In partnership with
The Last Time Stocks Were This Expensive Was December 1999.
"Right now, it's good. But it was in '72, '86, 2000, and 2007." - Jamie Dimon, May 2026.
The Shiller CAPE ratio just hit 42.3. The only time in 140 years it's been higher? December 1999.
Stocks can stay expensive for a long time...
It’s one metric to consider, but when your portfolio is built around the most expensive equities in modern history, what else you diversify with could really matter.
Blue-chip contemporary and post war art has shown near-zero correlation with the S&P since 1995.* Prices are largely driven by private collectors competing for a fixed supply of artwork by artists like Banksy, Basquiat, and Picasso.
Masterworks lets you invest in shares of that market.
$1.3B deployed across 500+ artworks
29 exits to date
Net annualized returns like 16.5%, 17.6%, and 17.8%, not including those unsold
*According to Masterworks data. Investing involves risk. Past performance is not indicative of future returns. See important Reg A disclosures at masterworks.com/cd.
iPrompt Signals
AI & ROBOTICS INVESTING — EXPLAINED SO YOU CAN ACTUALLY ACT ON IT
Who pays for the AI buildout? The funding ladder test
Oracle beat on everything and fell 9%. SpaceX asked the market for $75bn and got it. The same week, two inflation prints made money more expensive. Here is the four-rung test that explains all three, and tells you which of your AI holdings answers to the market.
THE THESIS IN ONE PARAGRAPH Every AI company is now climbing a four-rung funding ladder: growth paid for by your own cash flow, by your customers' commitments, by lenders, or by selling new shares. The lower the rung, the less the market can touch you. The higher the rung, the more every CPI print, every Fed meeting and every rival IPO becomes your earnings event. This week the market started pricing the ladder — Oracle's 9% drop on a record quarter is what repricing looks like. The test below takes five minutes per holding and uses only numbers companies must publish. |
THE WEEK THE BILL ARRIVED
You read the issue, so the scene is familiar: Oracle posted a near-perfect card on Wednesday night — record revenue, an earnings beat, backlog up $85bn to $638bn — and fell 9%, because alongside the beat came the bill. Capex of $55.7bn this year, roughly $70bn net pencilled for next, and about $40 billion of fresh debt and equity to raise, on top of $43bn of debt already taken on and a $20bn at-the-market share programme. Management said margins would “step down” before they recover. The market heard: the growth is real, and somebody else has to pay for it first.
And Oracle wasn't alone. Alphabet has announced an equity raise; Meta is reported to be weighing one; SpaceX priced the largest IPO in history at $135 a share. All of it landed in the week the ECB hiked, two inflation prints ran hot, and a December Fed hike became fully priced. Money got more expensive in the same week the AI trade asked for more of it. That collision is the story — and the rest of this article is the tool, so let's get straight to it.
THE FUNDING LADDER
Sort any AI company by one question — where does the next dollar of buildout come from? — and four rungs appear. Each rung down transfers risk away from shareholders; each rung up hands the market a vote on the company's roadmap.
Two classification rules before the rungs. First: grade the marginal dollar, not the historical average. Companies straddle rungs, and the market prices where the next dollar comes from — Oracle funded its past from operating cash; its next $40bn is borrowed and issued, and that is what Wednesday repriced. Second: the ladder measures two exposures at once — the direct source of a company's funding, and the indirect source one step upstream, at its biggest customer. We'll use both.
The four rungs, with this week's casualties and survivors sorted onto them.
RUNG 1 — SELF-FUNDED
Growth paid out of operating cash flow, with money left over. Nvidia is the archetype: it raised its dividend this month, mid-buildout, which is a company advertising that the premium is already earned. Microsoft, Alphabet and Amazon have historically lived here too. A rung-1 name can be wrong about AI demand and still never need the market's permission to keep going.
RUNG 2 — CUSTOMER-FUNDED
Growth paid by buyers before the product exists. This is memory's quiet superpower: Micron's remaining FY26 HBM output is committed under long-term contracts, SK Hynix just pledged to triple HBM capacity inside a multi-year Nvidia pact, and Broadcom sits on more than $30bn of bookings its CEO calls “simply insatiable.” Customers carrying the risk is the strongest position in a funding squeeze — it's why Micron erased a 9% weekly drop in a single session.
