16: Nvidia Q2 Highlights, $9 Billion from LOL Dolls, and a Typo that Created a 212-Story Building
16: Nvidia Q2 Highlights, $9 Billion from LOL Dolls, and a Typo that Created a 212-Story Building
"Remember the rotation to value?"
Truth is much too complicated to allow anything but approximations. —John Von Neumann
One thing I like about writing here is that I’m not trying to sell anything to anyone, so I don’t have to follow the usual conventions of corporate/marketing writing. It’d be a lot more boring to write if I had to polish it up to those standards, which usually include the removal of almost all personality.
I figure I may lose some people who have no time for anything that isn’t 100% down-to-business (“where’s the actionable idea here, Liberty???”), but that those who enjoy the kind of eclectic things that I like may find it more fun that way.
Life is trade-offs. If you try to please everyone, you usually end up pleasing no one, especially yourself.
Don’t get me wrong, I understand why people who run investment funds and are trying to attract investors, or want to get hired by some prestigious firm, have to be pretty buttoned down. You have to seem very serious and responsible to be entrusted with people’s hard-earned money.
But even if someday I decide to have a subscriber-model here, it’ll be more on the understanding that we’re sitting across a table from each other, and I’m talking (probably too much) and sharing stuff that interests me, and you’re buying me a scotch to keep me talking.
I’d be quite happy with a Talisker 10, but the more generous of you should know, a dram of Ardbeg Uigeadail will extract the biggest smile out of me (enough of these, and I’ll probably even tell you what’s in my portfolio).
Speaking of Scotch, a good follow on the topic is @BadaBingWhiskey.
Investing & Business
This is the "rotation to value" that made many headlines a few weeks ago
I’m sure someday there will be a longer-lasting inflection point in the dance-of-the-factors, but trying to guess when that’s happening in real-time is above my pay grade.
Nvidia Q2 Highlights
Revenue of $3.87 billion, up 50% from a year earlier, up 26% sequentially
Second-quarter revenue [for the data-center segment], which included Mellanox, was $1.75 billion, up 54% from the previous quarter and up 167% from a year earlier
Gross margin (non-GAAP): 66% (up 590bps from 60.1%)
Powers eight of the top 10, and two-thirds of the total systems, on the latest TOP500 list of the world’s fastest supercomputers.
Set 16 AI performance records on the latest MLPerf benchmarks, eight on a per-chip basis based on the A100 Tensor Core GPU and eight “at scale” using the DGX A100 SuperPOD system (details here)
Segment revenues:
Data Center: +167%
Gaming: +26%
Pro viz: -30%
Automotive: -47%
(obviously the last two segments are much smaller than the first two)
From the call (highlights are mine):
Data center is diverse, consist of cloud service providers, public cloud providers, supercomputing centers, enterprises, telecom and industrial edge. Q2 revenue was a record $1.75 billion, up 167% year-on-year and up 54% sequentially. [...]
Mellanox contributed approximately 14% of company revenue and just over 30% of data center revenue [...]
Mellanox feels like such a bullseye adoption to me. Interconnect tech for improving lateral traffic should be an increasingly important and differentiated piece of the datacenter going forward.
[Ampere adoption:]
The first Ampere GPU, A100, has been widely adopted by all major server vendors and cloud service providers. Google Cloud Platform was the first cloud customer to bring it to market, making it the fastest GPU to come to the cloud in our history. And just this morning, Microsoft Azure announced the availability of massively scalable AI clusters, which are based on the A100, and interconnected with 200 gigabyte per second Mellanox InfiniBand networking. A100 is also getting incorporated into offerings from AWS, Alibaba Cloud, Baidu Cloud and Tencent Cloud. [...]
