212: Snowflake Q3, Cloud Scale, Spotify vs Comedians, TSMC's Morris Chang on Intel, Chinese Billionaires, and California's Last Nuclear Power Plant
"...appeared out of nowhere, Beetlejuice-style."
To assess which journeys are worth taking, you must first understand what destinations are desirable.
🛀 Knowing “the right answer” and “what’s going to happen” in theory vs having the bullets fly by your head…
The real stuff always kinda feels different.
⚫️ 🚫 What’s your personal decarbonization plan?
I mean, countries have ‘em, companies have ‘em, why not you and I?
I’ve never thought about it in quite those terms before, but I think it’s a neat project, and brings clarity to a bunch of stuff I may otherwise do piecemeal.
For example, my next big step is getting an electric car.
My current vehicle is from 2010, two years before the Model S came out, so there wasn’t much available at the time. But my next one will be electric for sure.
What else? What are the biggest contributors to my personal/family’s carbon footprint?
Biggest one is definitely heating our house and water. We have a natural gas furnace and natural gas water heater. I’ve been looking into converting the water heater to a high-efficiency electric one when it’s due for replacement, but the furnace is a bigger challenge.
I suppose I could also convert it to an electric furnace, since where I am electricity is 95% hydro and 4% wind. It would be a big improvement on gas.
But hey, I’m an optimizer, I like to go the extra mile… So I’ve been looking into closed-loop ground-source heat pumps.
They’re also powered by electricity, but because they *move* heat rather than *generate* it, they can be much more efficient (for every unit of energy you put into it, you get ˜4 back), and they act as both A/C and heat, and are silent unlike the more conventional air units.
🛀 📝📝 🌏 We should make two lists:
One for the most important problems in the world.
The other for the problems that the most brains are working on.
Next step is to compare the two lists and see what’s being under-served.
💉 👹👹👹 If you want to see a technique used by antivaxers on social media, check out the replies in this thread.
I posted about my kid getting the COVID19 vaccine, which I’m very happy about and is a totally innocuous thing (it’s probably the 8th vaccine he gets in his life, or in that ballpark).
Vaccines are objectively one of humanity’s greatest achievements, and for this pandemic, we’re extremely lucky to even have vaccines this quickly, with such high effectiveness and safety (thank you modern technology and the hard work of scientists around the world — and don’t forget the engineers and technicians, because we had to scale production up really fast).
Millions of people will live and see their kids and grand-kids grow up thanks to these vaccines, and millions more would suffer unnecessarily and possibly have long-term effects without them.
As an aside, isn’t it bonkers that even praising vaccines is seen as a kind of political statement by many now. You can say you’re a fan of all kinds of techs — solar power or semiconductors or space rockets — and people will still think at least semi-rationally about what you’re saying. But say you dig vaccines, and it won’t engage the same part of the brain. 🤔 Can you imagine if through some accident of history, this cultish group was anti-antibiotics, and they were dying left and right from preventable sepsis and pneumonia while posting stuff on Facebook about how bacterial infections aren’t real and antibiotics don’t help anyway…
Back to my story: Almost instantly after posting my good news, the mouth-breathing, anti-science/functionally-pro-virus brigade appeared out of nowhere, Beetlejuice-style.
To be very clear: It’s ok to be uncertain about all this, to have questions, to be confused by the fog of war, to be angry at the bungling authorities, to not understand the details of what people are talking about because you have no interest in biology, etc. There’s plenty of good faith people out there. But that’s *not* who I’m talking about here — I’m talking about people who’s opening line is “Did you ask your son what his last wishes are? Poor kid, you’re a terrible parent, you’re everything wrong about the world today…”
Almost all of them are people who don’t follow me, so they’re actively looking for keywords in Twitter search to go and harass people and create this astroturf campaign.
I’ve been online long enough to know how this works and just laugh it off, but I can’t imagine how a civilian, some casual Twitter user with 23 followers and not much understanding of the medical facts, deals with 30 people trying to brow-beat them with emotional attacks and scary-looking screenshots of “studies”…
It’s the usual talking points: “It’s dangerous, it doesn’t prevent infection, it doesn’t prevent transmission, the virus is harmless for kids anyway” (isn’t it contagious, though? How about being in a transmission chain that gives it to someone vulnerable? Plenty of older vaccinated people don’t have as robust an immune system and aren’t as protected, plus all the people who just won’t take the vaccine because they did their medical research on Fox News)...
Some even claimed “the injection is not a vaccine” (???) and that it’s “experimental gene therapy” (it turned hilarious when I had that person try to explain what gene therapy was).
