Browse all takes

Uniswap v4 Hooks transform AMM pools from fixed rules into programmable infrastructure, enabling pools to execute custom logic before and after swaps. 0xMedia highlights uPEG and Slonks as breakthrough examples: uPEG generates on-chain SVG unicorn images from swaps themselves, while Slonks uses a Hook as fee collector to fund buying and voiding NFTs tied to CryptoPunks, replacing opaque token taxes with pool-layer mechanics. The trade-off is that v4 Hooks eliminate safety by default—they can hide fees, enforce transfers, or contain malicious logic, requiring new market literacy to distinguish safe implementations from exploitative ones.

Guy argues that finance has largely escaped the digital transformation that reshaped other industries, with institutions still dependent on fragmented systems and constant reconciliation. Blockchains solve this by creating a Schelling point for counterparties to agree on shared state without trusting a central controller, addressing practical Wall Street concerns around counterparty risk and fair ordering. As financial institutions adopt blockchain infrastructure for digital assets, they'll inadvertently inherit crypto's composability ethos.

Mesky explains HIP-4 and BTC outcome markets on Hyperliquid as tools for individual traders to buy mispriced probabilities, positioning them as binary options without the casino dynamics. The guide frames outcome markets as practical instruments for trading specific probability outcomes rather than speculative gambling.

Yiannis argues most commentary on Canton Network misses the fundamental point because observers are trying to understand it through the wrong lens rather than from first principles.

Pink Brains explains that Hyperliquid's HIP-4, which launched May 2nd with a daily BTC binary as its first mainnet market, functions as an options layer rather than a prediction market. The distinction matters for understanding the protocol's architecture and trading mechanics, though the full implications require examining how this positioning affects $HYPE's ecosystem development.

Most people's mental model of Tether is 3-5 years stale. Here's what it actually is now: $10B profit in 2025 with ~300 employees ($33M/employee), $122B in direct US Treasuries (more than Germany), holds 96K BTC + 140 tons of gold, zero external investors, zero transaction fees on secondary USDT transfers. Business model = world's largest money market fund that keeps all the yield, not a payments company.

Scale: 550M+ estimated users globally. 2025 USDT volume = $13.3T onchain, but McKinsey pegs identifiable real payment activity at ~$390B annualized — the "value moved" gap is real. The product isn't a transfer mechanism, it's a savings account in countries where local rails are 20% efficient (Argentina, Nigeria). Ardoino's framing: US financial system is 90% efficient, stablecoins push it to 95%; in emerging markets where efficiency is 10-30%, USDT pushes it to 50%. The 5% margin game in America doesn't interest him.

Three layers to the company now:

The money machine — yield-on-float economics protected by Tether's organic distribution. Less than $10M total marketing spend 2020-2024. Parabolic 2020 growth came from Latin American black-market dollar rails moving onchain when COVID lockdowns shut physical kiosks.

Bifurcation strategy — USA₮ (federally regulated, Anchorage-issued, Cantor-custodied, run by the former White House Crypto Council director Bo Hines) for US institutional onshore. USD₮ for offshore monopoly. USD₮'s zero-yield position is monopolistic offshore because users have no better alternatives. USA₮ can't win on margin ("race to the bottom"); has to win on programmability + Tether's distribution.

Operating conglomerate — $20B portfolio increasingly taking control: 70% of Adecoagro (board overhaul, Sartori as Executive Chairman), 30%+ Be Water, board seat at Gold.com, plus physical bodegas / kiosks / phone-credit shops across LATAM/Africa/Asia. Tether owns the literal cash-to-crypto on-ramps in emerging markets, bypassing banking systems entirely.

Real risks: rate sensitivity (rate cuts compress the float, profit already dropped from $13B to $10B in 2025), TRON dependency (44% of supply, $82B), the persisting audit gap (no Big Four; new CFO from LetterOne hired for "contentious audits"), USDC overtaking USDT in adjusted volume, opacity-of-USD₮ contaminating USA₮ by association.

But the volume flip doesn't translate into a profit threat: Circle surrenders ~60% of revenue to distribution partners (Coinbase took $900M+ in 2024). Tether owns its distribution organically and is now physically buying more of it. Tether's $10B profit dwarfs Circle's $1.7B revenue by an order of magnitude. They're playing different games. The right comparison isn't Circle or Paxos — it's Berkshire Hathaway (yield-generating float funding a diversified conglomerate) crossed with Visa (settlement rails).

Translation + commentary on Bittensor founder Jacob Steeves's Tsinghua University talk. Cameron walks through Jacob's framing of "incentive computing" as the universal pattern behind both Bitcoin and AI. Five-step argument:

(1) One pattern underlies every powerful adaptive system: state · objective · feedback · adaptation · loop. AlexNet 2012 broke MNIST not by hand-coding what digits look like, but by letting the network self-adapt to a target. The same loop describes RL, genetic algorithms, slime molds finding shortest paths through mazes, river deltas, the structure of leaf veins.

(2) Bitcoin is the first production-scale implementation of this pattern — not as money, but as a self-adaptive computer that produces hashes. The numbers are absurd: 1000x the compute of America's six largest cloud providers combined, 10²¹ hashes/sec, 23GW continuous power (Thailand-scale). 700-9000x more efficient at producing hashes than centralized cloud — because it's borderless, always-on, autonomous, and permissionless. Bitcoin is the world's largest supercomputer, optimized purely for hash production.