Right now, A100 probably represents less than 1/4 of our data center revenues. So we still have a lot to grow. [...]
in the latest A100 training benchmark, MLPerf 0.7, NVIDIA set 16 records, sweeping all categories for commercially available solutions in both per chip and outscale performance based on the A100. MLPerf offers the industry's first and only objective AI benchmark. Since the benchmark was introduced 2 years ago, NVIDIA has consistently delivered leading results and record performance for both training and inference [...]
We have now shipped a cumulative total of 1 billion CUDA GPUs, and the total number of developers in the NVIDIA ecosystem just reached 2 million. It took over a decade to reach the first million and less than 2 years to reach the second million [...]
The software/developer ecosystem is a very important piece of the puzzle. At first most people focus on the hardware, like with Apple, because it’s what’s most visible and what they charge for more directly, but a lot of the differentiation for these businesses comes from the software and from the relationships with devs.
[Gaming:]
Gaming's growth amid the pandemic highlights the emergence of a leading form of entertainment worldwide. For example, the number of daily gamers on Steam, a leading PC game online distributor is up 25% from pre-pandemic levels. And NPD reported that U.S. consumer spending on video games grew 30% in the second calendar quarter to a record $11 billion. [...]
We're expecting a really strong second half for gaming. I think this may very well be one of the best gaming seasons ever.
I’m almost surprised it hasn’t grown more than that during the pandemic… The Nintendo Switch spent almost the whole period supply-constrained and ‘out of stock’.
The pandemic will have a lasting impact on how we work. Our revenue mix going forward will likely reflect this evolution in enterprise workforce trends with a greater focus on technologies, such as NVIDIA laptops and virtual workstations, that enable remote work and virtual collaboration. [...]
Nvidia’s been pretty aggressive at branding laptops from other vendors as “Nvidia laptops” in a kind of variation on the classic “Intel Inside”. I haven’t looked super deeply at this and how they do it and what kind of deals they have with various manufacturers, but it’s an interesting strategy to try to get the brand to travel for the whole computer and not just for the GPU.
landmark partnership with Mercedes-Benz which, starting in 2024, will launch software-defined intelligent vehicles across an entire fleet in using end-to-end NVIDIA technology. Mercedes will utilize NVIDIA's full technology stack, including the DRIVE AGX computer, DRIVE AV autonomous driving software and NVIDIA's AI infrastructure, spanning from the core to the cloud [...] making the turning point of traditional vehicles becoming high-performance, updatable data centers on wheels. It's also a transformative announcement for NVIDIA's evolving business model as the software content of our platforms grows, positioning us to build a recurring revenue stream. [...]
Q&A:
The way that a cloud data center is built and the way that an enterprise data center or a cluster is built is fundamentally different. And it's really, really beneficial to have the ability to accelerate applications that cloud service providers would like to offer, which is basically everything. And we know that one of the most important applications of today is artificial intelligence. It's a type of software that really wants acceleration, and NVIDIA's GPU acceleration is the perfect medium, perfect platform for it. [...]
the architectural change from hosting applications to hosting services that's driving this new type of architecture called disaggregation versus hyper converged. And the original name of hyperscalers is a large data center of a whole bunch of hyperconverged computers. But the computers of today are really disaggregated. A single application service could be running on multiple servers at the same time, which generates a ton of east-west traffic, and a lot of it is artificial intelligence neuro network models. And so because of this type of architecture, 2 components, 2 types of technologies are really important to the future of cloud. One of them, as I mentioned, was -- is acceleration, and our GPU is ideal for it. And then the other one is high-speed networking. And the reason for that is because the server is now disaggregated, the application is fractionalized and broken up into a bunch of small pieces that are running across the data center. And whenever an application needs to send parts of the answer to another server for the microservice to run, that transition is called east-west traffic, and the most important thing you could possibly do for yourself is to buy really high-speed, low-latency networking. And that's what Mellanox is fantastic at. [...]