I also saw a lot of bad understanding of risk — the choice isn’t between doing something with X risk and zero risk, but rather between something with X risk and something else with what I judge to be much greater than X risk. So I’ll take the lower between the two. Not making a choice would still be a choice…
I had fun counter-trolling some of them for a bit — my litmus test is I won’t take medical advice about mNRA vaccines from people who don’t know what a ribosome is — but after a while it got boring and I just blocked the lot.
Anyway, my point is that social media is clearly still a huge battleground for a propaganda campaign by a functionally-pro-virus cult. It may not be happening quite the way it was last year, but it’s still going on in a more targeted way.
P.S. If you want to learn more about vaccine safety and the anti-vaccine cult, these very thoughtful interviews by Peter Attia are a good starting point:
💚 🥃 I'm thinking of forming a DAO to buy this building and turn it into the Liberty's Highlights HQ.
After the certain success of this venture, we’ll all be able to meet there to drink green tea and/or scotch. 🍵 🥃
Maybe we can have a big room with a bunch of PCs so we can have LAN parties of Starcraft, PUBG, and Hell Let Loose, and another big room where we install a nice projector and Dolby Atmos sound system to do marathons of Denis Villeneuve, Wes Anderson, David Fincher, and Coen Brothers films…
Who's with me?
Liberty’s Highlights is reader-supported. To support my work, consider becoming a paid subscriber. 💙
🧩 A Word From Our Sponsor: Heyday 🧠
Do you have 15+ browser tabs open right now? 😬
Give your memory a boost with Heyday so that accidentally closing them (😱) doesn’t feel like the end of the world.
The Heyday browser extension is like cheat codes for your memory.
It automatically saves content you view and resurfaces it alongside your Google search results. 👩💻
Investing & Business
Cirrostratus, Cirrocumulus, Cumulonimbus, and other clouds
AWS kept hold of 33 percent of the market, followed by Microsoft Azure with 20 percent and Google Cloud at 10 percent. Even as the top three continued to bring in more market share, there was still growth to be had in the space, with the next 10 largest cloud providers seeing 28 percent revenue growth and revenue for small to midsize providers jumping 25 percent. (Source)
Speaking of the scale of the cloud, Benedict Evans recently posted this slide:
Snowflake Q3 Highlights
I’ve been following Snowflake since a little before IPO because I think their tech is cool, Frank Slootman is an interesting CEO, and their operational performance so far has been pretty 🤯.
I don’t think I need anything more to be curious about a company… So let’s have a look at their Q3 highlights:
Product revenue of $312.5 million, representing 110% year-over-year growth (that’s +22.9% sequential growth and an acceleration from last Q)
Net revenue retention rate of 173% (!)
Revenue from the EMEA and APJ regions up 174% and 219% year-on-year, respectively
Non-GAAP product gross margin are now 74.6%, which is a YoY bump of 420bps (meaning they’re getting nice leverage even though they operate on top of the hyperscalers — their scale must allow them to get volume discounts)
All their other metrics lower down (operating income, CFO, FCF, etc) swung by ˜+20-30% in one year
“For the full year fiscal 2022, we expect product revenues between $1.126 billion and $1.131 billion, representing year-over-year growth between 103% and 104%”
Some highlights from the transcript:
Vertical industry focus is an important evolution of our selling motion, especially in global enterprise accounts. During the quarter, we announced 2 industry data clouds.
The financial services data cloud brings together the Snowflake platform, partner solutions and industry data to help financial services organizations mobilize their data. Customers can launch products, build fintech platforms and accelerate their compliance on top of Snowflake. […]
We also launched a media data cloud, which enables media and advertising companies to share data for audience insights and measurements. With data clean rooms enabled by Snowflake, advertisers, agencies and publishers can design their own collaborative environment.
The way they target these industry verticals reminds me a bit of Adam Selipsky’s first re:Invent keynote as AWS CEO, and how AWS seems to want to do something similar, at least when it comes to how they package their very horizontal platform for their sales motion.
To date, there are over 175 Powered by Snowflake companies who have access to technical resources to design their applications. Most recently, we announced Securonix, UiPath, VideoAmp and ZoomInfo as Powered by Snowflake Partners.
It’s a bit like Matryoshka dolls…. You have AWS or Azure at the bottom, then Snowflake builds on top of it, and then other companies build on top of Snowflake (and on hyperscalers too, to make for a more tangled web).
This approach certainly must help make the service a lot more sticky. It’s one thing to use Snowflake as a data warehouse or lake on the side, but it’s another to basically build part of your app right on top of it.
Snowflake's Data Marketplace grew 41% this quarter, now with more than 900 data sets from over 200 providers.
We also saw a more than 130% annual increase in so-called stable edges. Stable edges are ongoing Snowflake data networking relationships between providers and consumers. One of the feature data sets is FactSet's tick history data feed. It provides asset managers real-time data from over 200 exchanges. ZoomInfo is another feature data set.
