330: Nvidia GTC, Preparedness paradox, Stratechery Plus, Cloudflare Deep Dives, RISC-V + NASA, AI Voice Disease Diagnostics, and Bladerunner 2099
"TIL about the existence of Detroit-style pizza"
A simple question asked may look foolish for five minutes, a simple question never asked my look foolish for the rest of your life.
—Original source unknown, found via friend-of-the-show Laurence Endersen
🍕⚙️ TIL about the existence of Detroit-style pizza. It looks 𝕘𝕠𝕠𝕕:
Very on-brand too, using industrial pans:
Detroit-style pizza is a rectangular pizza with a thick crust that is crispy and chewy. It is traditionally topped with Wisconsin brick cheese that goes all the way to the edges and caramelizes against the high-sided heavyweight rectangular pan. This style of pizza was originally baked in rectangular steel trays designed for use as automotive drip pans or to hold small industrial parts in factories.
The pizza was developed in 1946 at Buddy's Rendezvous, a former speakeasy owned by Gus and Anna Guerra located at the corner of Six Mile Road and Conant Street in Detroit
Now I really want to try it, but I have no idea where to get it around here. Maybe I need to find a good recipe and try making it myself 🤔
🛀 💥📊 Reality has a way of rudely blowing up the "forecast" part of financial models
“Expect the unexpected” is probably a good motto. ¯\_(ツ)_/¯
🕳🔥🌍 You know what I’d like to see?
A big Manhattan Project-type push for deep geothermal (not literally structured like the Manhattan Project, but the spirit of urgency and resource commitment). It’s dispatchable, clean, and doesn’t make us reliant on dictators. I’m not talking about only places where it’s easy to access near the surface, like Iceland — if you drill deep enough anywhere on Earth, you can find enough heat to boil water and spin turbines.
There’s orders of magnitude more energy than humanity needs beneath our feet, everywhere.
There should be a lot more invested in developing the technologies needed to make it happen sooner than later. In other words, geothermal is on a certain development and growth path right now, but with some leadership, we could no doubt significantly bend that path in the right direction.
Stuff like this rock-melting gyrotron that I wrote about in edition #300. Or as I wrote about in edition #219, there’s a lot of oil & gas machinery, skills, technologies, and workforce that can be reskilled to exploit geothermal, so it’s elegant on that front too.
It may even be a source of heat for industrial uses, to displace fossil fuels that are otherwise hard to replace (SMRs should be good for that too).
🎥 🪵🪓🌲🔨 If you need some relaxation, you could do a lot worse than watching the adventures of Erik Grankvist, a Swedish teen who ventured into the woods with the dream of building an off-grid log cabin using ancestral tools and techniques.
He filmed his journey over the years, as he and his dog show us just how much hard work is required to build anything in this world (especially if you do it all yourself and don’t “outsource” the work to groups of specialized professionals with mechanized tools).
Here is a chronological playlist that goes from chopping down the first tree to putting the finishing touches on the cabin.
h/t friend-of-the-show Doomberg (🟩 🐓) for telling me about Erik
💚 🥃 If you want to become a paid supporter but every time you click on the subscribe link you don’t see options to do so, it’s because you’re not logged in your Substack account on that browser/device.
Liberty’s Highlights is reader-supported. To support my work, consider becoming a paid subscriber. IT MAKES A BIG DIFFERENCE 🦸🏻♂️
📊 A Word From Our Sponsor: Stream by AlphaSense 🕵️♂️
👉 Power your investment decisions with quality expert insights on demand 👈
Need help making investment decisions based on vetted market insights? 🤔 Look no further! Stream by AlphaSense helps investment research teams access unique, high-quality expert insights faster and more cost-effectively than most traditional expert networks. 🚨 Sign up for a free 14-day trial here! 🚨
With Stream, you’ll have instant access to:
20,000+ on-demand transcripts of interviews between experienced buy-side analysts and company experts provide first-hand perspectives on the companies and sectors that matter to you.
The fastest-growing library in the industry, you're bound to find what you need, when you need it!
Insight from experts with 500+ hours of industry interview experience.
Instead of using the same pool of experts for multiple interviews, we custom source and recruit experts for every request. Each expert has recent involvement in their industry and our rigorous vetting process helps create the highest quality interviews, so you can make truly informed decisions. 🕵️♂️
🏦 💰 Liberty Capital 💳 💴
🔥 Nvidia 2022 GTC Keynote 🔥
There’s just too much for a full recap, but here are some highlights: NVDA
Honoring Ada Lovelace! Very nice historical reference (TIL: She died at only 36 years old from uterine cancer 😢).
76 billion transistors!
What stands out to me is how GPUs — which are accelerators over more ‘general’ CPUs in the first place — are getting more and more accelerators themselves. It’s turtles all the way down!
