center investment pauses. AI model companies were predicted to face heavy regulation with government-dictated pricing, similar to utilities, rather than capturing all economic activity. Trump is in discussions with labs about whether taxpayers should hold an equity stake in frontier AI companies, and frontier model providers were described as unable to be allowed to go bankrupt and sell assets to China for national security reasons.