Personal essays on Web3 strategy, ecosystem bridges, and the patterns I see across 60+ portfolio companies. Distinct from my work on Forbes and Cointelegraph — closer, more current, more personal.
Between August 1 and August 22, 2026, crypto funding trackers reported roughly $8.8 billion. Primary venture rounds into actual Web3 companies came to about $93 million. The gap is the story.
On July 21 Robinhood said its new chain added 240,000 stock-token holders in a month. The number that tells you whether the composability bridge got built is far smaller — and it's the one nobody put in the headline.
In June 2026 Mastercard, Visa and Coinbase all shipped ways for AI agents to pay. The press calls it cards versus stablecoins. I think the real prize is the layer that bridges both rails.
Coinbase launched 1:1 tokenized U.S. stocks on June 16. Robinhood and Interactive Brokers followed. Issuance is now table stakes — composability is the bridge nobody's finished building.
Stripe, Visa, and Mastercard are close to launching a joint stablecoin platform. A field note on what the incumbents co-opting the rail means for founders building payment infrastructure.
In 2026 crypto venture funding split in two: a handful of mega-funds writing nine-figure checks, and a starving long tail raising under $2M from angels. The bridge between them runs straight through revenue.
I've lived through enough Web3 cycles to see the same three patterns repeat. In 2026, two of them are running on schedule. The third has changed shape — and most of the market is still pricing the old version.
Decentralized AI talks a lot about compute and not enough about proof. Verifiable inference and zkML are the trust layer the category has been missing — and the protocols building it are the ones I'd back into 2027.
Intents quietly won the DeFi orderflow war. UniswapX, CoW, Across, 1inch Fusion route a majority of retail volume. The interesting question for 2026 isn't whether intents won — it's what compounds on top.
Two decentralized-AI founders in one week. Different categories, different traction, different teams. Same blind spot — the question neither could answer about model dependency.
Compute DePIN and telco DePIN look similar on a slide. The unit economics tell you they're two completely different businesses — and the winners, timelines, and capital needs don't compose.
Tokenized money-market funds quietly crossed $10B in 2026. The shift in what 'crypto rates' means — and what that does to stablecoins, treasuries, and DeFi yield — is the most underpriced story in RWA right now.
Four LP calls in five days. The questions allocators are asking in Q2 2026 are nothing like the ones from last spring. What's changed — and what it tells you about how Web3 capital is being underwritten now.
EigenLayer crossed $20B, then ate itself. The second act of restaking isn't a TVL race — it's a yield-attribution problem. The protocols I'm watching now look nothing like the ones that won the first.
The AI x Web3 thesis gets pitched as one trade. In 2026 it's three different markets with three different unit economics. I've been wrong about which one matters most — here's where I've landed.
I took three DePIN pitches in 72 hours. None of the decks looked alike. What they shared — and what I almost missed — is what I'm taking into next week's calls.
Tokenized real-world assets crossed $17B in early 2026. The three bridges that actually decide whether this becomes infrastructure or stays a press cycle are liquidity, regulation, and settlement — and only one is half-built.