A big part of what Snowflake is trying to build, and what distinguishes it from a lot of what has come before, is this “Data Cloud” concept with a vibrant data marketplace that allows Snowflake customers to sell access to part of their data to third parties, in a way that is easy and secure.
You wouldn’t want to send your raw data to someone else, because you could never be sure if they actually deleted it when the deal is over, but inside of Snowflake, a third party could run queries against your data, get their answers and pay you, and when the deal is over, Snowflake simply cuts off access to that data set (which never left Snowflake and couldn’t be downloaded in raw form).
So this 130% increase in “stable edges” is a way to show that different Snowflake users are making deals to use each other’s data, and that some of these relationships are long-lasting (not just one-offs). It’s kind of a health metric for the data marketplace.
Of course, all this also makes Snowflake more sticky for users, and this data gravity can attract new customers because it’s much easier to buy & sell access to data inside of Snowflake than outside of it.
we are just seeing the tip of the iceberg. Snowflake was built from the ground-up as a data sharing platform, and we've been at it from the beginning [...]
people look at modernizing legacy workloads. And those kinds of things often have priority over getting to data sharing because we can't even consider data sharing unless we get our data to the cloud. We start moving that workload, we migrate the databases and so on [...]
we sort of haven't reached that tipping point yet, where sort of the floodgates are open and things are just expanding at a meteoric rate. But we're anticipating that, that will happen at some point. It's very nonlinear in the way the adoption is going to develop.
Slootman seems to think they’re still in the early phase of having customers move existing workloads to the cloud, but that over time, as they discover new possibilities that they didn’t have access to before, they won’t just *replace* what they did before, but *expand* it quite a bit (incl. with data sharing).
you definitely see network effects, not so much by enterprise, [or] by institution, but really by industry. Because the industry and sub industry, they really induce and evoke the network effect because the entities have relationships and do business together. [...]
the financial services data cloud and media and advertising, there is a huge amount of network effect in that area. And advertising for a lot of reasons that people know is under enormous pressure.
One more reason for the industry-vertical focus is exactly this.
You may not have nearly as much data-sharing between industries, but inside of it, the network effect can help attract new customers and retain existing ones once you reach critical mass.
The sales organizations are -- over the last year, are being redirected really understand the customer's context, what are their challenges, what they're people doing, what are people not doing, how are they approaching it, so that we can bring that value to our customers.
So it's no longer hey, here's our architecture versus the next guy, let's benchmark and POC this workload, see how they do. That was historically what Snowflake did. Now we're completely leaned into the customers' context, what are their issues, their challenges. And we are becoming experts at their business.
Not about just selling software, but about knowing enough about your customers’ business to sell them solutions to their problems.
Q: this kind of revenue outperformance is kind of above the parameters we've thought about, certainly above what we've seen in the past few quarters.
A: I don't think you're going to see that same repeat of a beat. At least I'm not expecting that. I'm sure you guys would love it, but I just don't see that happening in this quarter. As I said before, a 5% to 7% beat is a big beat for us with our model.
So it was exceptional performance last quarter. I'm actually disappointed we outperformed that much, to be honest with you.
When the CFO almost apologizes for the company performing so much better than expected, haha.
In context, I think it’s more about their failure to forecast it and guide to it (so expectations are better calibrated) rather than about the actual performance.
the whole world going direct to consumer. Every industry is to -- even people that have historically not been direct to consumer are moving towards direct-to-consumer that will obviously triggers enormous investments and people trying to get up to speed on data operation, data science and being able to run a real data-driven enterprise. [...]
the ability to own your own data as an institution, as an enterprise and to fully operationalize and mobilize, is going to become incredibly important. I mean even industries that historically have not been direct-to-consumer, and think of retail and things of that sort, retail, obviously, in the brick-and-mortar type, they're all changing now under the influence of the likes of Instacart and DoorDash.
In the post IDFA world, owning your own first-party data, and being able to augment and enrich it with access to third-party data shared through something like Snowflake’s data marketplace is going to become more and more valuable.
on the topic of performance, we are very intentional that each time that we make the system faster, we're improving also the economics of Snowflake. And we hear very consistently from our customers that those are very welcome surprises, where the economics of the platform are getting better without having to do any upgrades or changes to their platform.
This is a bit similar to AWS’ price cuts over the year (especially aggressive in the early years).
By improving data compression and improving performance on the compute, customers can either get a smaller bill for the same workload, or run more workloads for the same amount… And by the nature of doing data analysis that has a good ROI, it tends to be more the latter than the former.
Spotify vs Comedians
Tell me if you’ve already heard this one before…
Spotify Technology has removed the work of hundreds of comedians, including John Mulaney, Jim Gaffigan and Kevin Hart, amid a new fight over royalty payments. [...]
