TSLA
PLTR
HIMS
SPCX
SOFI
META
BTCCase study: Palantir, step by step
Our framework applied to one company we follow, with the numbers from our dossier of 18 August 2026
Runway
Passes+93%sales growth in the second quarter of 2026, up from about 30% in 2024US commercial sales grew 149%, yet Palantir has only 653 US commercial customers so far. The open question is how much of that market there is to win.
Read ‘A long runway’Moat
Strong on four of five157%net dollar retention: what existing customers spend now, against a year earlier- Private data flywheel
- Regulation and licences
- Physical assets and networks
- Trust, brand and habit
- Owner-operators
Close to untouchable with the customers it already has, but in a fight for the ones it doesn’t. Rivals sell tools and models; Palantir models the business itself.
Read ‘Five moats AI can’t reach’Scenarios
Fair priceYearly return over three years, from a $430 billion market value at 153 times earnings
Over six years the base case is worth 4.9 times today’s price, but for the first three it may look like a mistake.
Read ‘Fast growers: value them like options’Price zone
Middle thirdAt $174 the price already assumes the base case, but not the good one: the middle third.
Read ‘How to set the range’Decision
Buy in batchesNot all at once: at this price there is no margin of safety. Buy in batches; for the part not yet bought, sell puts or set a Turbo Range position at a bottom-third price. Then stick to the plan.
Read ‘Buy: let the price set the pace’What would prove us wrong
Checked every quarter- Net dollar retention falls below 120%, or keeps falling for two or three quarters
- New customers slow down: the land grab is losing pace
- Sales grow no faster than headcount: growth is still bought with people
- Stock-based pay, now about 14% of sales, stops falling
- Companies start letting AI decide without people, at scale: rethink the whole case
A worked example of our method, not a recommendation. Read the full dossier