New improved cores for ray tracing, 4th gen tensor cores for certain ML workloads… If this keeps going, in a few years GPUs are going to have 10s of different kinds of specialized sub-cores (which isn’t a bad way to avoid commoditization, along with the software stack).
The demos of things created in Omniverse were very impressive — it feels like it could become a big pillar for the company if it becomes the standard in the space. The way it stitches together many other tools and formats — and possibly commoditize some of them eventually — is strategically smart.
Another clever move: Using the GeForce Now game streaming infrastructure to allow Omniverse to connect to Nvidia’s Omniverse Cloud to do all kinds of stuff, including run large digital twin simulations, stream complex 3D rendering to edge devices, etc
They announced a new insanely powerful automotive/robotics system called Thor with 2000 TFLOPS of compute that can simultaneously run Linux, QNX (a real-time automotive OS), and Android, and centralize a the disparate semiconductors content in a car unto itself (benefiting both the carmaker and Nvidia — but not so much those that make SoCs that specialize in infotainment, parking cameras, etc).
Another theme of the Keynote was Nvidia making more of their products available in the cloud, such as their Nemo, Bionemo (a model that can create new protein and chemical designs), and Megatron (530bn parameters!) LLM models.
Nvidia now has so many product lines, hardware, software, and services… It’s hard to keep track of everything, and even the summary at the end of the keynote was multiple minutes long.
Impressive how they went from a mostly just gaming company just 10-15 years ago to a kind of Microsoft of accelerated computing and AI today.
☯ 💪💣 The Preparedness Paradox
Do you know this one?
The preparedness paradox is the proposition that if a society or individual acts effectively to mitigate a potential disaster such as a pandemic, natural disaster or other catastrophe so that it causes less harm, the avoided danger will be perceived as having been much less serious because of the limited damage actually caused.
The paradox is the incorrect perception that there had been no need for careful preparation as there was little harm, although in reality the limitation of the harm was due to preparation. Several cognitive biases can consequently hamper proper preparation for future risks.
This applies to so many things.
Any *successfully* prevented or mitigated problem only gives ammunition to those who claimed that it wasn’t worth worrying about in the first place.
I’m sure there’s a bunch of people still saying “we used to hear about acid rain and the ozone layer, but now we don’t and it seems fine, so all that was for nothing”, overlooking the huge efforts to deal with those problems (upgrading equipment worldwide to reduce sulfur emissions, banning CFCs, etc).
If we avoid the worse impacts of climate change — I mean, even worse than what we’re already seeing around the world — we’ll never hear the end of it from people who go “see, it ended up not that bad”.
Same with things like AGI alignment. If we figure that one out, we can expect to forever after hear about how all the AI worriers got scared for no reason…
If being made fun of was the worst thing that happened here, it would be fine. The problem is that this creates a feedback loop that makes it harder to prepare and avoid future problems.
It’s the whole Minsky thing:
Stability is Destabilizing.
It’s an over-simplification, but sometimes it feels like the 20th century post-war generation built a lot of infrastructure that later generations were able to take for granted, and after a long period of stability, we’ve forgotten just how important things like a stable power grid are to our civilization, and what is required to have them — in other words, that problem had been ‘solved’ for so long in many countries (*cough* Europe *cough*) that it didn’t even seem like something worth worrying too much about….
Rich countries like the US are a lot more limited by leadership and imagination than money. How many nuclear power plants could’ve been built for the cost of the Iraq war (one estimate puts long-term costs at $1.9 trillion)?
🎁 Ben Thompson launching Stratechery Plus ➕ hints at expanding bundle… 🤔
As a long-time reader/listener and paid sub to Ben’s stuff (I first heard him on a short-lived podcast he did in 2013 called “Cubed” with three Bens…), my heart skipped a beat earlier this week when I got an email telling me I was getting a partial refund for my Dithering sub.
Oh no! Is Dithering ending?
Not only is it not, but Ben is going all-in on bundling by creating ‘Stratechery Plus’, a subscription that now includes the daily written updates + their audio versions + the Dithering podcast that he does twice-a-week with John Gruber (🍎), and a new podcast called Sharp Tech with Andrew Sharp “about how technology works, and the ways it is impacting the world.”
To me, the most intriguing aspect is this 🕵️♂️ :
This is, I hope, only the beginning for Stratechery Plus. Right now the content is obviously very Ben-centric, but my hope is to expand the offering over time. For now, I am delighted to be doing my part to make your Stratechery subscription more valuable than ever.
