239: Nvidia+ARM is Dead, Nvidia + RISC-V?, ARM IPO, Intel Foundry Services, Starlink, Apple + Peloton?, Flexport, Nuclear Shipping, Stripe, and Spielberg

"I’m just glad they didn’t ask me to fax something"

A truly great library contains something in it to offend everyone.

—Jo Godwin

💨🦠🪟 It will be interesting to see if a bunch of ventilation improvements made because of Covid make us realize that many other diseases were more airborne than we thought (ie. some of the seasonal variation that we attribute to other factors may be caused by the differences in ventilation during the cold season).

But even without taking pathogens into account, more fresh air would be a big win health and cognition-wise.

🐦 I guess I’ve officially arrived. Friend-of-the-show and supporter (💚 🥃) Willis Cap alerted me of an imposter account on Twitter pretending to be me, DM’ing people and asking “how’s your trading going?” (same avatar image, same name, only the username is different with “RFP” instead of Richard Feynman’s initials).

Unsurprisingly, twitter’s process to report a scammer is very clunky and manual. I’m just glad they didn’t ask me to fax something…

How hard would it be to write three lines of codes so that when a new account is created, if it has the *same avatar*, *same banner image*, *same name*, and an *almost identical username* (1-2 characters off, like replacing a lowercase “L” with uppercase “i”, etc), then automatically flag it for review.

Put the burden of proof and extra friction on scammers, not the people being scammed.

But hey, it’s Twitter, right? 🤦‍♀️

🦅 👀 Eagle-eyed reader Gary Chen (ok, am I getting too on-the-nose with the emojis?) noticed in the photo of my washing machine in edition #238 that I’m using rubber hoses to connect the water supply.

it would be worth considering changing that to stainless steel as rubber tubes are known to fail and could be a source of leak.

I had never thought about it before, but it makes sense.

Doing a bit of research, I discovered “auto-shutoff” or “flood safe” hoses:

These hoses have a shut-off mechanism at one end as an added safety feature. If the hose bursts, the connector can sense the water pressure change and it stops the flow of water. This auto-shutoff device can mean the difference between a little water on the laundry room floor and extensive flood damage.

“While typical water supply pressure ranges 50 - 80psi, our hoses are rated 290psi with burst strengh up to 1500psi.”

I also learned about small, battery-powered leak detection sensors that make a loud noise (90-100db) when they get wet. You can put them under sinks, near water heaters, etc.

You know what they say, an ounce of prevention is worth a pound of gutting out the water-damaged floors!

🚧 Friend-of-the-show and supporter (💚 🥃) Conor posted a very thoughtful update about his newsletter, and I think it shows great introspection and fortitude to be able to do course-corrections even when they are short-term painful, or go against the expectations that others may have.

Over long periods, what ends up mattering is the rate at which we grow and learn.

If Conor can increase that rate via a move like this, then I’m happy for him, because it’s a positive development! HGMI.

💚 🥃 If you’re trying to gauge in your mind how much work goes into this newsletter, don’t just imagine how long it’d take to type. That’s the quickest part.

The majority of the iceberg is reading & listening to lots of things, most of which never make the cut. Then thinking about what I want to say about the 10% interesting enough to include, then putting it on the page, and then re-reading it all to fix most of the typos and clunky sentences (sorry, can’t catch ‘em all).

To be clear, I’m not complaining, it’s a lot of fun, but it’s a lot of work too.

If you feel you’re getting value from it, please consider becoming a paid supporter, it’s the only way I can keep going:

---

A Word From Our Sponsor: ⭐️ Tegus ⭐️

What makes a great investor isn’t pedigree or diplomas 🎓, it’s quality thinking and information 🧠

Tegus can’t provide the thinking for you, but on the information front, they’ve got you covered! 📚

With 25,000+ primary source expert calls covering almost any industry or company you may want to learn about, and the ability to conduct your own calls with experts, if you so choose, at a much lower cost than you’ll find elsewhere (70% lower than the average cost of ˜$400), this is the service to turbo-charge the depth and breadth of your knowledge 📈

⭐️ Get your FREE trial at Tegus.co/Liberty ⭐️

---

Investing & Business

Nvidia + ARM deal officially dead ☠️, Softbank will IPO it with new CEO

Well, not a surprise, but it’s officially official now.

It’ll be interesting what kind of price Mr. Market puts on this company when it IPOs sometimes “before March 2023”, because Nvidia’s stock & cash deal that was originally ˜$38.5bn had ballooned up to over $80bn (depending on Nvidia stock price), yet ARM revenue has just been reported to be $2bn in the first 9 months of 2021. Even if you’re generous and put it at $3bn for the year, that’s still the kind of EV/S number that the current market may not be ready to pay for this asset.

SoftBank acquired Arm in 2016 for $32 billion. I’d be surprised if it traded for much more than $40bn, maybe $50bn (which isn’t a great return over 6-7 years), but who knows ¯\_(ツ)_/¯

The new CEO will be Rene Haas, replacing Simon Segars “immediately”. Softbank is getting a break-up fee of up to $1.25bn.