RUNG 3 — DEBT-FUNDED
Growth paid by lenders. Oracle has raised $43bn of debt this fiscal year and plans more. The CoreWeave-class neoclouds live here permanently. Debt is fine while rates fall; it reprices brutally when two hot inflation prints push the 10-year toward 4.7% and the Fed's December hike goes from possibility to consensus. Every rung-3 name now has a rates desk whether it wants one or not.
RUNG 4 — EQUITY-FUNDED
Growth paid by selling new shares — dilution. Oracle's $20bn ATM programme. Alphabet's announced raise. Meta, if the reports confirm. SpaceX's $75bn is the rung-4 trade at planetary scale. This is the only rung where the funding mechanism directly subtracts from your slice of the business, which is why the market prices it instantly and unkindly.
THE FIVE-MINUTE TEST
You don't need a model. Five numbers, all in the cash flow statement and share count, all published quarterly. Run them on each AI holding and the rung assigns itself.
CHECK | WHERE TO FIND IT | HOW TO READ IT |
Capex ÷ operating cash flow | Cash flow statement, both lines | Under 1.0 = self-funding. Over 1.0 = someone else is paying. Direction matters more than the level. |
Financing cash flow sign | Cash flow statement, bottom third | Persistently positive = the company is a net taker of market money. Negative = returning it (rung 1 behaviour). |
Share count trend | Income statement, diluted shares | Rising share count during a buildout = rung 4 in progress, whatever the press release says. |
Customer prepayments | Balance sheet: contract liabilities / deferred revenue | Growing prepayments = customers funding the roadmap. The rung-2 signature. |
The dividend/buyback tell | Capital returns announcements | Raising returns mid-buildout is a flex only rung 1 can afford. Cutting them is the first quiet step up the ladder. |
WORKED EXAMPLE: ORACLE
Run Wednesday's numbers through the test. Capex of $55.7bn against operating cash flow that doesn't cover it — ratio well above 1.0, and rising into next year's $70bn net plan. Financing cash flow strongly positive: $43bn of debt raised, $40bn more planned. Share count set to rise via the $20bn ATM programme. That's rungs 3 and 4, simultaneously.
But be fair to the bull case, because the test catches this too: Oracle disclosed roughly $75bn of prepaid and customer-supplied hardware offsetting the gross spend. That's genuine rung-2 mitigation — real customers committing real money against the backlog. The honest read is a company straddling rungs 2 through 4: contracted demand that's almost certainly real, funded by a structure the market now taxes. Both things were true on Wednesday night. Only one of them was new.
WORKED EXAMPLE TWO: MICRON, IN SIXTY SECONDS
To prove the ladder travels, run the same five checks on the week's other protagonist. Micron reads the opposite way on nearly every line: its remaining FY26 HBM output is committed under long-term contracts before the capacity exists — the rung-2 signature, sitting in plain sight on the balance sheet — with no announced raise and no debt binge funding the expansion. Then the upstream check: its HBM buyers are Nvidia and the rung-1 hyperscalers, names that pass the test themselves. Direct funding clean, indirect funding clean.
That double-pass is why the same selloff repriced Oracle for a week and Micron for a day. Same sector, same demand story, opposite rungs — the ladder, not the narrative, predicted which one snapped back.
THE CIRCULARITY PROBLEM
This is the second exposure defined at the top — the indirect one — and it stops rung 2 being a safe house. Ask who Oracle's customers are. The biggest name behind that $638bn backlog is widely reported to be OpenAI — a company that is itself a serial equity raiser. Which means some of Oracle's “customer-funded” growth is, one step upstream, market-funded after all. Rung 2 can be rung 4 wearing a contract.
The fix isn't to abandon the test — it's to run it twice: once on your holding, once on its biggest customer. Micron passes both (its HBM buyers are mostly rung-1 hyperscalers and Nvidia). Oracle passes the first and, on the reporting, partly fails the second. That difference is roughly the difference in how the two stocks traded this week.
THREE FALSE POSITIVES
If you keep one section of this article, keep this one. A framework applied mechanically is worse than no framework, because it manufactures false confidence. Three cases where the obvious ladder read is wrong:
1. THE SMART RAISE
Alphabet announcing an equity raise looks like rung 4, but a self-funder selling shares at record prices to bank cheap capital before the window shuts is opportunism, not distress. The tell: does capex still fit inside operating cash flow without the raise? If yes, the raise is a choice. That's why our framework keeps Platforms at HIGH conviction with a funding footnote rather than a downgrade.
