Valuations across technology, financials and consumer — built from primary filings, with the model attached. Each note ships its model as a workbook.
The quarterly report answers what the release left open: if a customer or a funding market fails, who gets paid first? We map every claim, trace the parent's own cash quarter by quarter, and name the events to watch. It follows The Floor Moved to the Fleet.
Read the note →·PDF·Model·The Floor Moved to the Fleet →Palo Alto Networks beat, raised and bought an AI start-up with stock, as a wave of software flaws made national news. The quarter strengthens its next two years; we show what the price needs after that. It follows our September update.
Read the note →·PDF·Model·The original valuation →The quarter beat our July top case, the founder left, and the stock fell anyway. What does $155 pay for now? One guided year, and almost nothing after it. We rebuild the valuation and name what moves it next.
Read the note →·PDF·Model·Priced for the Clock Not Breaking →CoreWeave's stock fell a third from July while its margin went from 1% to 10% and its older GPUs re-contracted above their first prices. The credit market and the business disagree about what the equity is. We find each is right about a different thing, and value what the fleet, not the signed book, is worth.
Read the note →·PDF·Model·The Thin Floor →Block now guides a margin our July valuation's best case did not reach, and the legal charge July called one-time came back three times larger. We value what the margin keeps after the bill, the loan book and Block's own risk, and name the one judgment the value rests on.
Read the note →·PDF·Model·The original valuation →Palantir's revenue grew 96% while the contract value it closed grew 6%. We sort out how much of that gap is real, and what the price asks of a business growing this fast.
Read the note →·PDF·Model·The Fade Line, Measured →Apple spends seven cents of capex per dollar of net income and still got the bill for the AI build, through memory. What a record year is worth once that bill is counted, and what the price asks of Apple's discount rate and growth.
Read the note →·PDF·Model·The Control Group →Amazon's cloud is accelerating toward 40% growth, on capacity customers reserve before it is built. The bill arrives first. What the faster utility is worth once that bill is counted, and what a dollar of new plant must earn for the price to hold.
Read the note →·PDF·Model·The Honest Machine →Coinbase's trading revenue hit a two-year low, while lower costs, prediction markets and stablecoin offset most of the damage. What a share is worth once two double counts come out, how likely a new crypto cycle is, and what return the price asks owners to accept.
Read the note →·PDF·Model·The Right Company at the Wrong Price →In July we valued Strategy at its net asset value and gave its credit franchise no value. Then it sold $5.41 billion of stock and bought little bitcoin. What that money bought instead, what it cost each share, and what the price now pays for.
Read the note →·PDF·Model·The Q2 update →·The original valuation →Robinhood's earnings fell because of the tax rate, as we said in July they would, and the business beneath kept growing. What a share is worth on the returns its equity earns, what a Supreme Court ruling would cost, and what the price asks of the next decade.
Read the note →·PDF·Model·The Price Is a Growth Commitment →Alphabet delivered half of what we said in July would close the gap to its price: a $561 billion backlog. The gap widened anyway. What the order book is worth once its bill is counted, and what the price asks of the years to 2040.
Read the note →·PDF·Model·The Convention Price →In July, Microsoft said it was stretching its building lives from 15 to 25 years. Its first 10-Q since says nothing about it, but the lease note moved. What the silence means, what the quarter did to the value, and what the price now asks for.
Read the note →·PDF·Model·The $116 Billion Question →GE Vernova's cash flow is mostly its customers' money, paid years ahead for turbines still to be built. We count it as timing, not earnings, and show how long the sold-out market has to last for today's price to hold.
Read the note →·PDF·Model·The Sold-Out Decade →Anthropic is cheap to serve and expensive to build: an 85% serving margin against $657 billion of compute already promised. Inside: what the S-1 leaves blank, who holds the vote, when insiders can sell, and what each price being discussed would require.
Read the note →·PDF·ModelTesla beat the estimate, and this time the profit was operating. Our fair value moved by less than three dollars. The price moved what it requires of robotaxis and robots by a fifth, and now implies a discount rate below the Treasury yield.
Read the note →·PDF·Model·Four Times the Finish Line →Meta’s quarter beat, and our fair value still fell. The filing behind it shows much of 2027’s bill already signed and a share class that can help pay it. Inside: why we stopped counting earnings that aren’t becoming cash, and what would bring them back.
Read the note →·PDF·Model·Paying Twice (July) →SpaceX tripled its compute revenue in one quarter, then doubled the capital it will spend in the next. Where the fleet’s cash catches its capital, what the share price requires of a 13-gigawatt plan, and the 1.6 billion shares that trade free this week.
