Stray Dog
EN中文
EN中文

Prices are Hyperliquid perpetual futures: contracts that track each stock and trade 24/7. They are not exchange prices and can differ from them, most of all while US markets are closed. Change is over 24 hours. For information only.

  • TSLA—
  • PLTR—
  • HIMS—
  • SPCX—
  • SOFI—
  • BTC—
…Loading prices
General

A five-layer framework for moats in the AI era

AI has made information, content and 'seeming to know a lot' infinitely cheap. What makes a moat scarce has shifted from 'scarce capability' to 'scarce credentials, assets and trust'. In one sentence: anything AI can learn is no longer a moat. Only what AI cannot learn, cannot get around and cannot buy still counts.

Read more

The priority for the next stock picks is a fintech or healthcare platform that has all three at once: a closed loop of private data, hard barriers from regulation, licences or physical assets, and a top-tier owner-operator.

Key evidence

  • Fake moats that have lost their value: information gaps, a pure software or algorithm lead, and content production capacity. AI has levelled them all.
  • Layer 1, private data flywheel: private data that AI cannot get, in a closed loop that grows the more it is used (HIMS's cross-product use and health data)
  • Layer 2, regulatory and licence barriers: hard credentials AI cannot pry loose, such as Nubank's US OCC licence, DLocal's emerging-market payment licences and Circle's first-mover position on GENIUS Act compliance
  • Layer 3, physical assets and networks: energy, power grids, data centres, rare minerals, logistics (the MIT data centre REIT collects picks-and-shovels rent from AI)
  • Layer 4, trust, brand and habit: Coca-Cola, Costco. AI cannot steal these with a line of code.
  • Layer 5, owner-operator capital allocation: the scarcest layer (Vélez with 75% of the voting rights, Musk, Dudum)
Show 2 more
  • An honest score for current holdings. TSLA's data flywheel is a 'story moat' that has not yet been realised. HIMS's physical assets are undervalued (352k sq ft in New Albany, peptide capacity in California and its own Hims Labs facility), and vertical integration hedges GLP-1 regulatory risk by turning it into the advantage of owning capacity. SoFi scores hard on licences, medium on its tech platform, and its cross-selling flywheel is still growing. Overall it is above average, but not a wide moat you can buy with your eyes shut.
  • The top layer of macro scarcity: the patience to focus for a long time on hard things with no immediate payoff. Value investing is a bet on exactly this.

Catalysts to watch

  • Apply this to every stock analysis video and short: score each stock on the 5 layers, separating 'realised moats' from 'story moats'
  • The next position decision is not 'which existing holding to add to' but 'which scarce dimension is missing'. The priority is a fintech or healthcare platform that holds all three layers, each of them fairly hard (Nubank is the closest candidate right now).
  • On the TSLA data-flywheel narrative: until Robotaxi actually makes money, it must not be treated as a verified moat, only as a story moat
  • On HIMS: vertical integration (its own drug manufacturing plant and lab) is the core narrative for turning regulation from a risk into a barrier. Future videos should tackle it head-on.
  • On SoFi: monitor loan default rates, member growth and net interest margin, the three signals of deterioration

Risks

  • The framework's biggest failure mode is scoring a stock highly because you already own it (putting effect before cause). Before every score, first ask 'where is this layer inflated?'
  • If TSLA's data-flywheel moat keeps failing to turn into profit, it stays at the 'story' stage and the bull narrative breaks down
  • HIMS's licence layer is blurry: GLP-1 regulation is more a threat than a protection for it (pressure from senators, FDA warnings, the Novo lawsuit). Vertical integration points towards turning the risk into a barrier, but the job is not finished.
  • SoFi's financial products are commoditised and switching costs are low (customers move for 50bps). Its moat rests on the efficiency and scale that keep CAC and funding costs low. Once growth slows, the moat gets thinner.
  • AI cuts both ways for SoFi. It lowers the bar for building a good finance app, so its UX lead will be caught up. What it can really defend is its licence, funding costs and scale.

A brief is a lead for our research, not a verified conclusion or advice.