What will be interesting to watch next is if the ARM mothership can compete with its customers on designing data-center CPUs, or if most of the value will be captured by the AWS’ and Nvidias of the world, as ARM’s own designs remain too vanilla to compete with those who have a lot more R&D resources to throw at the problem.

Looking back on the merger attempt, I can see both sides’ arguments — ARM customers didn’t want Nvidia to have more power over them, or for ARM’s position to become stronger, allowing it to capture more of the value it creates.

Nvidia knows the data-center will be a huge part of its future, and that ARM will keep growing in the space, so if it’s going to invest a ton there, it would prefer to have more control over the stack, more vertical integration, and capture more of the value created. It would also love one more channel for its IP (“anyone wants ARM chips with CUDA cores on die?” or whatever they’d call it…), which would’ve made ARM more competitive with x86 and third-party designs.

ARM-less Nvidia, does that push it in RISC-V’s arms? 💪

(push it in its arms? get it? 😬)

It’s hard to fit a lot of nuance in a headline, so let me first start by saying that Nvidia doesn’t *need* to own ARM to keep building its own ARM chips (it has an architectural license to the IP), or using ARM’s own designs in its products. It has already made the ARM-based Grace CPU, and I expect more over time.

But if Nvidia doesn’t own as much of the upside of investing a gazillion R&D dollars into ARM products, maybe it’ll make more sense for it to hedge its bets and do more on the RISC-V front.

The change wouldn’t be binary, from all ARM to all RISC-V, but the change in intensity could still matter over time, not only for Nvidia, but also for ARM and RISC-V (because Nvidia is big enough that when it moves around, smaller players feel it).

If someone has the resources to build the RISC-V software stack, which is largely what is missing for a lot of use cases, it’s Nvidia.

I was discussing this with Stephen Nellis a few days ago, and one of the questions that came to my mind was about the RISC-V IP licenses; are they closer to the GPL, where modifications have to be contributed back to the open-source IP pool, or is it closer to FreeBSD or Apache, where modifications can be kept proprietary or contributed back on a voluntary basis.

I suspect it’s more the latter, with the RISC-V foundation maintaining the common ISA, and with third-party developers being able to then do whatever they want as long as they maintain compatibility (ie. SiFive is making its own chips), but I haven’t dug deep into license details yet (yawn). If you know about this, please let me know.

Intel Foundry 😘 RISC-V

Speaking of RISC-V, Intel has announced a new $1bn fund to “strengthen Intel’s foundry business”, and that includes support for RISC-V IP and a partnership with a bunch of RISC-V designers (including SiFive, which I’m keeping an eye on):

[Intel Foundry Services] (IFS) is the only foundry to offer IP optimized for all three of the industry’s leading ISAs: x86, Arm and RISC-V.

Intel is joining forces with leading partners in the RISC-V ecosystem, including Andes Technology, Esperanto Technologies, SiFive and Ventana Micro Systems. IFS plans to offer a range of validated RISC-V IP cores, performance-optimized for different market segments. By partnering with leading providers, IFS will optimize IP for Intel process technologies to ensure that RISC-V runs best on IFS silicon across all types of cores, from embedded to high-performance.

Intel is also joining RISC-V International, the nonprofit in charge of the RISC-V instruction set architecture (ISA) and extensions.

🐴 SiFive:

SiFive has partnered with IFS to develop a RISC-V development platform, codenamed “Horse Creek,” featuring a multi-core SiFive Performance P550 processor, and implemented on the Intel 4 technology platform, on track for availability in 2022.

I’m a bit surprised it’s the P550 and not the P650 (“>50% uplift over the SiFive Performance P550”). Probably soon ¯\_(ツ)_/¯

Nasdaq’s 15-year drawdown & subsequent rise 📈📉📈

Friend-of-the-show and Extra-Deluxe (💚💚💚💚💚 🥃 ) supporter Byrne Hobart writes:

the Nasdaq Composite took 15 years to get back to its March 2000 high of 5,048. An unlucky tech investor could have spent almost half of their career in a drawdown! But since then, even including the fireworks of the last month, the total return from a Nasdaq 100 ETF has been 19.7% annualized.

Andrew Walker on Cable vs Fiber 📺 💻📱

Friend-of-the-show and OG supporter (💚 🥃 🎩) Andrew Walker has a very interesting series of posts diving pretty deeeeep into cable and the threat from overbuilding their market with fiber-to-the-home (FTTH) and fixed wireless:

• Cable: part 1: overview

• Cable: part 2: FTTH risk

• Cable: part 3: fixed wireless

Here’s a highlight:

Comcast says that right now, 40% of their footprint is overbuilt with fiber, and if you add up all of the FTTH announcements they think their footprint goes from 40% fiber overlap to 55% over time.

From an investor’s standpoint, that’s enormously good news. Based on my talks, I think a lot of investors are internally thinking about FTTH competition like cable companies are about to go from competing with fiber in ~20% of their footprint to 100% of their footprint in the next 5 years. That’s completely off

SpaceX Starlink Service Density 📡 🛰🛰🛰🛰🛰🛰🛰🛰

As part of the series of posts above, Andrew also published a really interesting transcript of a conversation about Starlink with a former SpaceX manager who spend about 5 years at the company until 2020.

