Gabor Giran

Making fun of business concepts until you actually understand them

[Gabor Giran @GGaborGiran]

Dear ladies: you can't price a 30 year bond off its day one yield.

Expecting a 25 year old guy to already have the same success his dad built by 55 is the same mistake. You're judging the principal before the compounding even started.

The first decade of any long duration position looks unimpressive. That's not a red flag, that's just the front end of an exponential curve.

His dad isn't ahead because he started richer. He's three decades further down the compounding curve, that's the whole difference.

Let the position run before you call the return.

[Gabor Giran @GGaborGiran]

The Fed cut rates and your mortgage rate didn't move. You're not being scammed.

The Fed sets the overnight rate banks charge each other. Your mortgage is priced off long term bonds, which move on what markets expect for the next ten years.

So a cut today can actually push mortgage rates UP, if the market reads it as "they're worried, inflation is coming back."

Short rates are a decision. Long rates are an opinion poll.

The Fed holds one end of the rope. Everyone else is pulling on the other.

[Gabor Giran @GGaborGiran]

a beat this thin on a company this big is basically the guidance call doing all the work. nobody trades AVGO on the print, they trade the number Hock Tan says out loud about custom AI chips forty minutes later.

i’d watch that XPU commentary. every hyperscaler designing its own silicon is a customer quietly building an exit from Nvidia, and Broadcom is the one selling them the shovel to dig it.

[Gabor Giran @GGaborGiran]

the $3 million salary is the decorative number. $55 million in equity is the contract, and it’s the part that tells you what the board actually wants.

what i’d want to see is the split between time-vesting and performance shares, and which metric the performance half runs on. Apple has historically used relative TSR against the S&P, which is a fine measure of the stock and a poor measure of whether the AI transition worked. pay the man on total return and you’ve paid him for buybacks, which Apple can do without inventing anything.

honestly, second-largest company on earth, and $58 million is roughly two hours of market cap movement on a quiet day. the incentive question is real, the size of it isn’t.

[Gabor Giran @GGaborGiran]

the interesting side of this deal is Oracle’s, not HPE’s. Oracle sold enormous amounts of future compute and now has to physically build it, and building it means buying racks from whoever can deliver on a schedule. that’s a nice contract for HPE and a reminder that OCI’s backlog is a capex problem wearing a revenue costume.

i’d separate the two economics. Oracle books multi-year contracted revenue, HPE books hardware once. one of those compounds and one of those ships.

so the number to follow isn’t the partnership, it’s Oracle’s debt issuance. the buildout is getting financed somewhere, and that’s where the strain shows up first.

[Gabor Giran @GGaborGiran]

OPEC is a cartel. That's not an insult, it's the definition: a group of producers agreeing to limit supply to hold the price up.

If a group of American companies did this with petrol, executives would go to prison. When countries do it, it's a summit with a group photo.

The genuinely interesting part is that cartels are unstable by design. Everyone benefits if everyone cuts production. Everyone benefits MORE if everyone else cuts and they quietly don't.

That's a prisoner's dilemma with oil tankers.

Which is why they keep needing emergency meetings.

[Gabor Giran @GGaborGiran]

Every single year somebody announces the dollar is finished.

Then you check the numbers. Just under 60% of global reserves. Most of global trade invoicing. A dollar on one side of the overwhelming majority of currency trades.

The dollar isn't dominant because America is popular. It's dominant because there's nothing to swap it for. The euro has a structural flaw, the yuan has capital controls, and you cannot wire gold to a supplier in Vietnam.

Being the least inconvenient option is an absurdly durable position.

The dollar isn't winning. Everyone else keeps forfeiting.

[Gabor Giran @GGaborGiran]

a clean beat on both lines, and the part that decides whether it matters is nowhere in the print: the mix. HPE sells AI servers at thin margins because most of the bill of materials is somebody else’s chip, so revenue can beat while the company keeps a smaller slice of it.

i’d read gross margin and the networking segment before the headline. Juniper is the reason the acquisition happened, and networking is where the actual pricing power in this business lives. servers are pass-through, switches are not.

beat vs whisper number, watch the reaction not the print. if the stock does nothing on a 19% EPS beat, the market is telling you it already knew what AI demand does to the top line and cares about something further down.

