$MSFT has said to think of capex growth in the mid-teens % in FY26 and for long-lived assets (datacenter shells, HVAC, Fiber, Power, etc.) spend as % of that capex to revert from current elevated ~60/40 ratio to as much as ~30/70 in favor of servers/semis in 2026+. The simple math of this makes its really hard to see how industrial revenues benefitting from the boom in the "long-lived assets" category can grow in 2026+, at least w/ $MSFT (but ratio of spend elevated elsewhere too). Simply put, MSFT has been double-ordering datacenters in L24mo to catch up to demand, so that category of spend will cycle earlier and much harder than semis and can decline severely even if MSFT capex continues to grow. I'm not sure industrial sellside covg. or mgmt teams have listed to the MSFT call or are truly thinking through what this means, bc i havent heard any caution on '26 growth decel. Rather, everyone continues to point to overall hyperscale capex growth in '25 as defense for the trend that has lifted stocks recently and saying the Deepseek selloff is an overreaction. While investors are beginning to be more visibly concerned, valuations and financial results (growth rates / margins) are still at hugely inflated levels. If DC revs broadly decline in '26 as MSFT math suggests is possible (likely?), DC-exposed industrial stocks could be down a lot from current levels. As good as shortage economics are on the upside, glut economics are absolutely brutal on the downside. And markets tend to overshoot either way on multiples. Here are two hypothetical scenarios for $MSFT capex (very high-level for simplicity) to think through implications for DC vs semis spending. Starting w/ now fully-guided FY25 capex of $88B and assuming +15% growth in Fy26 (in-line w/ IR's mid-teens framework from callbacks) and +10% in Fy27 (IR suggested a trendline decel) total MSFT capex could reach $111B in FY27. If DC mix went from 55/45 in FY25 to 30/70 as $MSFT IR has suggested, DC-related spending could decline significantly (30%+) from '25 to '27 even w/ growing overall capex. In a bear case where MSFT spend decelerates more quickly than expected and outright declines in FY27, DC revs could get cut nearly in half over 2yrs. Implied CY26 semis rev could still grow nicely in either scenario (note FY vs CY dynamics here for MSFT June year end). For $NVDA specifically this would be bullish but there is risk of share loss to ASICs that might cause them to under-grow these scenarios, plus idiosyncratic issues like China import bans, but this general framework is why ASICs stocks have performed better than $NVDA post $MSFT and $META earnings (plus META emphasizing broader use of ASICs in '25 and '26 for inference then training).

Pretty interesting data from Michael Nathanson today on $NFLX having lower content spend efficiency (in viewership per hour of content available) than $DIS DTC platforms. $NFLX has a lot of hits, because they put out a lot of content, but ultimately achieve a lower “hit rate” Paired with the additional windows that $DIS/$WBD films get, if a producer or star wants to get reach on a project, $NFLX isn’t the automatic destination of choice…

$RBA is the mkt leader (w/ a price umbrella) in a mature mkt where competition is ticking up. This is structural & why no CEO has been able to sustainably re-accel the biz for 15yrs. What looks like accel is just more risk-taking inventory purchases, w/ some tailwind from price. RBA made better than avg. inventory margins in '22 but incremental returns here are a headwind (easier to find $500M of bargains than $1B), & even at elevated margins inventory is much worse biz vs. auction/mktplce (10% GP on GTV vs. 14%) w/ more downside risk & capital intensity

“We think the synergy is much higher, but we’re not going to start announcing numbers. We’ve given numbers to the street, & in the last 8yrs we’ve only missed one number by 1%. One of the things about us is we say what we’re going to do, and we do it.”

open.spotify.com/episode/6lzw4w… “The goal for us now is to make and exceed every one of our numbers. And if we emerge and generate 6,7,8 Bn of FCF with great IP, this company is going to look and feel very different than it does now.” - David Zaslav $WBD

$ADSK isn’t cheap.

Nor should a quasi-monopoly growing 18%, with 92% Gross margins, enormous embedded income statement investment, and which likely grow substantially >GDP w/ high contribution margins until my 1yr old has a college diploma, be.

https://twitter.com/napkincapital/status/1464317558618742784

Adsk is ~24x FY2 FCF. Yes, much of that is deferred rev, & ~25% will be SBC

Beyond next yr, essentially all FCF growth will come from earnings

SBC has blown out in recent yrs, but this is partially due to a lot of m&a & headcount growth, and it will get operating leverage

$AMZN is changing the way they purchase servers, and this will have a material impact on FCF for the next few years. But its nothing to worry about. $AMZN typically finances server purchases w/ leases. While some people refer to CFO - Capex as Amazon "FCF," $AMZN doesn't & neither should you. This would ignore the very real economic cost of buying servers for AWS which is ~25-30% of AWS sales. FCF is the bottom line below:

Pinterest head of engineering discussing multi-cloud: by keeping AWS services used at the IaaS layer, they leave it open to go multi-cloud or to repatriate eventually. Harder to do so when you adopt the clouds PaaS or special services.

theinformation.com/articles/how-p… "If you stay at the IaaS layer you can take advantage of specific capabilities from other providers. Multicloud CTOs aren't load-balancing the same application on 2 clouds. They usually say we’re going to run this set of services on this cloud, & this other set on another cloud."

“$ADSK is so critical to the sectors it serves that most burgeoning engineers, architects or contractors are trained on its programs in college. In a field where coordination between different professions is required, the work mandates a de facto industry standard: that's $ADSK.” Great summary of the value Autodesk has provided to constituents for decades, and now it is repositioning for the cloud era.

www-protocol-com.cdn.ampproject.org/c/s/www.protoc…

Buffett, Malone, Diller all long in the tooth. Who are some great dealmakers still under 60?

