In February the market wiped $300B off software stocks and called SaaS dead.
The spending data never agreed. Companies still pay 'per seat.'
The weak apps died. The strong ones just got cheaper.
Time to buy $ADBE and $NOW ?
Japan at 1%: Gravity Comes Back to the S&P
The Bank of Japan takes its policy rate to 1% this week, the highest in thirty years. Almost every write-up reads it as a Tokyo story or a yen story. For a US equity book it is neither. It is a gravity story. And the thing about gravity is that it works slowly. This does not show up in one bad session. It shows up over quarters and years, in the price of the most expensive stocks you own.
Here is the whole argument in one picture. 👇
The patient buyer is going home
For a generation, Japan has been the largest foreign owner of US government debt, sitting on roughly $1.2 trillion of it. Picture Japanese pensions and insurers as a huge, patient buyer who has stood under the US bond market for years, quietly soaking up Treasuries and keeping America's long-term borrowing costs lower than they would otherwise be. Nobody talked about this buyer because he never left.
He is leaving now. With interest rates back home finally worth something, a Japanese insurer can earn a respectable return in its own currency without taking the risk of shipping money abroad. So the money is coming home. In the first three months of this year, Japan sold the most US Treasuries since 2022, and the pace is picking up.
Why your stocks care
When the biggest buyer steps back, US long-term interest rates drift higher. Not because the Fed wants them to. The Fed's next move is still expected to be a cut. They drift higher because the bid that quietly held them down is fading.
And long-term interest rates are gravity for stocks. It is the old line that rates do to share prices what gravity does to matter. When rates are low, gravity is weak and prices float up to rich valuations. When rates rise, gravity strengthens and pulls them back down.
The catch is that gravity does not pull on everything equally. It pulls hardest on the stocks whose value sits furthest in the future. A dull company that hands you cash today barely feels it. A company whose whole value is a promise of huge profits a decade from now feels it most, because a higher rate marks down that far-off promise much harder.
That is the exact description of the part of the market everyone owns: big-cap tech. And here is the quiet problem. The S&P is no longer a spread-out index. It is now roughly a third concentrated in a handful of giant, long-dated tech names. The index has quietly turned into a bet on distant cash flows. So when gravity strengthens, "the market" and "the most expensive stocks in it" have become almost the same trade.
Where the risk sits, by name
Most at risk, to the downside, are the long-duration megacaps.
Nvidia ($NVDA ) is the sharpest example. It is the most crowded stock on the planet, held by leveraged funds everywhere, and it is the face of an AI-spending story whose payoff is already being questioned. It gets hit from both directions: fast, if a stronger yen forces global funds to sell their best winners to cover losses, and slow, as a higher discount rate deflates the multiple over time.
Oracle ($ORCL ) is the one exposed twice. It is a long-dated AI bet, and it borrowed tens of billions to build data centers. Higher rates raise both the discount on its future profits and the cost of rolling that debt. Same move, two hits.
The quieter casualties are the bond substitutes. Utilities ($XLU ) and real estate (REITs) are bought mainly for their yield. When safe bonds start paying more, these suddenly look worse by comparison and bleed.
Most resilient, and the relative winners, are the banks. A world of higher long-term rates is good for lenders, who make money on the gap between what they pay short and earn long. JPMorgan ($JPM ) and the broad financial sector ($XLF ) are the natural beneficiaries. Next to them sit the unglamorous value names that pay you now instead of later. And cash, for once, is a real position rather than an embarrassment.
Timing, and the one number to watch
This is the part to hold onto. This is not a crash you trade around a single meeting. Japanese institutions turn like oil tankers, the repricing comes in over quarters and years, and the Fed can paper over it for a while.
The risk worth respecting is that the slow story gets front-run by a fast one. If the yen snaps sharply stronger, leveraged funds dump their most liquid winners to cover, and you get an air pocket in megacap tech of the kind we saw in August 2024, before the slow grind has even started.
So the number to watch is not the BOJ's rate. It is the US 30-year Treasury yield. The BOJ headline is the cause. The long bond is where it lands, and it is the early warning for everything expensive you hold. When that yield keeps grinding higher even while the Fed is cutting, that is gravity doing its work, and the top floor of the market, the priciest tech, is the part that sways.
Everyone is watching Tokyo. Watch the long bond.
Construir um datacenter de 1GW custa entre $40-80 mil milhões.
Metade são chips que derretem em 3-4 anos — e a renda a cair ~70%.. Será a $ORCL um bom negócio?
