Tech investor for over 20 years. Ran large hedge fund for 10 of those. Here to help. Not investment advice.
3 subscribers
Jan 1 • 4 tweets • 5 min read
Happy New Year.
My occasional reminder that the best gift you can give yourself and everyone around you is the awareness of index ETFs.
99% of investors cannot beat these incredible ETFs that provide exposure to the best companies in the strongest economy in the world, automatically rebalance to add winners / remove losers, can be held forever, and require you to pay little to no taxes.
Numbers updated for 2024.
SPY: 13.7% compound return over 15 years. Trades at 23x forward P/E vs 10 year mean of 19x.
QQQ (Nasdaq 100): 18.4% compound return over 15 years. Trades at 28x forward P/E vs 10 year mean of 24x.
SMH (Semi's): 22.9% compound return over 15 years. Trades at 25x forward P/E vs 10 year mean of 20x.
IGV (Software): 17.3% compound return over 15 years. Trades at 37x forward P/E vs 10 year mean of 35x.
(Please note this is not a recommendation to go buy ETFs today at these prices, but just trying to create awareness for the inevitable pullbacks).
How to think about buying indexes when trading at above average multiples?
The charts below plot the S&P’s forward returns over different time horizons versus the starting P/E.
As expected, there is a negative correlation indicating higher multiples correlate with lower forward returns.
There is little correlation between valuation and 1-year returns, and also plenty of cases where cheap multiples still resulted in negative returns.
Also worth noting that the SPY produces negative returns over 5 years anytime it is trading above 22x in this time frame (we are at 23x today).
Not sure exactly what to take from this other than to perhaps be a bit less valuation sensitive particularly for long-term ETF investments. And a reminder to buy ETFs aggressively when they are cheap (<15x).
SPY correlations since 1999
-1 year = -0.27
-2 year = -0.38
-3 year = -0.34
-5 year = -0.43
Dec 7, 2024 • 18 tweets • 15 min read
For kicks, I read Senator Rubio’s 58 page report on China. It was surprisingly well done – current, thorough, and balanced.
It was complementary of China’s progress across many fronts, and frankly was not particularly combative – more focused on what the US needs to do to be competitive (focus and subsidies).
Lots of excerpts in this thread covering the report's assessment of Chinese semis, EV's, nuclear, research capabilities, robots, etc.
$KWEB trades at 9x EPS excluding cash. And the President Elect has made it clear we are not going to war so left tail risk greatly diminished. Hmmm.
Background
- China released “Made in China 2025” document in 2015. "At the time, they were already the worlds’ factory across textile, apparel, furniture and other sectors."
- Of the 10 sectors targeted by the government’s 2025 plan, China can credibly claim to the be world leader in 4 (EVs, energy and power generation, ship building and high-speed rail). In 5, China has made substantial progress but is not yet the leader (aerospace and aviation, biotech, new materials, robotics and semis)
- China brutally honest in self-assessments: “Manufacturing industry is large but not strong. Capacity for independent innovation is weak, and key technologies and high-end equipment are highly dependent on foreign countries.
- New slogan introduced “new quality productive forces” by Xi. Documents indicate state will promote technologies including humanoid robots, quantum computers, brain-computer interface, 6G network equipment, deep-sea drilling and mining tech, and advanced aviation equipment.
- While China is still massive producer of cheap toys, apparel and electronics, it is now among the world’s most competitive producers of just about everything – from machine tools to dock cranes.
Report's Conclusions
-“US must act now to avoid living in a world where China is the teacher, and we are the pupil”
- It may be the case that China’s export- and manufacturing-oriented development model has been successful enough to propel China to the technology frontier in the short term, but not successful enough to help the country outrun its structural problems in the long term.
- Instead of being complacent, the United States must act urgently to shield itself from China’s predations and reboot its ailing industry. This will require a bold industrial policy to support sectors critical to our national and economic security, paired with an equally bold deregulatory agenda to clear the path for industry to build again
- We should be circumspect enough to realize that, in many areas, the days when China needed to steal from us are past. Now, in sectors as diverse as shipbuilding, EVs, and energy, China leads the rest of the world. Already, Beijing is deploying new resources to lock in its advantages in these sectors, and to use its technology and trade to remake the world according to its interests.
