Kll Research

Value Investor | Building a repeatable investment framework | Deep-diving one company at a time & sharing it publicly | 7 phases. 1 process. 0 emotions.

[Kll Research @KllResearch]

$MSFT might be one of the highest-quality businesses on Earth.

After running it through my investment framework, I still wouldn't buy it today.

Here's why. 🧵👇

[Kll Research @KllResearch]

@awealthofcs The more important question is whether anything about your thesis has changed.

Price volatility alone shouldn’t change your decision.

Check often enough to monitor the business, but not so often that the stock price starts influencing your view of the business.

[Kll Research @KllResearch]

The “always invert” point is especially powerful. Investing is often less about finding the perfect forecast and more about systematically asking what could make your thesis wrong. Incentives + inversion + mental models is a pretty powerful combination for stress-testing conviction.

[Kll Research @KllResearch]

The instrument doesn’t make the investment decision — the underwriting does.

Buffett wasn’t “betting on options”; he was pricing risk, demanding a premium, and making sure Berkshire could comfortably carry the obligation.

That’s the part retail investors should copy, not the option itself.

[Kll Research @KllResearch]

@PeterSchiff The tension is the interesting part. Treasury wants lower long-end yields, while a weaker dollar and higher oil make the Fed’s inflation problem harder.

You can suppress the symptom without fixing the underlying fiscal problem.

[Kll Research @KllResearch]

@wallstengine The real test isn’t how many people Prime Air can reach, but whether it can meaningfully lower last-mile costs.

If Amazon can make that economics work at scale, the strategic value goes well beyond faster delivery. $AMZN

[Kll Research @KllResearch]

@ftr_investors I’d add valuation to that already great list. Research gives you conviction, but valuation tells you whether that conviction is worth acting on.

A great business can still be a terrible investment at the wrong price.

[Kll Research @KllResearch]

@InvestingCanons “Money costs money” might be one of the most important concepts in investing. Growth is only valuable when the return on the incremental capital required to achieve it exceeds its cost.

[Kll Research @KllResearch]

@zerohedge Call it what you want, but when Treasury doubles its purchases of long-dated debt and the 30Y yield immediately drops ~10 bps, the distinction starts feeling increasingly semantic. Not QE™.

[Kll Research @KllResearch]

This is the part I’d be watching. Treasury may call it liquidity support, but when the long end comes under enough pressure that the government materially increases its own demand for long-duration bonds, the distinction between “market functioning” and active yield support starts getting pretty interesting.

[Kll Research @KllResearch]

Lo interesante es que el punto de Buffett tiene lógica económica detrás, no solo filosofía de vida: la utilidad de una experiencia como Disneyland tiene una ventana de tiempo (tus hijos son niños solo una vez), mientras que un dólar ahorrado casi siempre puede esperar.

Es utilidad marginal decreciente aplicada al tiempo, no solo al dinero.

[Kll Research @KllResearch]

Great analysis. One part I’d challenge is the assumption that AV operators will want to remain inside Uber’s demand funnel once they reach sufficient density in a market.

Waymo already operates its own consumer app alongside its Uber partnership, and Tesla is pursuing its own robotaxi app. If an AV operator can generate enough direct demand, the economics of giving Uber a cut of every ride become much less attractive.

Uber’s distribution layer could still be extremely valuable — especially for multi-provider coverage, airports, new markets, etc. — but I think the key question is: what keeps AV operators on Uber once they no longer need Uber to acquire demand?

[Kll Research @KllResearch]

Worth flagging scale — Mohamed El-Erian’s take on this exact move was that the purchases are small in both absolute terms and relative to net issuance.

$4bn per operation is a rounding error against trillions in long-end supply, so the yield move today looks more like signaling than actual demand absorption.

[Kll Research @KllResearch]

@jimcramer This is playing out in real time today — Treasury just doubled its long-end buyback size targeting the 10-30yr sector that’s been getting no bids since June, and yields dropped instantly on the news (30yr from 5.26% to 5.18%).

Textbook ‘buy the discount’ move.

[Kll Research @KllResearch]

One wrinkle: ‘lock in low rates now’ assumes rates stay elevated or keep climbing.

If they’re wrong and rates fall from here, terming out debt today at these levels becomes the expensive mistake in hindsight — it’s less a free lunch and more a bet on the future rate path either way.

[Kll Research @KllResearch]

Insider ownership: 0.12% of shares outstanding.

Low, but common at this size. No red flag on insider buying, because there basically isn't any — insiders sell for lots of reasons, they buy for one. Zero buying reads as "fairly valued," not "hidden bargain."

Capital allocation itself, though: disciplined, cash-funded, reasonably scaled.

[Kll Research @KllResearch]

Read $V's latest shareholder letter looking for one thing: candor.

Self-criticisms: 0. External excuses: 0.

Not a red flag on its own — but also a letter that doesn't engage with the antitrust case or fee pressure sitting right in front of it.

[Kll Research @KllResearch]

What the letter didn't mention at all: buybacks.

For a company funding a $20B repurchase program, that's a real gap — you have to go find the numbers yourself in the 10-K. Which is exactly the point of doing this instead of reading the press release.

[Kll Research @KllResearch]

The fine is probably not the part I'd worry about most as an investor. The bigger question is whether regulators eventually force changes to the recommendation systems themselves.

If engagement-driven algorithms have to become materially less optimized for engagement among younger users, that could have second-order effects on how $META monetizes attention.

That's a much more interesting risk than the headline fine.

[Kll Research @KllResearch]

This is the kind of competition I'd actually watch. Not “another GPU,” but a fundamentally different bet on where AI compute economics go next: specialized inference at cluster scale.

NVIDIA can remain the dominant AI platform while still losing a meaningful piece of the inference economics if the performance/$ and performance/watt gap is real.

[Kll Research @KllResearch]

The historical pattern is interesting, but I'd rather ask what would actually make a 10–15% drawdown likely from here than simply assume the calendar will deliver one.

And if it does, the more useful question is which businesses become genuinely undervalued — not whether we called the pullback.