[Gabor Giran @GGaborGiran]

Imagine trying to punish your flatmate by paying more rent yourself.

That's a tariff.

The actual flow: a Chinese factory ships a $100 lamp. The American importer collects it at the port and pays $50 to the US government right there. Then sells you the lamp for $150.

The Chinese factory received its $100. Same as always. It has never once seen a bill.

You paid the $50. Congratulations on the trade war.

[Gabor Giran @GGaborGiran]

One company got so large it broke a country's economic statistics.

Novo Nordisk, the Ozempic maker, grew to be worth more than Denmark's entire annual GDP. Danish growth figures now get reported with and without it, because otherwise the numbers stop describing the actual country.

That's concentration risk at national scale. When one company IS the economy, the whole place is one patent expiry away from a very bad decade.

Denmark is currently a pharmaceutical company with a flag.

[Gabor Giran @GGaborGiran]

the phrase carrying the entire document is “non-market policies,” and 19 countries agreeing on the wording is not the same as agreeing on the remedy. that gap is where communiqués go to be forgotten.

what i’d watch is point 5, because it’s the one with a real deadline attached. sovereign restructurings keep stalling on a specific problem, which is that China is now the largest bilateral creditor to the countries needing relief, and it hasn’t accepted the haircut framework the Paris Club members use. so a group of 19 pushing for faster restructurings is 19 creditors asking the twentieth to write things off. that’s the actual dispute, and it doesn’t get solved by being outvoted.

the Hormuz line is worth noting too, mostly for what it reveals. when navigation language makes it into a finance communiqué, the energy premium has stopped being a trading story and started being a fiscal one.

[Gabor Giran @GGaborGiran]

a 22% move on a beat means the beat wasn’t the story, the positioning was. nobody reprices a company a fifth higher because revenue came in slightly above consensus. that’s a short book getting run over and a crowd that had written the AI-loser thesis being told to unwrite it.

the thing that stands out is which layer of the AI stack this is. Snowflake sells storage and query on data, and the bear case was always that models make the warehouse a commodity. a strong print argues the opposite, that more AI means more data being moved, cleaned and asked questions, and someone has to bill for that.

picks and shovels, just a few floors up from the chips. i’d watch net revenue retention over the headline beat, because that number tells you whether existing customers are actually spending more or whether this quarter was new logos and enthusiasm.

[Gabor Giran @GGaborGiran]

“most fearful in two months” is a phrase that quietly admits how short the memory window is. two months ago was July, which was not a crisis, so the bar being cleared here is fairly low.

i’d separate sentiment indicators from positioning. surveys and fear gauges measure what people say, and they’ve been wrong at turning points forever because saying you’re scared costs nothing while selling costs a lot. the useful version is whether hedging actually got paid for, which shows up in put skew and vol term structure, not in a mood reading.

flows follow returns, and fear follows drawdowns. by the time a gauge tells you people are afraid, it’s usually describing last week’s price rather than next week’s.

[Gabor Giran @GGaborGiran]

When people say the Fed is "printing money," picture this instead.

The Fed buys a bond from a bank and pays in casino chips. The chips are real, the bank now has millions of them, but they only work inside the casino. You cannot buy bread with a chip.

2008 to 2015 was exactly that. Trillions in chips, sitting inside the casino. Almost no inflation, because barely any of it reached a shop.

2020 was a different machine. That time the government sent actual money to actual people, who spent it on actual things. Prices moved within the year.

Money in the banking system is not money in the economy. That one distinction explains a decade of bad takes.