[None of this sounds outlandish or warmongering to me - I tried to include the most "incendiary" language I could find - mostly sounds like respectful competitor discussion]
Semis
- China remains deeply reliant on foreign countries for key inputs, including lithography equipment, Electronic Design Assistance software, and advanced materials.
- Chinese semiconductor manufacturing companies like SMIC are pushing the boundaries of what is possible in the face of export controls, but are several generations behind the leading edge.
- It also runs a substantial trade deficit in semiconductors, relying on imports, mostly from Taiwan and Korea, to meet the voracious demands of its internal market and manufacturers.
- China has invested heavily to develop an indigenous semiconductor industry, spending potentially more than $150 billion in the past decade
- Recently, Beijing has concentrated control of its semiconductor investments in a secretive government-run working group, which indicates the importance the CCP places on semiconductor technology and its desire to avoid further sanctions from the U.S. government
- YMTC appears to be succeeding despite a U.S. government blacklisting, reportedly producing the world’s most advanced commercially available NAND memory product.
- China’s premier manufacturer of logic chips is Semiconductor Manufacturing International Corp. (SMIC), a state-owned firm with close ties to the Chinese military. SMIC has also battled U.S. sanctions with considerable success. Multilateral export controls on semiconductor manufacturing tools have cut off SMIC’s access to the most advanced chipmaking equipment produced by firms like ASML. In response, SMIC has pushed the boundaries of what is possible with older equipment. Last year, Huawei stunned industry observers by releasing a new smartphone powered by a SMIC logic chip at the 7 nm node of production
- Despite these advances, China remains substantially behind the cutting edge in virtually all areas of semiconductor manufacturing. In particular, China lags in lithography equipment and EDA software and remains reliant on a handful of Western firms for such products.
- SMIC and other Chinese chipmakers have been on a construction spree of fabs producing less advanced chips, which go into automobiles, airplanes, and countless weapons systems.
Oct 28, 2024 • 12 tweets • 9 min read
Consensus is that Chinese demographics are “terrible”, so I double-clicked on the numbers.
I'm not a demographer (so discount this work by a lot) but seems like the "demo" concerns are overblown and the Chinese economy will grow regardless over the next 10-15 years minimally.
Highlights
-US / China populations will be relatively consistent for next 10-15 years.
-The shape of cohorts by age suggests the workforce also will be consistent. China is actually increasing the number of people entering the workforce each year over the next decade whereas the US will see a slight contraction.
-The main Chinese red flag is birth rates have been depressed since 2020 – potentially due to COVID / economic malaise / trade wars. If this persists, then the population will contract down the road. But China is also a command country and might be more dynamic in alleviating the costs of having children.
-The migration to cities from rural areas continues in China which is a tailwind to GDP.
-The education system has expanded and China is growing the number of college and graduate students annually – tailwind to GDP.
-Real GDP / capita in China is growing at 2x the rate of the US. Chinese real GDP will potentially grow rapidly even w/ demographic headwinds relative to the US.
-The percent of the population over 65 years old is 15-18% in China and 25% in the US.
-In the US, people aged 65-74 spend $47k / year – less than middle-aged people bringing up families, but similar to people aged 25-34 who spend $48k / year. So if age cohorts are consistent (which they largely are), not that much of a spending drag from people aging.
-US savings rate is <5% while Chinese savings rate is >40%. So could be spending, even with an aging demo, is more insulated as the elderly in China spend both for themselves and descendants.
1. Total Population
China has 1,426m people w/ growth of -0.2% yy in 2023.
US has 340m people w/ growth of 0.5% yy in 2023 (1.75m increase in population. CBO estimates 3.3m of these were immigrants implying organic US population declined; SSA estimates 2.0m were immigrants so also implies US declined organically; Census estimates 1.14m which implies slight organic growth)
India has 1,429m people w/ growth of 0.8% yy in 2023
So China is flattish, US is flattish but growing a touch via immigration, and India is still growing.