Na 10Q fomos fazer a conta que ninguém faz. 🧊
There's one number that quietly decides whether the entire AI buildout actually makes money — and once you see it, you can't unsee it. So let's do the math, with real figures.
Building a one-gigawatt AI data center costs about $38 billion. The chips — the Nvidia accelerators that do the work — are the single biggest slice: roughly $20 billion, over half. The rest is the building, the power, the cooling, the networking.
Here's the catch. The building and the power last 15-20 years. The chips? Effectively obsolete in about three. $NVDA ships a faster generation on a relentless cadence, and older chips lose 40-60% of their value within 18-24 months of the successor arriving.
So more than half of the most expensive thing humanity is building right now melts in three years. Picture a $38B warehouse where $20B of it is ice sculptures.
That used to be fine — because the rent was insane. In 2023-24, an Nvidia H100 paid for itself in under a year. Easy money.
Then everyone built. Capacity flooded in and the rent collapsed: H100 rental rates are down roughly 64-70% from their peak, to around $2-4 an hour. At those prices the payback period has stretched from under a year to seven-to-ten years — on an asset that's obsolete in three. You can't earn back a three-year chip over ten years of rent. At today's prices, the math just doesn't close.
Don't take our word for it — read the purest example's own filings. $CRWV does nothing but rent GPUs. Last quarter revenue grew 112% to $2.1 billion. Spectacular, right? Except depreciation alone — the chips wearing out — ate $1.15 billion, more than half of all revenue. The result: a $740 million net loss. Booming demand, losing money on every dollar, because the melting eats more than half the rent.
Then there's the accounting. Most big players write these chips down over five or six years, not three — which makes reported profits look fatter than the cash reality. One prominent short-seller estimates roughly $176 billion of "missing" depreciation across the industry through 2028 — enough to flatter the reported profits at names like Oracle by 20%+.
The demand is unquestionable. But the returns, at today's collapsed rents and honest depreciation, are underwater for the pure players and propped up by generous depreciation schedules for the big ones.
This is exactly why our system keeps rejecting the debt-funded builders. $ORCL sits below our quality floor not because demand is weak — but because borrowing tens of billions to buy a melting asset whose rent is falling is a fragile way to make money.
To be fair to the other side — and we always try to be — this flips if a few things go right: if rental prices stabilize as demand finally outruns the supply glut, if the chips stay useful past three years (older ones still rent), and if utilization stays near-maxed. Any of those, and the math closes again.
But the honest read today: the prettiest demand story in tech is sitting on the fastest-melting asset in tech, and the rent is heading the wrong way. So forget revenue growth — everyone has that. Watch one thing: whether GPU rental prices stop falling. The day they stabilize is the day this becomes a business instead of a race.
Would you borrow billions to buy something half-gone in three years — while the rent keeps dropping?
https://t.co/cYs46lMFyy
Next - my interview with @PataoPT, first time ever of having the incredible @RektHQ on my show and on @leviathan_news!
The "Rektrospective" -
https://t.co/M0A6IHRnIy
Rekt has always been a journalistic view of the space. As people change and social changes, so does the format.
@PataoPT, General Manager at @RektHQ, on why Rekt is moving into podcast, video, and live stages: the conversations happening in private channels with security teams should be available to everyone.
Hear it from Diogo directly. Full episode out now:
SpaceX. $1.75T. Maior IPO de sempre. Mas o agente canónico da 10Q passa.
Quality screen falha. Preço a 90× receita falha. Kill condition disparava no Dia 1.
A disciplina é o produto — frameworks dizem não onde os founders dizem sim.
https://t.co/sC4ZoAFBOZ
⸻
$SPCX prices in 22 days at $1.75T. Largest IPO ever recorded. Our canonical agent has read the S-1 — and her quality+price framework says she won't touch it.
The math fails twice before it has a chance to work.
𝗠𝗔𝗥𝗞𝗘𝗧 ��𝗢𝗡𝗧𝗘𝗫𝗧
SpaceX filed May 20. Goldman, MS, BofA, Citi, JPM as joint book. Nasdaq debut June 12 under SPCX, $75B raise — bigger than Aramco ($29.4B) and Ant ($34.5B) combined. For the first time we can read the actuals, and they split three ways inside one ticker:
- 𝗦𝘁𝗮𝗿𝗹𝗶𝗻𝗸 (connectivity): $11.4B rev · $4.4B op income · 50% YoY · 10.3M subs · the engine that funds everything
- 𝗦𝗽𝗮𝗰𝗲 (launch): $4.1B rev · -$657M op income · 80%+ of global mass to orbit · reinvesting every dollar into Starship
- 𝗔𝗜 (xAI / Grok / X): $3.2B rev · -$6.4B op income · $12.7B capex · the gravity well
Consolidated 2025: $18.7B revenue, -$4.9B net loss (vs +$791M in 2024, pre-merger). The xAI absorption turned a profitable company into a loss-maker by design.
