The full faceted feed. Filter by source, type, nature, vertical, time window.
All takes are our summaries. Tap View on Xfor the analyst's original words.
Nikshep argues NEAR is positioned as the coordination layer for the emerging agent economy, citing Anthropic's advances with Claude Opus 4.6 enabling autonomous agents to sustain long-horizon tasks and execute 100+ tool calls across teams of subagents—capabilities that require a blockchain to securely coordinate AI agents at scale.
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.
Bull pitch on NEAR at $1.28 / $1.67B mcap, ~94% off ATH. The setup nobody is pricing in — vesting fully completed Oct 12 2025 (no more cliff unlocks; the 4-year supply overhang is gone), inflation halved 5%→2.5% Oct 30 2025 via protocol upgrade v81, 70% of fees burn permanently (with sufficient activity NEAR is structurally net deflationary), House of Stake/veNEAR governance went live.
Founder asymmetry: Illia Polosukhin is one of the eight co-authors of Attention Is All You Need — the Transformer paper that powers GPT-4/Claude/Gemini/Llama. Co-founder Alex Skidanov was Engineer #1 at MemSQL, a two-time ICPC World Finals medalist, designed the only sharded distributed DB that worked at scale. The market is currently valuing their company at less than the seed-round valuation of half the AI agent startups in San Francisco.
Real thesis: agents can't use Visa. When autonomous agents replace humans as users, the entire payment stack breaks — weekend bank hours, KYC for every counterparty, days-to-settle, not programmable. NEAR has shipped more agent-native infrastructure than any L1 competitor:
Catalysts: Bitwise + Grayscale spot ETF filings (Grayscale to convert GTAO Trust on NYSE Arca with Coinbase Custody), NVIDIA Inception membership, Brave private-inference partnership, fee switch revenue.
Honest bear case: $117M TVL is small (RHEA Finance is concentration risk). Governance controversy — Chorus One opposed the inflation halving as forced through despite a failed initial governance vote. Memecoin overhang on AI/crypto narrative. Execution risk vs Solana's deeper liquidity and consumer DeFi. ETF filings ≠ approvals.
Asymmetry: at $1.67B with vesting done, halved inflation, fee burn, ETF filings in flight, $13B+ routed cross-chain volume, transformer co-author at the helm — downside bounded by L1 floor, upside multi-X if the agent thesis lands.
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:
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.
Victor opens a $CARDS allocation on the thesis that Collector Crypt — a Solana protocol tokenizing PSA/PWC-graded trading cards as NFTs — is structurally mispriced at a $23M circulating cap on $584M annualized revenue. Q1'26 gross revenue: $146M. Top-10 Solana app by revenue, sitting in the same band as Phantom and Jupiter, but the only one capturing demand from outside crypto (eBay/conventions/local card shops, a $25B global TCG+sports market growing to $43B by 2031). Existing rails take 10–30% per transaction; CC charges <2% with instant settlement.
Demand signals: weekly volume scaled 7–8x in 15 months during a crypto drawdown. Gacha machines were stocked only 29% of hours one recent week — the platform is supply-constrained, not demand-constrained (operationally easier to fix than user acquisition). Whale concentration is meaningful (top 3.3% of users → 81.5% of revenue, 58 wallets >$1M lifetime spend) but less concentrated than Hyperliquid at the same stage. Pyramid critique fails: 10K+ small users prove funnel reach, 1.8K mid-tier wallets are tomorrow's whales.
Catalysts: $1,000 Pokémon packs are now the largest weekly contributor (zero in late '25), $250 One Piece launched in early '26 already top-3, $100 Sports just live, fiat on-ramp (cards/Apple Pay/bank, USDC settled via Coinflow) just shipped — opening the much larger pool of card collectors who'll never own crypto. Marketplace V2 ships May. Disclosure: Victor is starting an allocation.
One Sunday email with the week's most interesting takes — handpicked, not algorithmic. Skip the timeline.
Sundays only. One email a week. Unsubscribe with one click.
Bull thesis on Bittensor / TAO at ~$3B mcap. Frame: "TAO 2026 = ETH 2016 = BTC 2013."
Core mechanic: Bitcoin paid miners to produce hashes that secure the network but are otherwise worthless. Bittensor pays miners — data scientists, ML engineers, AI researchers — to produce useful AI work. Validators score outputs via Yuma Consensus; TAO flows to whoever produces the most valuable work. Network is organized into 128+ subnets, each focused on a specific task (trading signals, LLM training, computer vision, code generation, financial forecasting). Some subnets generating millions in revenue, with Intel and PwC partnerships.
Tokenomics mirror Bitcoin: 21M fixed supply, no pre-mine, no VC allocation. First halving Dec 14 2025. BTC price went 83x in the year after its first halving in 2012.
Smart-money signals: Barry Silbert / DCG launched Yuma Group dedicated to accelerating Bittensor. Grayscale filed Form S-1 to convert GTAO Trust into a spot ETF. Stillcore Capital (Mark Jeffrey, Jason Calacanis, Rob Greer) targeting $1T mcap by 2030, aiming to own 1% of all TAO. Unsupervised Capital projects $4,800 by Dec 2027 (19x), bull case $10,800 — and that's before Covenant-72B, Jensen mentioning Bittensor, and PwC's formal alliance.
Subnet-level conviction picks:
Frame: Bitcoin = money. Ethereum = apps. TAO = intelligence. The gap between what TAO has built and how it's currently priced is one of the most asymmetric opportunities in crypto.
Magic introduces BAM's Maker Priority Plugin, enabling sub-slot deterministic execution for onchain market-making on Solana. The plugin addresses a fundamental limitation in current Solana market-making infrastructure that isn't about AMM design or throughput constraints. Magic positions this as solving a subtle but critical gap in how onchain market-makers can operate.
Baheet argues Sui's object-centric architecture, Move language, 390ms finality via Mysticeti, native DeepBook v3 CLOB, and March 2026-launched USDsui stablecoin create an underutilized technical foundation for prediction markets as the category scaled to $20-27 billion monthly volumes across Polymarket and Kalshi in 2026. While Polymarket's VP of Engineering acknowledged infrastructure strain from rapid traction—citing on-chain latency, transaction cancellations, and CLOB stability issues—Sui remains absent from the dominant prediction market apps, presenting a first-mover opportunity for builders prioritizing high-frequency scalar markets and institutional settlement over ecosystem maturity.
Sam argues Solana's perps problem runs deeper than liquidity—the chain lacks execution guarantees market makers need for tight quotes, while Hyperliquid processes 5-10x Solana's entire perp volume. Bulk's answer is a validator-native sidecar network handling matching and risk separately from Solana's leader-based execution, paired with a SPAN-style portfolio-aware risk engine that cuts margin requirements 70%+ on hedged books—the institutional standard CME has used for decades but no live crypto venue currently offers. The model preserves composability by keeping collateral productive on Solana while supporting trades, with mainnet targeting this half.
Solana hosts crypto's deepest retail user base but has ceded perpetual futures dominance to Hyperliquid, which runs 5 to 10x the volume of Solana's entire perps complex. Sam Schubert attributes this to Solana's general-purpose design lacking the execution guarantees perp makers need—non-deterministic ordering, opaque fees, and rotating validator leaders every 1.6 seconds make quoting impractical. Three new protocols (Phoenix Perps, Bulk, Bullet) are attacking the execution gap with different approaches, but closing that gap may not matter if Solana can't convert its memecoin-focused retail base into active perps traders.