This is important stuff. I think a lot of people who don’t swim in tech waters too much may be missing how the abstraction layers have been changing in the cloud era, mutating and re-congealing into new shapes over the past few years, and going in the future. It’s very interesting to watch happen.
the dynamics that I'm describing are permanent, and it's just been accelerated to the present because of everything that's happening to us. But this is the future, and it's not -- there's no going back, and we just found everything accelerated [...]
The computing unit is no longer a microprocessor or even a server or even a cluster. The computing unit is an entire data center now [...]
Question about ARM rumors, with a non-answer that is still kind of interesting:
you asked a question about ARM. We've been a long-term partner of ARM, and we use ARM in a whole bunch of applications. And whether it's autonomous driving or a robotics application, the Nintendo Switch, console business that we're in. And then recently, we brought CUDA to ARM and to bring accelerated computing to ARM. And so we worked with the ARM team very closely. They're really great
guys. And one of the specials about the ARM architecture that you know very well is that it's incredibly energy-efficient. And because it's energy-efficient, it has the headroom to scale into very high-performance levels over time. And so anyways, we love working with the ARM guys. [...]Turing is probably not even close, not even 1/3 of the total installed base of all of our GeForce GPUS
The Turing generation is the first one with RTX, their ray tracing implementation. This is interesting because it lines up fairly well with a new generation of consoles that will increase the minimum GPU requirements for cross-platform games, which should help the upgrade cycle:
now with the new console generation coming, every single game developer on the planet is going to be doing ray tracing, and they're going to be creating much, much richer content. And because of multi-platform, cross-platform play and because of the size of the gaming platform, PC gaming platform, it's really important that these game developers bring the latest generation content to PCs, which is great for us. [...]
On drivers for data-center demand:
our data center trend is really tied to a few factors. One is the proliferation of using deep learning and artificial intelligence and all the services [...] over the last several years, the number of breakthroughs in artificial intelligence has been really terrific. And we're seeing anywhere from 10x, 10x more computational requirement each year to more than that. And so in the last 3 years, we've seen somewhere between 1,000 to 3,000x increase in the size of models
That’s it for that call. Hope that wasn’t too long for you. Let me know if you like these transcript highlights, or if it’s too much..
Costco to Brita: Your Margin is My Opportunity
Yes, I’m aware that there’s a decent chance that Brita is actually manufacturing the Kirkland version, and that Brita is making good money from the huge volume.
But I’m pretty sure that this is still kind of plan B for Brita, and they’d still rather only have their branded product on the Costco shelves without a cheaper alternative without their brand next to it, even if they also make money on it, because their long-term economics in good party depend on their brand, and that gets eroded in this model.
MGA Entertainment: $9 Billion in Revenue on L.O.L Dolls
Julie Young with an interesting post on one of these private businesses that nobody ever talks about, but that is surprisingly large:
I’ve become fascinated with MGA over the past few years as I’ve watched them gradually capture the zeitgeist of Gen Alpha (the generation following Gen Z, currently age 10 and under) with “L.O.L Surprise!” dolls.
If you have a daughter between the ages of 4 and 10, you probably know exactly what I’m talking about.
MGA reported that these tiny plastic dolls did $4 billion in revenue in 2018, $5+ billion in 2019, and they projected increased growth in revenue for 2020 (!).
9+ BILLION DOLLARS IN DOLL REVENUE!!!!
OVER JUST A FEW YEARS!!!!
This company’s existence is so completely batshit insane to me, I can hardly wrap my head around it. (For comparison, in 2019 Hasbro reported TOTAL global sales of $4.72 billion and Mattel reported $4.5 billion—MGA is a behemoth in toy world.)
A big innovation was making “YouTube native” toys that are built around the whole ‘unboxing’ experience. Very clever.
You can read the whole thing here (a history of the company and a lot more details on the techniques and tricks they’re using to make their toys more desirable). H/T Matt Ball
More TikTok Drama…
TikTok will continue to operate its hit music video app in the U.S., whatever comes of the threatened Trump administration ban on its business, senior executive Vanessa Pappas said.