𝗢𝗨𝗥 𝗣𝗢𝗦𝗜𝗧𝗜𝗢𝗡𝗜𝗡𝗚 (𝗖𝗹𝗮𝘂𝗱𝗲 𝗦&𝗣, 𝗰𝗮𝗻𝗼𝗻𝗶𝗰𝗮𝗹 𝗳𝗿𝗮𝗺𝗲𝘄𝗼𝗿𝗸)
SPX strategy has a quality floor before anything else.
Consolidated burns $4.9B annually — fails the screen on entry. Strip Starlink out and you have a $4.4B op-income business growing 50% YoY. That is a real business worth real money — maybe $200-400B standalone at any reasonable multiple. But you can't buy Starlink. You buy the consolidated entity at $1.75T.
That's 𝟵𝟬× 𝗿𝗲𝗰𝗮𝘀𝘁 𝗿𝗲𝘃𝗲𝗻𝘂𝗲. Forward P/E doesn't exist (net loss). PEG breaks the math. Our canonical agent writes the kill condition before the position opens — and on this name, 𝘁𝗵𝗲 𝗸𝗶𝗹𝗹 𝘄𝗼𝘂𝗹𝗱 𝘁𝗿𝗶𝗴𝗴𝗲𝗿 𝗗𝗮𝘆 𝟭.
𝗛𝗢𝗧 𝗧𝗔𝗞𝗘
Bull case requires three things ALL to be true: Starship to orbit on schedule (2H 2026, 11 flight tests so far), Starlink ARPU stops compressing ($99 → $66 over three years, still falling), AI segment burn stops accelerating ($7.7B Q1 capex alone, $30B+ annualized). Each unproven. The optionality is real, but the $1.75T price already embeds full success on every leg.
The deeper question: who is this IPO for? Retail access via Schwab, Fidelity, Robinhood at the same price as institutions. 𝗠𝘂𝘀𝗸 𝗵𝗼𝗹𝗱𝘀 𝟴𝟱.𝟭% 𝘃𝗼𝘁𝗶𝗻𝗴 𝗽𝗼𝘄𝗲𝗿 (10 votes/share Class B). Anthropic pays $1.25B/month to rent COLOSSUS compute — terminable on 90 days notice. The S-1 reads like a venture deck wearing IPO clothes, sold to public investors at venture pricing.
𝗙𝗢𝗥𝗪𝗔𝗥𝗗 𝗟𝗢𝗢𝗞
SPCX won't enter Claude S&P's watchlist post-IPO under current framework — q-score on consolidated financials fails the floor. Three things would change our read: a price range materially below $1.75T (sub-$1T gets interesting), a Starship slip that forces a re-rating, or a Starlink spin-off as separate ticker (not contemplated in the filing).
Watching the pricing amendment. The honest framework says: 𝗽𝗮𝘀𝘀.
The question we're carrying for the audience: 𝗮𝘁 𝘄𝗵𝗮𝘁 𝘃𝗮𝗹𝘂𝗮𝘁𝗶𝗼𝗻 𝘄𝗼𝘂𝗹𝗱 𝘁��𝗲 𝗰𝗮𝗻𝗼𝗻𝗶𝗰𝗮𝗹 𝗳𝗿𝗮𝗺𝗲𝘄𝗼𝗿𝗸 𝘀𝗮𝘆 𝘆𝗲𝘀? Replies welcome.
It was a pleasure to share my thought about @RektHQ Blockhain Security and Investigate Journalism in Crypto with @leviathan_news
Rektrospective: Crypto's Investigative Journalists https://t.co/PugSIvAoDE via @YouTube
Ethereum funds rescued during the 2016 DAO hack are now being used to fund Ethereum security.
On CypherTalk podcast (Ep. 5), @PataoPT
from @RektHQ joins @beyer_st and @pumpkinGMIto discuss the DAO Security Fund and how quadratic funding helps support security initiatives across the ecosystem.
REKT SECURITY SUMMIT
March 27, 2026 | Cannes, France
Learn from teams who survived major exploits. Hear from researchers who've documented $2B+ in losses. Understand what actually breaks and why.