⛅️ Cloudflare Cloudbursting 🌧
When it rains, it pours. ☔️
(in a good way)
Two in-depth deep dives on Cloudflare came out at almost the same time — probably kinda making both authors feel a bit like their thunder was stolen a bit (ok, I gotta stop with the cloud-related puns) — giving those of us interested in the company with plenty of reading material. NET
My friend MBI (💎🐕) wrote:
Cloudflare: Helping to Build a Better Internet (sub $ required)
And The 2nd Man (👋) published on the 10th Man Blog (he’s the new Dread Pirate Roberts manning the ship now that the original Roberts retired):
These just came out and I haven’t had time to read them yet. I’ll circle back with highlights.
📡 🛰🛰🛰💥🛰🛰🛰🛰🛰 Russia implies it may target civilian satellites, Starlink 🇷🇺
A Russian diplomat said civilian satellites could be legitimate military targets in a statement that seems to refer to Starlink providing broadband access in Ukraine. Civilian satellites "may become a legitimate target for retaliation," the Russian official said in a statement to the United Nations' open-ended working group on reducing space threats.
Here is the translation of the statement from Konstantin Vorontsov, head of the Russian delegation to the United Nations Office for Disarmament Affairs (UNODA — almost did a spit-take on the Russian head of “disarmament affairs”):
We would like to underline an extremely dangerous trend that goes beyond the harmless use of outer space technologies and has become apparent during the events in Ukraine. Namely, the use by the United States and its allies of the elements of civilian, including commercial, infrastructure in outer space for military purposes. It seems like our colleagues do not realize that such actions in fact constitute indirect involvement in military conflicts. Quasi-civilian infrastructure may become a legitimate target for retaliation.
Actions of the Western countries needlessly put at risk the sustainability of peaceful space activities, as well as numerous social and economic processes on Earth that affect the well-being of people, in particular in developing countries. At the very least, this provocative use of civilian satellites is questionable under the Outer Space Treaty, which provides for the exclusively peaceful use of outer space, and must be strongly condemned by the international community.
Ah, yes, Russia complaining about risks to peace and the violation of international treaties. 🤥
🧪🔬 Liberty Labs 🧬 🔭
👀 See the innovation of Tesla’s structural battery packs (only on new Model Ys for now) 🔋🔋🔋🔋
These use the new 4680 batteries with the battery pack as a structural element in the vehicle. You’ll see in the video that the seats are mounted directly on the pack, which also acts as the floor of the vehicle, reducing the number of parts by a significant amount, saving weight, lowering manufacturing cost, etc.
SiFive’s RISC-V processors are going to Space 🚀
[SiFive] has been selected by NASA to provide the core CPU for NASA’s next generation High-Performance Spaceflight Computing (HPSC) processor.
HPSC is expected to be used in virtually every future space mission, from planetary exploration to lunar and Mars surface missions. HPSC will utilize an 8-core, SiFive Intelligence X280 RISC-V vector core, as well as four additional SiFive RISC-V cores, to deliver 100x the computational capability of today’s space computers.
This massive increase in computing performance will help usher in new possibilities for a variety of mission elements such as autonomous rovers, vision processing, space flight, guidance systems, communications, and other applications.
👩🏻⚕️🗣🎤 Using machine-learning to diagnose diseases using people's voice
Researchers are building a database of human voices that they’ll use to develop AI-based tools that could eventually diagnose serious diseases; they’re targeting everything from Alzheimer’s to cancer. [...]
The research team will start by building an app that will collect voice data from participants with conditions like vocal fold paralysis, Alzheimer’s disease, Parkinson’s disease, depression, pneumonia, and autism. All the voice collections will be supervised by a clinician. “So for example, somebody that has Parkinson’s disease — their voice can be lower and the way they talk is slower,” Bensoussan says. They would be asked to say sounds, read sentences, and read full texts through the app.
Very interesting, if it works, though I worry that such tools can then be used on almost anyone without their consent — soon we’ll have remote diagnoses of politicians, celebrities, and world leaders… (with plenty of false positives, I’m sure)
But also, insurance companies could potentially misuse this to deny coverage if there are no protections.
Elemento says he has a friend who is an oncologist and often treats patients who have metastases — cancer growths — in their brains. “He told me that he could actually tell from the changes in people’s voices that they were having metastases to the brain,” Elemento says. “I was blown away.”
🎨 🎭 Liberty Studio 👩🎨 🎥
🕵️♂️ Blade Runner 2099 mini-series 🦾🤖💭🐑
A Blade Runner TV show is in the works at Amazon Studios, the media giant has confirmed, with the original film’s director Ridley Scott on board as executive producer.
Blade Runner 2099 will be a limited series, and is envisioned as a sequel to the original 1982 film and its 2017 sequel, Blade Runner 2049, which was directed by Denis Villeneuve.
I’m skeptical that they can pull it off, and still wish 2049 had been a huge box-office success so that Villeneuve could’ve made a sequel… but we’ll see! 🍿
h/t my friend JPV