Currently, comics are paid as performers through their label or distributor and digital performance rights organization SoundExchange when their recordings are played on a digital service. They aren’t directly compensated as writers of that work (Source)
It’s hard being a billionaire… in China
Written in 2011:
I'm no statistics whiz, but it seems to me that a Chinese billionaire dies every 40 days.
China Daily reported Friday that unnatural deaths have taken the lives of 72 mainland billionaires over the past eight years. (Do the math.)
Which means that if you're one of China's 115 current billionaires, as listed on the 2011 Forbes Billionaires List, you should be more than a little nervous.
Mortality rate notwithstanding, what's more disturbing is how these mega wealthy souls met their demise. According to China Daily, 15 were murdered, 17 committed suicide, seven died from accidents and 19 died from illness. Oh, yes, and 14 were executed. (Welcome to China.)
I don't know about you but I find it somewhat improbable that among such a small population there could be so many "suicides," "accidents" and "death by disease" (the average age of those who died from illness was only 48). I'm only speculating but the homicide toll could really be much higher.
The above stats are a decade-old, but things don’t seem to have changed too much, as we’ve seen from recent developments…
The nature of the Chinese system means that steering clear of politics is no protection. The business world is opaque and runs on connections, so everyone is operating in what Shum calls the “grey zone”, and is therefore vulnerable to a charge of corruption. All institutions are under the sway of the Communist party. (Source)
Liberty’s Highlights is reader-supported. To support my work, consider becoming a paid subscriber. 💙
Science & Technology
California’s Last Nuclear Power Plant (scheduled to close in 2024-2025) 🔌
The photo above shows the Diablo Canyon plant in California.
It has a nameplate capacity of 2.25 gigawatts, or about 16,165 GWh/year (in 2019). That’s a lot of ⚡️
Scientists from the Massachusetts Institute of Technology and Stanford University have published a 113-page report outlining ways to retrofit California’s Diablo Canyon nuclear power plant, which is currently scheduled to be decommissioned and closed in August 2025.
The plan would save up to $21 billion in power grid costs if Diablo Canyon were kept open until 2050.
In addition, keeping the plant open to 2050 would preserve 90,000 acres of land that would otherwise need to go to be covered in photovoltaic solar panels to meet California’s climate goals. The scientists also propose that Diablo Canyon could be used as a site for a desalinization plant to provide much-needed water to the state, and a plant to produce hydrogen for use in clean-energy solutions.
The plant is already there and operating.
This isn’t even about building something new (which is almost impossible these days)… The anti-nuclear sentiment in the state is the main reason why it’s being shut down, not technical reasons.
Even with all the extreme regulatory costs of ALARA that I wrote about a few times in the past, the plant is still producing electricity for about 6 cents per kWh, which is pretty low for California.
PG&E, the owner of Diablo Canyon, told CNBC it is not reconsidering its decision to decommission and close the plant. But the scientists behind the report hope another utility might take it over.
If you want to check out the report arguing for keeping Diablo open, it’s here.
TSMC Founder Morris Chang on Intel
Via Dan Nystedt:
TSMC founder Morris Chang said he doubts Intel CEO Pat Gelsinger can restore Intel to its former glory, media report. Intel's mandatory retirement age is 65-years old, and as Gelsinger is already 60, Chang does not believe he has enough time.
TSMC's founder also reportedly said Gelsinger's recent comments about Taiwan being unsafe, and his criticisms of TSMC and other Asia chip makers are ways to try to get more of the U.S. government's subsidy money for Intel, media report.
Dr. Chang also reportedly said Intel sought TSMC years ago for chip production, but Chang told TSMC Chair Mark Liu not to get excited about it because technology is in Intel's soul, so very likely nothing would happen. Chang was correct.
Liberty’s Highlights is reader-supported. To support my work, consider becoming a paid subscriber. 🤠
The Arts & History
Sequel to Spider-Man: Into the Spider-Verse (2018)
I’m not really big into super-hero stuff, and I know very little about Spiderman in general (just stuff that has become part of the popular culture, but I haven’t really seen the movies or read the comics).
But I saw ‘Into the Spider-Verse’ in 2018, and was so blown away by it that I actually saw it twice in theater.
It’s such a visual feast, with so many innovations, that I think the theater probably added a lot to it — and because they play a lot with framerates, with some characters being animated at say, 12 fps while others were at 24, and some others at 12 too, but on the opposite 12… If you saw it home on a TV screen that had any motion-smoothing going on, it ruined the whole effect.
Anyway, this is just context to say that when I saw this teaser of the sequel, I had to check it out. Hopefully the film will be as fun and innovative as the original.