“We believe we have multiple paths forward to ensure that we continue to provide this amazing app experience to the millions of Americans who come to rely on it every day," Pappas said Thursday (Source)
Not sure what’s the plan, here. Maybe this is just bravado to try to get a better price out of Microsoft (or Oracle… *shudder*). Because if they get kicked out of the Google Play Store and the Apple App Store, it’s pretty much lights out.
I guess they could still have a web presence and try to get users to go through there, but it would be a much diminished experience, and probably not a sustainable way to operate, especially if the US government starts trying to keep advertisers from spending on the platform. Maybe the plan is just to hang tight until after the U.S. election to see is things change and the ban is reversed or modified?
Interesting data point from the article: TikTok has 1,500 U.S. employees.
Science & Technology
I never thought I’d write this many times on Microsoft Flight Simulator, but it goes to show that you never know where a story will take you.
But I just love this, so I had to share:
A year ago, "nathanwright120" made an edit to OpenStreetMap [an open source, crowd-sourced, Wikipedia-like map], adding a tag that indicates that a building in the suburb of Fawkner in Melbourne, Australia, had 212 floors instead of 2. All his other edits of OpenStreetMap seem legit, so it appears to have been a typo...
The error was later corrected by another OpenStreetMap user, BUT, in the interim, Microsoft took an export of the data and used it to build Flight Simulator 2020. The result... this incredible monolith
You can read the original Twitter thread where this was discovered for the whole story.
Psychedelics Research Finally Getting $$$
The time to do this was probably 60 years ago, but I suppose the second best time is now.
A group of Silicon Valley and Wall Street executives has raised $30 million to speed the development of a closely watched psychedelic-drug therapy using the key ingredient of the party drug Ecstasy to treat trauma patients.
The Multidisciplinary Association for Psychedelic Studies [MAPS], a nonprofit advocating for psychedelic research since the 1980s, is conducting its last phase of clinical trials to research the efficacy of using MDMA-assisted psychotherapy for post-traumatic stress disorder. [...]
In 2017, the FDA designated MDMA as a breakthrough therapy for PTSD, meaning it would expedite review of the drug. MAPS said a recent interim analysis of its Phase 3 clinical trials, conducted by an independent data-monitoring committee, showed a very high likelihood the therapy will be effective for treating PTSD.
Kudos to Michael Pollan and Tim Ferriss for helping popularize and fund this research, and destigmatizing these potentially very helpful compounds for which there is very little health risk compared to many legal substances (take too much and chances are, you’ll never want to touch it again, rather than become addicted).
This episode of Peter Attia’s podcast is a good place to start if you want to learn a little more about psychedelics research and MDMA in particular:
I’ve also heard good things about the book ‘How To Change Your Mind’ by Michael Pollan, but I haven’t read it yet.
Baby Watermelons!
Earlier this year, my 8yo niece planted watermelon seeds in my mom’s garden, and now they have these cute things growing. Never really had thought before about what watermelons must look like in the early stages of growth… Just one of these “uh, kinda cool” things.
The Arts
Ponyo
Watched 'Ponyo' (2008, written and directed by Hayao Miyazaki of studio Ghibli, Japan) for the first time with my family recently.
The 6yo loved it, even the 2yo was captivated for whole thing.
Wife and I really enjoyed it too. Wonderfully inventive, great animation and aesthetics.
Recommended, especially if you can watch with young kids. If you want to watch other Miyazaki films but aren’t sure which ones are age-appropriate for your kids, you may find this blog post useful.
I’m looking forward to watching the 4-part documentary by NHK on Hayao Miyazaki and the creation of this film:
Miyazaki allowed a single documentary filmmaker to shadow him at work, as he dreamed up characters and plot lines for what would become his 2008 blockbuster, "Ponyo on the Cliff by the Sea." Miyazaki explores the limits of his physical ability and imagination to conjure up memorable protagonists.
Thanks to Kirk Simon for pointing it out to me.