Dan Taren

Co-founder of Scrimmage (acquired) & former BofA analyst

[Dan Taren @DanTaren]

Iq test

[Martin Shkreli @MartinShkreli]

Rumors:
SALP down more than 50% MTD
SALP was up 200% YTD and is now down more than 30%
GSCO has liquidated some SALP positions

Source: Godel News
*want to be clear we wish SALP well and hope all of our clients and non-clients do great. just reporting what we're hearing.

[Stock Talk @stocktalkweekly]

What Microsoft tried to do on this earnings call is clear-cut:

They understand everyone is concerned about the pricing-power for frontier labs in light of the open-weight trend & the emergence of much more cheaply-built, yet still competitive, Chinese models. They understand the frontier labs' ability to meet spending commitments is in question as a result of that (at least, until they figure out better business models).

Microsoft wants to distance themselves from the frontier labs for this reason. They explicitly note that their cloud-revenue is 90% "non-frontier model customers". They also explicitly note that their +$50 billion backlog, growing +25% is "excluding OpenAI". The language behind these statements is not ambiguous. Microsoft does not want to be connected to frontier lab risk.

[threadguy @notthreadguy]

everything trades like crypto now welcome to hell

[Hamid @hamids]

🤣 For those who think Michael Burry is a genius right now because he shorted $MU and $NVDA, based on a few days of market downturn, you might want to consider this:

[Dan Taren @DanTaren]

@metaproph3t @ripcarsio Nice, thank u

[Dynamicmoats @DynamicMoats]

The whole "the market will force hyperscalers to cut capex" argument is a bit perplexing to me

When there's existential race to your co., you're not going to give up in the early to mid innings of a race,
especially when your stock price is:

$MSFT flat since 2024
$GOOGL up a lot
$META flat since 2024
$AMZN flat since 2024
$SPCX Elon chooses his strategy, not his stock price

The stock prices aren't even down 50%

No CEO is planning existential business strategy from a price action of "15% drawdown from ATH"

Not to mention all the metrics they're tracking are all very strong

Doesn't mean semi stocks are all pound the table longs as they're exposed to the 2nd derivative

[Tyler Neville @Tyler_Neville_]

Over the next ten years America will realize the repercussions of boomer political force on housing was the inevitable castration of their lineage.

Boomers sacrificed grandchildren so they could keep their cost of capital low and their equity high.

[Brett Erickson @BrettErickson28]

As reporting is circulating this morning about potential discussions for a new deal between the United States and Iran, I want to discuss one of the core problems to what I am seeing.

Trump isn’t being realistic… at all…

Based at least on the initial reporting, President Trump is trying to essentially strike a deal that reverts back to the terms of the original Memorandum of Understanding. Put nicely… that’s ridiculous.

Unfortunately for Trump, unfortunately for everyone, the Trump Administration took guidance from the Lobby and Hawks that called for Washington to revoke the oil sanction waivers (General License X) that were issued under the MOU. Even worse, during the 10-day wind down period, Treasury added an escrow account mechanism. This means that any oil sold during this window would be controlled by the United States. Of course, Iran would never willingly do that, so it was essentially torched without notice.

This is significant because the oil waivers were the only real concession that Iran received up front in the original MOU. Because the Trump Administration revoked them without notice, they cannot simply “reissue them” again in a future MOU. This is because countries that MIGHT have purchased Iranian oil under the original deal know that the US is willing to revoke them without notice. Therefore, there is no durability anymore, and they effectively lose all value as a future concession.

On top of this, the United States also never released any of Iran’s frozen assets during the 27 days the MOU was in effect, and this only further degraded any trust that the United States will hold up their end of the bargain in any deal.

To take it a step further, before the MOU was “officially” destroyed, the United States imposed new sanctions on Iran, which explicitly breaches Article 9 of the MOU.

President Trump putting forth an “offer” that essentially reverts back to the prior terms of the original MOU… that’s simply delusional. He knows it. Iran knows it.

It’s not a serious offer.

The problem with this though, is it fundamentally guarantees that if Iran holds the cards, which I think most everyone understands that they do, President Trump WILL be forced to agree to far worse terms under a future deal than he received in the original MOU.

As much as I hate to say this… the odds the Iranians accept these terms are zero. Nada. Zilch. It’s an entirely unrealistic set of terms from Trump, and this will only signal to Iran that the United States is not serious, and not willing to make a real deal to end this war.

[Dan Taren @DanTaren]

@Biotech2k1 Why would they go bankrupt? Their legacy businesses would also go to zero? Lmfao

[Banana3 @Banana3Stocks]

$GOOGL

Being punished for spending more money

This capex spend is utterly misunderstood as half the analysts raised price targets and the other half lowered price targets

In 25 years of doing this it’s the most bifurcated markets I have ever seen

As the Blackstone CEO just said, many investors are treating this like the telecom boom where they overspent money on overbuilding in hope of demand, the case here is demand is so strong that they continue to heavily build to capture already high demand

$AMZN and $META being hit from the pin action in $GOOGL in worries that there capex spend is going up as well

The spenders are being punished and the vendors are being rewarded

I posted a piece recently saying that $AMZN and other hyperscalers are not the spenders, their customers are as they are demanding more compute from them and companies like $AMZN are charging them and will pass along higher cost and collect a toll for the service

There’s times where the market has schizophrenia and this is one of those times

[The Long Investor @TheLongInvest]

I want to be clear

If the market is irrational enough to allow $GOOG to drop to $250

You take full advantage and build your portfolio around this position.

[Dan Taren @DanTaren]

@NoahRubin22 Fox at 90 is an unfathomable overreaction to the playoffs lol

[Dan Taren @DanTaren]

@zeroxkyle I think the market considers capex a call option that if it works fantastic and a high multiple makes sense. If it does not work then cut capex and you are back to being a high FCF and high buyback mag7.

In what scenario do the mags do high capex forever w minimal ROIC?

[Dan Taren @DanTaren]

@The_RockTrading what exactly do you mean by “bad”

[Jukan @jukan05]

BofA’s Vivek Arya made an excellent point, which I’d like to share.

Is open-source AI bearish for memory?

Closed models amortize global demand across shared HBM pools concentrated in a handful of data centers, whereas open models create a new memory footprint with every deployment. If 10,000 companies self-host the same open model, the model weights must be replicated across 10,000 separate HBM pools, with each deployment also requiring its own KV cache. As 128K–1M token contexts become commonplace in 2026, the KV cache alone can exceed 40GB per active session.

Low-cost Chinese APIs drive greater inference demand, broader enterprise self-hosting, and more memory sockets worldwide. In short, closed models concentrate memory demand, while open models multiply it.

[Dan Taren @DanTaren]

@evrgn11112231 @riddle245 @blknoiz06 they sure do

[Kyle Harrison @kwharrison13]

https://t.co/KFLDnaWWmP

[Dan Taren @DanTaren]

@Midnight_Captl there’s no incentive for those analysts to be anything but conservative, you kinda are forced to be

[Dan Taren @DanTaren]

Meta can’t lose. If they compete on frontier models, awesome. If they lose and cut capex, awesome. If open source wins and they are the orchestration layer, awesome, + they are then a pretty solid neo-cloud

Really cheap on fwd multiples, it’s free on a long enough horizon. The discount it trades at due to Zuck is insane