Read the note →·PDF·Model·The First Print, Scored →Anyone can launch an ETF; few can build BlackRock’s volume. Commoditization has slowly eaten at ETF fees for years — can scale defend the premium? We value the whole company, ETFs, active funds, private markets and Aladdin, each on its own growth, and show what today’s share price assumes.
Read the note →·PDF·Model·US Banking Before the Print →Goldman earned 22% on its tangible equity this year, and the five big US trading banks’ own filings show a real franchise behind it. Return the market to normal and that franchise earns about 16%. At $892 the shares need more than this year’s market, forever.
Read the note →·PDF·Model·US Banking Before the Print →JPMorgan is on a fifth straight year at or above its 17% return target, and just cut buybacks 84% rather than buy its stock at 2.9 times tangible book. The price needs a 24% return forever. Is the best decade in the company’s history the steady state?
Read the note →·PDF·Model·US Banking Before the Print →Micron beat, guided 7% above the Street, and signed ten more take-or-pay contracts. It also committed more than $50 billion of capital spending at a record margin. The scorecard we froze before the print, graded, and a fair value that turns on one undisclosed number.
Read the note →·PDF·Model·The original valuation →Micron guides a quarter to $50 billion at an 86% margin, and the stock has sat below its June high for thirteen weeks. The last two times the company printed a record margin, the stock had already peaked. A full valuation through the memory cycle, and an eight-gauge scorecard frozen before Wednesday’s print.
Read the note →·PDF·Model·Scored after the print →Nike delivered the first gross-margin expansion in eight quarters, then guided a year below our own downside case, with earnings under the dividend. The scorecard we froze before the print, graded: three cleared, four missed, one not given — and what the company is worth now.
Read the note →·PDF·Model·The original valuation →Nike at $36 is neither the bargain the yield advertises nor the trap nine sell-side cuts imply: the price requires a 9.3% terminal operating margin, halfway between last year’s 6.1% and the 12–14% of 2019–2024. Eight gauges frozen before the October 1 print.
Read the note →·PDF·Model·Scored after the print →Palo Alto Networks beat, raised on every line, and fell 9.3% — the fourth straight beat the market has sold. Six of eight frozen questions cleared; the two misses are the ones the multiple rests on. The scorecard we froze before the print, graded — our misses first.
Read the note →·PDF·The original valuation →Palo Alto Networks is executing almost flawlessly — and that is not the question. What does $372 have to believe, what is the execution actually worth, and what can Tuesday’s print settle? A full valuation, with a scorecard frozen before the answer arrives.
Read the note →·PDF·The sector thematic →·Scored after the print →NVIDIA beat, guided $108bn, and jumped 8.7% — and the quarter’s biggest number was in a footnote: $108.5bn of guarantees now underwriting the AI buildout. The scorecard we froze before the print, graded — our misses first.
Read the note →·PDF·The July valuation →NVIDIA reports Wednesday at a price above every scenario our July valuation models. Before the print, we freeze nine gauges, what silence on each will mean, and the outer bound of any fair-value move. The commitment comes first; the scorecard follows.
Read the note →·PDF·The July valuation →Five gauges published the night before SpaceX’s first-ever earnings report, scored with no edits: three cleared, two missed. The guide we were told did not exist is the quarter’s most aggressive number — $100bn of annualised revenue by December against $23bn today — fair value rises 18% to $78.51, and the AVOID stands. And 911.5 million shares unlock 6 August into a float of 646 million.
Read the note →·PDF·The pre-print gauges →The sequel scores the print: every pre-registered threshold cleared — revenue $1.94bn (+93%), guide $8.15bn (+82%) — and fair value restates $112 → $149 under rules written before the answer. By the open the market had repriced through it: at $154 and 59× sales, richer than before the print, the burden of proof now rides with the bulls.
Read the note →·PDF·The original framework →SpaceX reports tomorrow — the first earnings print in its 24-year history — down 31% from its first close toward the $66.50 fair value of our June initiation. What the print must show to justify even $111: five gauges pre-registered the night before, a first-print straddle with no base rate for anyone, and the Tesla-merger arithmetic the descent is repricing weekly.
Read the note →·PDF·The original valuation →Palantir reports tonight at 56× sales. We run the valuation backwards: the price demands ~31% revenue growth for a decade, and the whole bull–bear fight compresses to one number — the fade rate. Breakeven ×0.78; weighted fair value $112. Falsifiable within hours.
Read the note →·PDF$115.9bn of equipment in one year, Big Tech's widest server depreciation range, and $17.95 of EPS that our three-year clock re-marks to $16.96. On the day it filed “no change,” it stretched a different clock instead.
Read the note →·PDFThe only company in Big Tech that ever shortened a server clock — and it paid double its own guide for the honesty. AWS grew its fastest in eighteen quarters; trailing free cash flow went negative anyway.
