TLDR Crypto 2026-07-21

Strategy sold 2.7 million MSTR shares for roughly $263.5 million between July 13 and July 19 without buying or selling any bitcoin, pushing its USD reserve up $225 million to $3.225 billion. The firm's holdings remain at 843,775 BTC (~$54.7 billion, about 4% of bitcoin's supply cap) bought at an average price of $75,476, carrying roughly $9 billion in paper losses. CEO Phong Le said Strategy would only reassess debt-related risk if bitcoin fell to $8,000-$10,000, while MSTR stock closed the week down 4% at $94.85, off 38.6% year-to-date.

AZ-COM Maruwa, Amazon's largest delivery partner in Japan, will pay 2,300 carriers and independent contractors in JPYC, a yen-backed stablecoin, in what the companies describe as Japan's first large-scale corporate stablecoin payroll deployment. The rollout reflects Japan's amended Payment Services Act, which since 2023 has permitted licensed entities to issue yen-pegged tokens, giving JPYC a regulatory footing absent in most other major economies. The partnership moves stablecoin payment rails into recurring enterprise payroll at logistics scale, a use case distinct from the crypto-native trading and DeFi activity that has dominated prior stablecoin volume metrics.

Hyperliquid will roll out permissionless deployment for HIP-4 outcome markets, launching on testnet first, letting deployers instantiate validator-approved templates rather than requiring validators to create markets directly. Deployers must stake 500,000 HYPE, locked for six months, with stakes subject to slashing for poorly defined markets, bad settlement, or missed settlement deadlines, and can charge fees up to 50%. Each deployer starts with an allocation for 100 outcomes. HIP-4 launched in May and drew about $100 million in volume in its first month.

EIP-8130 is a protocol-native account abstraction standard developed with Optimism, Coinbase, and WalletConnect set to ship with the Cobalt hard fork in September and extend to all OP Stack chains later this year. By embedding AA at the protocol layer rather than running it as an ERC-4337 overlay, EIP-8130 cuts USDC transfer gas from 125,000 to 46,000 gas (63%) and calldata from 1,090 to 180 bytes (83%), while delivering batch calls, gas sponsorship, session keys, sub-accounts, cross-chain portability, and quantum-ready key rotation as native protocol features. Standard EOA accounts gain sponsorship and batching without code changes. Existing ERC-4337 smart accounts require a one-time migration to access the lower cost structure. Developers can test EIP-8130 on Base Vibenet, an ephemeral devnet available ahead of the Cobalt activation.

"Inference farming" may be the successor to perps yield farming, as decentralized AI inference platforms will accrue revenues to token holders the way Hyperliquid channeled trading fees to HLP/HYPE participants. Open-source models like Kimi K3 reaching frontier performance at reduced cost are the catalyst, with decentralized inference networks able to price 30%-plus below centralized providers as a result. Farmers participate by renting GPUs and contributing compute to these networks. Early project picks named include Dolphin AI, Bittensor subnet 53 (Engy), and AntSeed AI. A quoted thread identifies settlement standardization, reference pricing, and compute derivatives (futures and forwards) as the largest unaddressed gaps across the inference capital markets stack.

Crypto's foundational infrastructure problems are solved, with L1s, L2s, DeFi, and stablecoins now functioning as mature settlement rails, and the next opportunity lies in building financial infrastructure for autonomous machine actors rather than humans. The firm's "machine economy" thesis holds that AI agents, warehouse robots, and automated research systems require open, programmable, permissionless payment rails that legacy financial infrastructure was not designed to provide. Emerging agent payment standards including x402, MPP, and AP2 are maturing alongside this demand. Coinbase, Robinhood, and Binance have each shipped agentic trading infrastructure in recent months.

Robinhood and Coinbase already run significant core operations on Ethereum L2s, and this represents the opening move of a broad corporate onchain adoption pattern that platforms like Airbnb should follow. The proposed Airbnb blueprint includes a proprietary Ethereum L2, a platform stablecoin for host and guest payments and deposits, and tokenized ABNB equity natively integrated into host and guest rewards, making the stock function as a dual-purpose utility token. A cross-corporate onchain reputation network would feed underwriting for adjacent lending and mortgage products, with host transaction history serving as a proxy for borrower creditworthiness. Corporate crypto/onchain functions are the next standard internal department, comparable to IT or Legal, with Ethereum L2s and RWA infrastructure as the dominant infrastructure layer and ETH as a primary demand beneficiary.

Michael Saylor published a 110-point essay, "110 Reasons BIP 110 Is a Bad Idea," arguing consensus rules shouldn't judge the purpose of valid, fee-paying transactions, drawing over 840,000 views. BIP-110 is a temporary one-year soft fork restricting data-heavy transactions, built on Bitcoin Knots software from backer Luke Dashjr, with a mandatory signaling window opening around August 7. Miner support currently sits at just 0.86%, far below the 55% threshold needed for early lock-in. If signaling stays low, BIP-110-enforcing nodes risk splitting onto a minority chain when the rules take effect around September 1.

Distribution over features, compliance as a core requirement, banking businesses over consumers, and building multiple revenue streams beyond interchange from day one are some of the many ways to quickly, effectively, and sustainably scale a neobank.

Former New York Governor Andrew Cuomo, who has advised OKX on US regulatory strategy since 2023, joined the exchange's board to support its US expansion and will co-chair the OKX-ICE joint venture.

14 onchain businesses now generate $200M+ in two-year revenue, up from almost none in 2021.

A new proposal would block BTC transfers to quantum-vulnerable addresses, phase out legacy signature schemes over five years, and explore a recovery mechanism for users who don't migrate funds in time.