2. THE FRAGILE SELF-FUNDER
A rung-1 badge is one guidance cycle from expiring. Meta funds from cash flow today; if the reported raise confirms and capex guidance jumps again in July, it steps onto rung 4 in a single press release. Treat the rung as a reading, not a rating.
3. THE CHEAP-DEBT ILLUSION
Investment-grade borrowers will tell you debt is cheaper than equity, and at issue it is. But debt's cost is set on the day the market is angriest — at refinancing. With December's hike fully priced and PPI at 2022 levels, every rung-3 maturity schedule deserves five minutes of your attention. The coupon is not the cost; the rollover is.
THE SPACEX GAUGE
From today you also get a free, real-time reading on the whole question. SpaceX at $135 is the largest single ask the equity market has ever fielded — $75bn, priced fixed, into a tape that just got two hot inflation prints. Its first weeks of trading are a live gauge of the funding window:
Holding comfortably above $135 = the pool is deep, equity raises clear easily, rung-4 names breathe. Breaking below $135 = the marginal dollar is exhausted, and every announced raise — Oracle's ATM, Alphabet's, Meta's if it comes — gets harder. You don't need to own SPCX to use it. If you hold any rung-3 or rung-4 name, SPCX's tape is effectively their cost-of-capital ticker: it tells you how much dilution and refinancing the market will tolerate before it starts taxing yours.
WHAT TO WATCH
WHEN | EVENT | WHAT IT SETTLES |
Wed 17 Jun | FOMC decision + dot plot | The price of every rung 3 and 4 dollar. A hawkish dot plot re-runs Friday's sort within hours. |
Wed 24 Jun | Micron fiscal Q3 | The rung-2 showcase. FY27 HBM contract pricing commentary is the single line that settles the cycle-vs-structure debate. |
Daily | SPCX vs the $135 issue price | The funding-window thermometer. Watch the level, not the headlines. |
July | Hyperscaler capex guides | Who stays on rung 1? Any self-funder guiding capex above cash flow steps up the ladder in real time. |
Ongoing | US 10-year above 4.7% | The level where rung-3 maths visibly breaks for the levered builders. Currently ~4.53%. |
TURNING RUNGS INTO RESEARCH POSTURES
Rungs 1–2: your research question is demand — is the order book real, is pricing holding? Funding takes care of itself.
Rung 3: your question is the maturity wall — when does the debt roll, and at what rate does the project maths stop working?
Rung 4: your question is the share count — how much dilution is announced, how much is implied, and is the market still paying record prices for the new paper? None of this says sell. It says know which question your holding has to answer, because the market is now asking it weekly.
THE RULE THAT CHANGED (AGAIN)
Issue 13 said the market had started sorting numbers from noise. That rule survived the week — but it grew a second clause. Numbers now come with an invoice attached, and the market reads the invoice first. Oracle proved a perfect quarter can't outrun a funding plan. Micron proved sold-out capacity can outrun anything. Between those two results sits every AI name you own.
And to be honest about what would disprove all this: if SPCX holds well above $135, Oracle's raise clears at tight spreads, and the December hike gets priced back out, then funding never became the axis and Issue 13's demand sort still rules alone. That is a checkable outcome, not a hedge — we'll call it either way in a future issue.
THE ONE-LINE TAKEAWAY In a rising-rate AI buildout, the question isn't who has the best roadmap — it's whose roadmap doesn't need the market's permission. |
YOUR NEXT MOVE
Tonight, take your single largest AI holding and run the five checks — five minutes, all public filings. Write the rung on a sticky note with the date. Then make it a ritual: each Friday after the issue, grade one more holding — a month from now your whole AI book is laddered, and every earnings season re-grades it for free. After the Fed on Wednesday and Micron on the 24th, run the first one again. If the rung moved, you'll know before the market explains it to you.
See you Friday.
FROM ISSUE 14 This article accompanies iPrompt Signals Issue 14, which covers the full week — the $1tn chip round trip, the Nvidia–SK Hynix pact, Intel's quiet streak, and three funding-sorted ideas to research this weekend, led by Micron into the 24 June print. |
Disclaimer: This article is for informational and educational purposes only and does not constitute financial advice. iPrompt Signals is not a registered investment advisor. Always conduct your own research and consult a qualified financial professional before making investment decisions.