6 min read · 12 sources

The NYSE has spent the past year testing Avalanche technology and probing its economics for a tokenized equities platform, but has yet to name a settlement chain.

NYSE Spent a Year Testing Avalanche for Tokenized Equities

Source: theblock.co ↗

The New York Stock Exchange has spent the past year testing Avalanche technology and probing its economics as it builds a platform for onchain trading and settlement of tokenized US equities and ETFs. Ava Labs President Charley Cooper described a “close working relationship” with NYSE parent ICE but stopped short of saying Avalanche had been selected. ICE’s Michael Blaugrund said the firm remains “very engaged” in evaluating blockchain platforms.

For engineers, the interesting bit is what “testing economics” means operationally. Settlement finality, validator costs, and per-transaction fees at NYSE scale are nothing like a DeFi app. If Avalanche’s subnets can’t hit the throughput and cost profile ICE needs for lit-order trading, all the relationship goodwill in the world won’t make it production. The fact that this is still a test after a year tells you the bar is high.

Visa Closes the Memecoin Rewards Loophole

Source: theblock.co ↗

Visa is moving to stop payment processors from categorizing memecoin purchases under merchant code 5815, the “digital media” category that lets credit card buyers earn ordinary points or cash back. The change follows The Block’s investigation and a direct warning to Checkout.com, Crossmint’s processor. Visa’s grace period is expected to end next week, after which purchases must be processed under standard crypto transaction rules.

Crossmint already pulled tokens GENIUS and DEGEN from its Apple Pay checkout after The Block questioned their classification as memecoins. New York Attorney General Letitia James’ office said it is looking into the matter. Mastercard has not disclosed whether it will follow Visa’s lead. For anyone running a payments stack, this is a reminder that merchant category codes are a regulatory surface, not just a billing field. If your processor lets you pick the MCC, expect that flexibility to shrink.

First LLM Runs Entirely Onchain via Solana SVM

Source: threadreaderapp.com ↗

The first LLM inference run executed as pure onchain Solana transactions, with model weights, activations, attention calculations, and token outputs each processed as standard SVM program calls. No trusted black-box attestation - the inference is an auditable, verifiable ledger entry. That breaks with most current AI agent and oracle architectures, where inference results are attested after the fact rather than proven in place.

The operational question is cost. Per-inference gas on SVM is nontrivial, and the thread doesn’t pretend otherwise. But if the pattern compresses to practical levels, it changes the trust model for AI agents holding funds or signing transactions. You no longer need to trust an offchain model provider; you can verify the exact computation that produced a given output. That is a meaningful shift for anyone who has ever debugged an agent that hallucinated a transaction.

S&P Global Acquires OpenZeppelin

Source: threadreaderapp.com ↗

S&P Global has acquired OpenZeppelin, extending its ratings, benchmarks, and risk assessment infrastructure into blockchain. Financial terms were undisclosed, but OpenZeppelin brings 900+ security engagements and surfaced 10,000+ pre-production vulnerabilities through its audit practice. The firm’s open-source tooling remains free and publicly maintained on GitHub.

The rationale is straightforward: S&P wants to rate tokenized assets and DeFi protocols, and it needs the security signal that comes from auditing code at scale. For engineers, the risk is strategic drift - will OpenZeppelin’s audit arm stay independent, or will S&P steer it toward rating products the parent wants to push? The acqui-hire of a security team by a ratings agency is a test of whether “independent auditor” survives contact with a commercial parent.

Post-Quantum Security Is Now a Procurement Problem

Source: x.com ↗

Google’s 2026 quantum computing progress report set a 2029 PQ readiness target, converting the migration from deferred planning to current procurement. NIST has finalized ML-KEM (FIPS 203), ML-DSA (FIPS 204), and SLH-DSA (FIPS 205). The “harvest now, decrypt later” attack vector is operational today, with encrypted traffic being recorded against the day a fault-tolerant quantum computer exists.

Blockchain infrastructure faces the same ECC replacement pressure as traditional PKI. ECDSA and EdDSA wallet key material and transaction signatures are exactly the kind of long-lived secrets that will be harvested and broken. The migration concentrates in four areas: hardware security modules and trusted execution environments, crypto-agile embedded devices, machine-to-machine authentication, and the actual signature schemes on chain. If your chain doesn’t have a plan for post-quantum signatures, you are assuming Q-Day is further out than your users’ funds will live.

Bybit Quietly Opens Registration for Chinese Users

Source: threadreaderapp.com ↗

Bybit, the third-largest offshore exchange, has quietly opened registration for Chinese users despite a long-standing ban. Competitors have already done so, and the thread notes the number of Chinese users could be in the tens of millions, causing internal employee panic. A secondary thread discusses contract approval risks in major wallets, highlighting that Binance and KuCoin have unlimited approval limits and failed to clean up BUSD approvals, with a specific mention of Multichain’s event.

The regulatory implications are severe - serving Chinese retail users while pretending not to is a classic enforcement trigger. For engineers, the approval-risk thread is the more actionable item: unlimited token approvals are a footgun that keeps paying out. If you are running a wallet or a dapp, assume every contract you approve will eventually be exploited, and design for revocation.

Blockchain Capital Allegedly Running Quiet Coup at Polymarket

Source: threadreaderapp.com ↗

Rumors allege Blockchain Capital is running a coordinated play at Polymarket, using the same playbook it used at Bluesky. Shah took a board seat after leading the Series A, and Jay Graber was removed from founder Shayne Coplan’s unilateral control over budgets. Dalio’s former Bridgewater Chief of Staff and an ex-DoorDash GM have ring-fenced day-to-day operations. None of the claims have been independently verified.

The pattern, if true, is familiar: a lead investor uses a down round or a governance crisis to insert operators and strip the founder’s control. For anyone building on Polymarket’s prediction market infrastructure, the risk is that governance changes alter fee structures or market resolution rules. Unverified, but worth watching if you have capital at risk in these markets.

Robinhood Chain Collected $4.5 Million in Daily Fees

Source: threadreaderapp.com ↗

Robinhood Chain collected $4.5 million in daily fees while paying Ethereum only about $398 for data posting and proof costs. That is a roughly 11,000x margin between revenue and settlement cost. The chain is capturing the spread between what users pay in fees and what it costs to post data to Ethereum.

For L2 operators, this is the business model working as intended - but it also highlights how much value accrues to the sequencer. If Robinhood can sustain that fee capture while competitors undercut, the moat is distribution, not technology. If not, expect fee compression as other chains copy the model.

When Every Trader Uses AI, Nobody Beats the Market

Source: threadreaderapp.com ↗

When all retail traders use the same LLMs with the same basic prompts, the result is herding rather than efficiency gains. Classical EMH assumes independent actors with diverse information; homogeneous AI agents break that assumption. Momentum strategies then amplify consensus before it forms, creating crowded trades that unwind together.

For quant engineers, this is an argument that alpha will shift to data and execution speed, not better reasoning. If everyone’s model says the same thing, the edge is in who can front-run the consensus. That is a structural change in how retail alpha is made, and it is happening whether or not the market has priced it in.

EtherFi Cashback Claims and Plasma Card Ordering

Source: threadreaderapp.com ↗

First results from ETHFI cashback claims for EtherFi Cash users are in, and the Plasma Card physical card ordering process has opened. Both are consumer features, but they matter for engineers building on EtherFi because they represent real usage patterns - cashback claims stress the claim contract, and card ordering stresses the KYC flow. If either breaks under load, that is signal about the team’s operational maturity.