(3) Incentive computing generalizes the pattern by replacing "reward = a number in a computer" with real money. ML's reward signal can't pay 200 countries' worth of contributors; Bitcoin's can — that's why the entire planet became a mining network. But hashes are useless outside Bitcoin. The question is whether the same mechanism can mint anything.

(4) Bittensor is the generic version — replace "miners produce hashes" with "miners produce any useful work": storage, compute, ML models, gradients, data, robotics. Validators score, network mints. PyTorch for incentive computing.

(5) Five proven examples already running on Bittensor:

• SN62 Ridges (SWE-Bench coding agents) — top miner makes $60K/day. The agent that beat Claude/OpenAI on SWE-Bench was 7,000 lines written by an unknown person. "An AI lab with no engineers — it doesn't define how to solve the problem, it only defines the incentive."
• SN3 τemplar (cross-internet collaborative pre-training) — successfully trained a 70B-parameter model across the open internet. Has never been done before. Cameron notes the founder later "ran away" — full piece coming.
• GPU markets (SN51 Lium, SN4 Targon) — borderless permissionless GPU rental → world's lowest GPU prices.
• SN64 Chutes (open-source inference) — #1 open-source provider on OpenRouter, 9.1T tokens. Briefly served more DeepSeek queries than DeepSeek itself.
• Robotics + long tail — drone simulation, US stock signals, sports betting, drug discovery, weather forecasting, quantum compute, commodity trading.

dTAO (live since Feb 2025) makes the network self-referential — subnets compete in capital markets for emission allocation. The market itself decides which incentive mechanisms get the next round of TAO.

The deeper point: AI is being captured by a tiny number of closed labs (OpenAI, ~3K employees, you'll never own any of it, your data goes who knows where). Incentive computing distributes ownership and makes the rules visible. Anyone can enter, contribute, and own a piece — even if Bittensor isn't the project that wins, the shape of the AI economy will change because of this idea.

Mesky frames HIP-4 not as a Polymarket clone but as a missing payoff layer for Hyperliquid: bounded, dated, fully-collateralized outcome contracts that settle at a date or event with no leverage and no liquidation engine. Where spot trades ownership and perps trade direction, HIP-4 trades states of the world — turning event risk into a composable financial object on the same execution engine that already prices crypto.

The real bull case is not "capture prediction-market volume" (~$240B est. 2026, per Bernstein). It's that HIP-4 expands the addressable market into short-dated convexity and event hedging — analogous to 0DTE options, which now do ~59% of SPX volume. At a 7 bps base spot-taker fee on chargeable close/settle notional, $25–100B/mo of HIP-4 flow becomes one of the platform's most material revenue lines.

Strategic edge: Hyperliquid isn't bootstrapping a venue — it already has $183B/30d perp volume, $643M annualized revenue, and the maker base. HYPE captures value through (1) Assistance-Fund buyback/burn from incremental fees, (2) staking-collateral demand if HIP-4 deployers require staked HYPE like HIP-3 (500K HYPE), (3) staking discounts (up to 40%), and (4) USDH demand as the native unit of account for event risk.

Mesky's prescription: don't out-Polymarket Polymarket. Sequence rollout toward crypto-native, recurring, hedgeable templates (BTC weekly thresholds, Fed decision markets, token unlock outcomes) where market makers can build inventory — not viral one-offs. Repeatability beats virality.

Real risks: ambiguous resolution, regulatory perimeter (CFTC v Wisconsin, Brazil's blanket ban), insider trading (DOJ Polymarket case, Kalshi candidate suspensions), long-tail spam, and perp cannibalization. Mainnet HIP-4 spec/fees/deployer rules still aren't formalized in the Hyperliquid GitBook.

Harry's thesis: the real "DeFi meets TradFi" story isn't JP Morgan on a blockchain — it's an emerging infra layer that lets neobanks ship "earn" and "savings" features backed by DeFi/RWAs without becoming DeFi engineers themselves. Early DeFi was monolithic (Aave, Compound, Maker each owning UI + liquidity); the new layer abstracts chain routing, normalizes onchain liquidity + tokenized funds, and handles KYC/AML/1099s at scale.

Reference architecture: @blend_money offers white-label earn infra where each user gets their own self-custodial smart-contract account (no co-mingling, funds remain accessible even if Blend disappears), purpose-built earn pages with T-bill yields + DeFi lending, risk ratings translated for compliance officers, and out-of-the-box reporting. The unlock for neobanks: "we'll handle the chains, protocols, bridges, KYC vendors and reporting — you focus on customers."

Market context: DeFi TVL hit $237B in 2025, RWA market grew 380% in 3 years, 400M+ people use neobanks (projected $6.5T deposits by 2030), Standard Chartered projects RWA could hit $30T by 2034. End users want a savings-account experience that pays better — they don't care that crypto is the substrate. The infra companies that absorb the complexity and "let someone else put their logo on the home screen" are the leverage point binding chains, protocols, and consumer trust.

Modular Capital argues the U.S. power deficit of 50 GW through 2028 makes Bitcoin miners' converted data centers the lowest-cost path to AI infrastructure, with energized capacity priced at $30–50/MWh versus $80–150/MWh for alternatives like nuclear and fuel cells. Bitcoin miners holding 10+ GW of approved grid interconnection now command structural economics worth $5–10M per gross megawatt in HPC colocation deals, with recent transactions clearing at $1.24–$2.17 per critical IT watt annually and 80–97% EBITDA margins, while the sector at $4M equity value per approved MW implies only 50% conversion pricing, leaving asymmetric upside for operators with large approved portfolios and credible execution track records.