Mar 11, 2024 • 7 tweets • 9 min read
The $GOOGL bear case is pretty loud right now (Search monopoly facing first real threat from GenAI).
Predicting the future here is tough but worth at least exploring a strawman for a more constructive outcome in search for $GOOGL.
Some thoughts and search comparisons:
#1 : Search queries are growing. Some have stated this is not the case but pay attention to footnotes in these exhibits ("does not represent any specific data" ?). Last week, $GOOGL Chief Business Officer stated "Search query growth has been consistent across all major markets over the past 12 months". This is during ChatGPT's meteoric rise...
#2: GenAI searches are neither 100% cannibalistic nor 100% new so question is how additive they might be. GOOG blog post from 8/2023 suggests they are finding GenAI searches useful for complex questions they wouldn't typically think to search. See image.
#3: Impact of GenAI on monetization remains unclear - certainly could be negative, but also might be positive. Bears state 7 blue links goes to 0-2 in GenAI and that's that. Sure, but those 7 links were presented to provide the consumer with sufficient choice to better meet their need. Now, the search provider has a more detailed search request and can narrow the result further - and offer the advertisers a more qualified search. So if the 7 links previously paid $1.00 in aggregate to reach a prospect at a certain point in the buying journey, why wouldn't 2 links pay the same aggregate amount to reach the same, but more qualified individual (Nike will pay X for "shoe" but a higher price for "running shoe for trail-running", and against only one other option)? Also, GenAI searches tend to be iterative. So maybe 2 links over 3 iterations results in 6 ad slots overall, all of which might be worth more than 7 generic ad slots given the increased targeting [Again, I don't know the answer - just trying to highlight it's not that simple].
#4. $GOOGL has a pretty notable advantage in currently being the default search status across Android, Apple, Chrome, etc. Users don't change their behavior unless an alternative offers a meaningfully superior proposition. $GOOGL has an advantage in that it is a directory (get me to Nike's website), it is a concierge (tons of info for various searches on the first page like weather, math, Football schedule, etc.), it can get you to resources you need quickly, and now provides GenAI results on the first page as well. So a positive view would be that, so long as GOOG GenAI results are at least reasonably competitive with others, users will just default to $GOOGL. It's just all there for you. Will be interesting to see ChatGPT user / query growth going forward to see if their answers are sufficiently different / better than $GOOGL to keep their momentum going.
#5. $GOOGL has advantaged data which provides better search results. Local data, real-time data, user data (gmail, search history, maps, Android) are all real advantages. Data suppliers also tend to update $GOOGL first given its omnipresence (restaurant updating hours for holiday, etc.).
#6. Commercial searches are some small percent of total. The most expensive keywords include insurance, loans, mortgage, attorney, credit, lawyer, etc. (see image). A table from the EU anti-trust trial indicates the largest revenue generators include iphone 8, auto insurance, car insurance, online colleges, etc. As the exercise below will indicate, GenAI today can help users conceptually think about how to shop in a given category but is pretty far from helping users research a specific vendor, interact with the vendor, buy the thing. Again, I would expect GenAI to develop that functionality over time but worth noting $GOOGL has that today, and also will be working on enabling GenAI to assist.
#7. Desktop vs Mobile. $TRIP always had issues because users would research a trip on mobile, and then just go directly to $GOOG on desktop to actually make the reservation. Feels like today a similar dynamic may exist. GenAI can provide some suggestions today, but then you have to go to $GOOG to see the restaurant's current menu and book a table.
Ok, let's go through some examples. I'm using the most current version of ChatGPT and Perplexity.
"Can you help me buy car insurance"
They all seem pretty comparable to me except with $GOOGL you can iterate / continue your shopping journey by searching for "car insurance" which shows you a page of ads.
GOOGL: Pretty good result page - sort of a literal interpretation of how to mechanically get it (phone, online, etc.).
ChatGPT: Good result more focused on the important consideration points when shopping for insurance.
Perplexity: Someplace in between. A bit cleaner layout.