Read the note →·PDFFour of the Mag 7 spend $1–2 on AI hardware per dollar earned; Apple spends seven cents, and its capex is falling. The house depreciation re-mark rounds to zero. The cleanest earnings in the index, at its most expensive price.
Read the note →·PDFEvery server is paid for twice: cash today, earnings tomorrow. The first payment hit $31.1bn against $15.8bn of net income — and the second, on the longest server clock in Big Tech, has barely begun.
Read the note →·PDFOur June valuation called the risk a slow funding drain, not a liquidation. Q2 ran it in real time: the dollar reserve fell to half a year of coverage, bitcoin was sold at a $203M realized loss to fund a preferred dividend, and STRC broke to $70. Eight of our own claims scorecarded — one of them wrong — plus two claims from the call that do not survive scrutiny.
Read the note →·PDF·The original valuation →Record market share, record USDC, a new $100M product line — against a revenue miss, a third straight GAAP loss, and a worse July. Trigger check on our July valuation: nothing tripped; fair value holds at ≈$102 vs ~$155.
Read the note →·PDF·The original valuation →Alphabet trades at a reported 17× earnings. Strip out the $99.0bn securities gain and the clean multiple is 33×. Capex now exceeds operating cash flow. The distance from our fair value to the market's price is not a disagreement about the business — it is three accounting conventions.
Read the note →·PDFEighty-six percent of Tesla's $1,088bn enterprise value is the price of robotaxi and Optimus. That price embeds roughly four times the company's own board-set milestone of one million robotaxis — at one-seventh the valuation the award attaches to it. Unsupervised vehicles counted today: about 21.
Read the note →·PDFThe purest winner of the AI power buildout beat Q2 and raised guidance — and fell 8.7% anyway, because at ~25× its own 2028 target the good news was already paid for. Backlog $176bn, turbines sold out to 2030. The multiple, not the business, is the risk.
Read the note →·PDFThe best business in the AI complex, at a price that already assumes the buildout never plateaus. A reverse-DCF says $203 requires ~$1tn of revenue by FY2031 — and NVIDIA's revenue is, by identity, other companies' capex. Great business, full price.
Read the note →·PDFA 64% fall took Oracle from pricing its nameplate AI backlog to pricing our base case. The legacy software annuity is worth ~$72 a share alone; at $124 the market pays ~$52 for an AI bet whose defining margin Oracle will not disclose — full value for an outcome that must go right.
Read the note →·PDFThe neocloud thesis in one security. Revenue more contracted than its peers' — take-or-pay, prepaid, Microsoft-anchored — but the signed book covers a seventh of the price, and the rest is a levered option on how long a GPU stays useful, behind $25B of debt.
Read the note →·PDFThe price is a growth commitment. Robinhood broke its crypto dependency mid-winter — crypto revenue halved while total revenue grew — but at ≈62× normalized earnings, $110 banks a 20%-a-year compounding machine for four uninterrupted years, with its fastest-growing line awaiting a Supreme Court petition.
Read the note →·PDFThe most misread name in crypto. Reported revenue overstates the business (a third is low-margin bitcoin pass-through) and a tripled loss line is mostly accounting timing, not a credit blow-up — underneath sits a mid-teens gross-profit compounder priced at roughly fair value, conditional on one credit signal.
Read the note →·PDFThe price of distribution. Circle earns the interest on $73B of USDC — then pays nearly two-thirds of it away to the partners who distribute it. A July OCC national trust-bank charter widens the moat but leaves that split — the reason it screens rich — untouched; fair value holds at ≈$50.
Read the note →·PDFThe right company at the wrong price. Coinbase is two businesses — a hyper-cyclical trading franchise and a quietly compounding “financial plumbing” annuity — and at today’s price the market pays a peak year as if it were permanent, returning less than simply holding bitcoin in every scenario.
Read the note →·PDF·Q2'26 addendum →What are you really buying? After a brutal de-rating the famous NAV premium is gone — MSTR trades at ~1.0× net asset value, no longer the “$2.50 for $1 of bitcoin” the bears shorted. What’s left is a ~1.7× levered, convex bet on bitcoin carrying a permanent preferred coupon.
Read the note →·PDF·Valuation model (XLSX)Priced for Mars: the $1.77 trillion listing embeds the bull case, and then some. A full valuation built from the S-1/A puts fair value at $66 against a $135 offer — a structurally rich entry that leaves no margin of safety for execution risk on Starship or Starlink.
Read the note →·PDF·Valuation model (XLSX)Coverage is initiated selectively, where a primary-filing-driven valuation adds something the sell-side consensus misses. New reports are added here as they publish. For broader thematic and empirical work, see the research library.