- Andy Florence, CoStar
- Neil Hunn, Roper
- Jensen Huang, Nvidia

Dark Horses
- Josh Silverman, Etsy
- Larry Culp, GE
- Jeff Lawson, Twilio

Zuck’s likely done doing deals

Who else? H/t @soonervaluecap

$ADSK just acquired Upchain, a cloud PLM & Data management company. After mostly focusing on building up, integrating, and marketing the construction SaaS business in @andrew_anagnost first few years, the company has now made two acquisitions in manufacturing in the last 12mos. I love that ADSK has the confidence & awareness to disrupt themselves. On this slide, "Traditional Mfg. Processes are Inefficient," one of the logos is their own.

w/ F360 & now Upchain ADSK is trying to build the common data environment for cloud-based mfg. design & engineering.

ROIC matters. $INTC announcement is much better r/r for semicap than Intel. Why? $INTC takes the risk, Semicap has certainty of high-ROIC, aa'l installed base to drive services, gets paid on front end. INTCs ultimate result will be delayed and could be low ROI or even big losses. $20bn is a lot of capital, and if EU build is similar-size, earning attractive ROI on $40bn will require near perfect execution and significant market share for Intel Foundry Services. There are several reasons why INTC's foundry might not look at financially sexy as TSMC's:

Great interview by @AndrewRangeley of @borrowed_ideas on $ADSK.

Here is my response to Andrew’s two primary concerns:

podcasts.apple.com/us/podcast/yet… His biggest concerns were:
1) Anagnost didn’t orchestrate the SaaS transition, he just benefits from it already starting before his tenure. So is he truly proven?
2) R&D spend stalled for years during the transition

Investing wisdom Life wisdom

After-tax IRR determines net worth/spending power.

If >100% turnover (<1y avg. hold period) can add 400bp+ of add'l performance, more power to you. For most people I'd run high turnover ideas in tax-free account and have a high bar for taking gains <a year in taxable accounts: I sketched a quick illustration comparing 3 strategies over 10y: 1st an index fund earning 10%, 2nd a 100% turnover portfolio earning 20% gross IRR, 3rd a 20% annual turnover portfolio that earns 15% gross IRR. Each portfolio has a 38% marginal tax rate, NYC state and local taxes

Broken clocks are right twice a day and Jeremy Granthams perma bear calls are right once a cycle. A timeline and cautionary lesson for investors:

2010: Grantham sells stocks out of the financial crisis, saying stocks are now overvalued: cnbc.com/id/40131748 2014: We’re in a stock bubble and it will end badly in 2016: thinkadvisor.com/2014/05/04/gra…

2018: The market is a bubble and will soon crash 50%: institutionalinvestor.com/article/b1743d…

Bull case for branded hotels w/ successful loyalty programs ( $mar, $hlt )

Airbnb pressures RevPAR at low end, esp. unbranded hotels. These units are dependent on OTAs to drive traffic, & w/ inflating labor/marketing costs, damaged balance sheets, they continue to be squeezed This increases the relative value of branding hotels. Compared to an independent, your ad, tech, and loyalty spend gets benefits of scale. You tap into the power travelers in loyalty programs, attractive economic splits with OTAs, a great mobile app, streamlined operating model.

Below is a long thread on 7 key levers that will drive growth at for the next decade. Investors have been fixated on mgmt.’s guided $2.4bn FCF by Jan ‘22, but I believe there is vast opportunity beyond that point. First, a summary of the drivers. I will expand on each one: It may not be immediately obvious why an already dominant, mature CAD business is an attractive growth story 40 years after its founding. Hopefully this thread makes the thesis clearer. This is just a fingerpainting exercise to size up some of the important initiatives underway.

Uber Freight w/ $500mm raised from a group led by Greenbriar, a transpo specialist which owned Transplace.

At $50mm/quarter burn rate, this gives Uber 2-3 more years of aggressive run-rate to keep selling freight at cost of PT.

Not good for the $CHRW, $JBHT, $LSTR of the world google.com/amp/s/www.frei…

on getting too big:

“We decompose the growth drivers and our growth algorithm over the last 17 years, it's been basically 19% TSR. And the organic revenue growth, the organic operating cash flow growth, the M&A layer, and then improvement attributed to better CRI,” “As we look through that model and apply it forward the next seven years, that's just the planning horizon that we work on, there is – we're deploy – have to deploy about $2.5 billion a year. Right now, at the end of seven years, it's about $3.5 billion.”

1/ "And I do think there will be some more flexibility on whether we all go to the office every day when we're not traveling and we'll see people that can sort of further mix, to some extent, work and leisure. I think there's a piece of that which will be good for us." - MAR 2Q20 2/ "So, imagine that a year from now or two years from now that week in Florida or a week in the Caribbean, which would have been 100% vacation and I could only do it once a year, I might be able to do twice a year now because I can go down there for a week"

1/x Touring some historical sites this week and unexpectedly stumbled on a nice business lesson in disruption theory.

This is the now-defunct Chesapeake and Ohio (“C&O”) canal. It is 185 miles long stretching from Washington DC to Cumberland, MD. 2/x Construction of the C&O canal began in 1828, 5 years after the completion of the famous Erie Canal which inspired it. The Erie was such a success that businessmen south of New York wanted to copy its success and connect Eastern cities to the Midwest.