[Gabor Giran @GGaborGiran]

lumber is the honest one in the commodity complex. it can’t be stockpiled cheaply, it doesn’t have a geopolitical premium, and nobody buys it as a store of value. the price is just how many houses are actually getting framed.

so a 17% drawdown while mortgage rates sit near 7% isn’t a lumber story, it’s a confirmation. builders order wood weeks before they break ground, which makes this a forward look at starts, and the forward look is soft.

i’d pair it with the ARM data from earlier this week. lenders stretching to make payments work on one side, framing demand falling on the other. those two describe the same problem from opposite ends of the transaction.

[Gabor Giran @GGaborGiran]

Temu spends more to get you than it makes from you.

Free shipping. Free returns. $0.80 phone cases. Ads during the Super Bowl.

That's customer acquisition cost, and right now theirs is higher than anything you'll ever spend there.

The bet is simple: get you opening the app on reflex, then quietly stop being that cheap.

You're not the customer yet. You're the down payment.

[Gabor Giran @GGaborGiran]

the market cap to GDP comparison is the one i’d retire from these threads. it’s comparing a stock to a flow, a valuation of all future cash flows against one year of output, and NVDA sells globally while US GDP is domestic. impressive number, wrong denominator.

what’s actually structural is the 8%. index funds don’t ask whether that’s justified, they buy the weight, and the weight came from the price. so every dollar into an S&P fund now sends eight cents to a company whose revenue concentrates in a handful of hyperscaler customers who are also index constituents. the buying is circular in a way nobody chose.

the index is ten stocks wearing a costume, and this is the one holding the costume up. worth knowing whether your diversified exposure is a bet you’d make deliberately.

[Gabor Giran @GGaborGiran]

agreed, and i think the honest answer is that the real value isn’t in the wrapper, it’s in what the wrapper lets you skip.

collateral is the one that would actually justify it. a tokenized share that can post as margin against something else, instantly, across venues, without a custodian confirming settlement three days later, is a genuine improvement over the current plumbing. same with fractional access to things that are currently gated by minimums.

but both of those require the legal claim to be clean, and right now it isn’t. so we got the part that’s easy to build and skipped the part that makes it worth building.

[Gabor Giran @GGaborGiran]

both sides falling at once is what makes this a different kind of weak. hires down 278k and separations down 265k isn’t a labor market shedding people, it’s one that stopped moving entirely. nobody’s hiring, nobody’s leaving, and the net print is a small number sitting on top of two large ones going nowhere.

i’d underline the quits piece inside that. voluntary quits are the confidence indicator in the whole release, because leaving a job requires believing there’s another one. when that dries up, the unemployment rate can stay flat while the actual experience of the market gets much worse, especially for anyone trying to enter it.

the awkward part is what a negative print does to the rate debate right now, with hikes being priced. a frozen labor market and hot inflation is the combination where the Fed has no good option, only a choice about which mandate to disappoint.

[Gabor Giran @GGaborGiran]

the timeline is the argument, and a timeline isn’t causation, but i’d note that exemption lists are exactly where this question always lives.

what makes tariff carve-outs different from tariffs is that nobody votes on them. the schedule gets written by a small number of people, industry by industry, and the ones that win are almost always the ones with the best access rather than the best case. that’s true across administrations, which is why the mechanism deserves more attention than any single ring.

i’d watch the exemption list as a lobbying scoreboard, honestly. it tells you who has the phone number, and it’s public.

[Gabor Giran @GGaborGiran]

“every bull run” overstates it, and i think the overstatement makes the real point weaker than it needs to be. the carry trade was one funding channel among several, and equities rose for earnings reasons too. but the direction of the argument is right.

what i keep coming back to is that the carry unwind isn’t a decision, it’s arithmetic. borrow in yen, buy something abroad, and you’re short yen without ever calling it a currency position. when JGB yields rise and the yen firms at the same time, both legs move against you simultaneously, and you don’t exit because your view changed, you exit because the margin desk called. the same pain trade again, just denominated differently.

August 2024 was the trailer for this. one week, no fundamental news, and correlations went to one across every asset that had nothing to do with Japan. i’d treat that as the map rather than the exception.