Feb 29, 2024 • 7 tweets • 6 min read
$CRM / @Benioff reported consistent growth in-line w/ the street w/ expanding margins ahead of the street. Story is getting better on improved execution / AI and numbers should be moving higher over this year. Risk / reward is decent here. Upside of 30% to 44x my CY2025 GAAP EPS / 25x FCF, and 56% upside to same multiples on CY2026. Downside of -15% to 20x street CY2025 FCF.
Qtr was fine:
Bookings indicators: cRPO grew 13% yy cc which has been consistent > guidance for 11%; current deferred grew 51% qq (which is seasonally weak relative to typically growing 55-59% in the biq Q4) / 9.4% yy down from 11-12% earlier in the year.
Subscription rev of 8.75b grew 12% yy > street at 8.6b which has been consistent w/ net rev delta of 607m higher than 556m last Q4 and 449m the Q4 before that.
Sales cloud is 21% of subscriptions and grew 10% yy which was consistent / slow decel, service cloud is 23% of subscriptions and grew 11% which is also consistent / slow decel, marketing cloud is 14% of subscriptions and grew 7% which was a slow decel (was high-teens in prior year and 10% earlier this FY), data cloud is 19% of subscriptions and growing 21% yy cc consistently, and platform / other is 20% of subscription revs growing 10% yy cc which is a slow decel.
Company highlighted their success cross-selling w/ 8 of top 10 deals containing 6 or more clouds, and highlighted big deals w/ 78% growth in deals worth over $10m for the year.
Growth by geo was also consistent w/ US growing 9%, Europe 11% and APAC 19% (highlighted India growing net new business 35% yy).
Sales productivity has been improving and 0.76 in Q4 similar to 0.76 last Q4 and up from 0.63 the Q4 before that.
One perplexing disclosure was that <50% of sales cloud customers also buy service cloud (would think that would be an easy / natural cross-sell – could be isolated purchases are more prevalent at low-end of customer base), and that vice versa was true as well. They pitch that as opportunity but seems like suboptimal execution to me (should have a more compelling product integration value prop to customers at this point and thus more customers buying both solutions).
PF OM of 31.4% was > street at 31.6%. SBC of 7.2% was down nicely from 9.4% last Q4.
Shares outstanding decreased 1% for the year from the buyback.
FCF of 9.5b for the year was 27% of revs up from 20% the prior year and grew 51% yy.
Nice job on cost control and capital return.
Feb 28, 2024 • 6 tweets • 7 min read
$SNOW reported an ok quarter (strong bookings but compromised on pricing, decelerating revenue growth, subdued customer addition metrics), guided well-below, and announced CEO change w/ new CEO lacking ANY enterprise software experience. Story is getting worse,and numbers are going down (though likely bottom here for a bit). Very expensive valuation.
Upside using even stale street numbers is -7% using 12x CY25 revs / 42x FCF and 19% using same multiples on CY26. Downside of -50% using 8x my CY24 revs / 40x FCF and -40% using same multiples on CY25.
New CEO: The CEO change is not that surprising as Slootman had been running the company from Bozeman (and remote work is not this guy’s style) but the timing does seem abrupt. New CEO is not from central casting – 15 years at GOOGL primarily in engineering but then ran Ads / Commerce for 5 years. Then cofounded Neeva which was a consumer-focused search start-up he ran for 4 years. Slootman is an enterprise software carnivore, and the new guys has modest relevant experience. He still has “Learning Snowflake” as his current role on LinkedIn. Oh boy. Doesn't strike me as normal CEO change so either CEO did something untoward or board fired him for basically missing the AI boat and now having to play catch-up from far behind.
CFO did re-commit for another 3 years.
Feb 21, 2024 • 9 tweets • 4 min read
$LCID missed and guided below. Deliveries, ASPs, margins, FCF and liquidity are bad, bad, bad, bad, and scary respectively (and respectfully). (1/9)
Deliveries -> Bad. 1,734 were up 19% qq but missed guidance of 1,800, and well below production of 2,391 in the qtr. This is a new car company in a scale game, so they need deliveries to be exploding right now for this to work. Deliveries for the year were 5,997 up 37% - they need to be growing >100% at this small scale. (2/9)