Some highlights:

A: with the current satellite design, roughly approximating here, for an area the size of like Seattle or Toronto, that large of a surface area, you can really only put like 300 to 600 users. So, this isn't, as you've said, honestly, that big of a competitive threat to like raw servicing of highly dense areas, it's a spatial play. [...]

Q: if you can only fit 600 people in like a Seattle-based area, and maybe that goes up to 1,000 over time, I don't know. But like yes, that's nothing, right? Whoever is servicing Seattle probably has three million broadband users. If they're only going to lose a 1,000 to Starlink, that's not a worry for the city franchise. It might be a worry for their rural franchises, but it's absolutely not a worry for their city franchise.

A: That's correct. Yes, and it will go up. I think it will go up. It basically goes up as a function of the aperture of the antenna, like the aperture diameter of the antenna on orbit. That means, basically, you have a smaller beam size and a smaller service zone per beam from the satellite so you can fit more users in a given area without breaking the FCC rules.

But it's not like this cornucopia of solutions where you're going to have the exact same competitive user density threat as a telecom traditional terrestrial provider would have, it's going to maybe get better by like at most a factor of 10 or something like that. [...]

I think five years from now, totally guessing when I'm speaking, three years from now, five years from now, you're getting to the point where you have enough satellites in orbit and there's a newer generation with the larger aperture diameters antennas where you can fit, call it, 3,000 to 6,000 people in the same service territory where you only used to be able to fit 300 to 600.

Interview: NZS Duo, Jon Bathgate & Brad Slingerlend

Friends-of-the-show and supporters (💚 💚 🥃 🥃) Jon and Brad spoke with friends-of-the-show Ben and David (💚 🥃) for a fascinating conversation that, of course, went into the case for why the semiconductor industry is so important:

• NZS Capital Part II: What’s Going on in Today’s Markets? (Plus more Semiconductors!)

The link above includes a transcript, for those of you who prefer textual titillation.

I thought it was pretty funny when Jon said something about how Amazon hadn’t raised Prime prices in a long time, but “probably as soon as I say this they will”, and they did! The episode was recorded just a few days before the price increase was announced.

Apple + Peloton Rumors 🍎 🚴🏻‍♂️

This may turn out to be wrong by the time you read this, but for now, I agree with John Gruber’s take:

Apple gets mentioned frequently when potential Peloton suitors are pondered, but I don’t think that’s likely.

On the hardware side, bikes and treadmills are specific; Apple Watch is upstream of specific exercise machines. (I wouldn’t be surprised if a majority of Peloton owners have an Apple Watch.) On the service side, Apple already has Fitness+. If they were going to buy Peloton to turn Peloton’s online training classes into an Apple service, I think they would have done it two years ago.

Apple bought Beats to jump-start Apple Music; Fitness+ is already off the ground, and Apple has deliberately made choices that are different from Peloton’s (e.g. Peloton conducts live classes; Fitness+ is entirely pre-recorded.)

Amazon and Nike, I could both see.

Horace Dediu puts it well too:

What Apple does not buy:

1) Earnings
2) Revenues
3) Customers
4) Brand
5) Processes

What Apple buys:
1) Teams
2) Technology

After a few hundred examples this still seems to be a puzzle for many.

Apple Enables Contactless Payments on iPhones, Partners with Stripe

Once Tap to Pay on iPhone becomes available, merchants will be able to unlock contactless payment acceptance through a supporting iOS app on an iPhone XS or later device. At checkout, the merchant will simply prompt the customer to hold their iPhone or Apple Watch to pay with Apple Pay, their contactless credit or debit card, or other digital wallet near the merchant’s iPhone, and the payment will be securely completed using NFC technology. No additional hardware is needed [...]

Tap to Pay on iPhone will be available for payment platforms and app developers to integrate into their iOS apps and offer as a payment option to their business customers. Stripe will be the first payment platform to offer Tap to Pay on iPhone to their business customers, including the Shopify Point of Sale app this spring. (Source)

That’s probably not good for Square, or Cube, or Block, or whatever they’re called now…

⏰ Tiktok tic toc 🕰

The little app for teenagers doing dance moves is growing up fast:

Gavin Baker @GavinSBaker

2) TikTok went from experimental budget allocation in 2021 to mainstream budget allocation in 2022. Most surveys showed it going from 2-3% of social media budget in 21 to 7-10% in 2022. All of the TikTok increase came from the largest player in those same surveys.

Flexport & Ryen Petersen Profile 📦

I’ve mentioned him and his company a few times, and now Forbes has put them on the cover:

Flexport was founded in 2013 to automate paper customs forms. It now does much more, helping customers like Georgia-Pacific (Brawny paper towels, Angel Soft toilet paper), baby food giant Gerber and speaker maker Sonos handle all the headaches of shipping inventory from factory to warehouse to store. Flexport’s software analyzes and optimizes a customer’s supply chain, then automates it, often coming up with ways to shave days off delivery and save customers millions in late fees. Flexport’s centralized tracking and messaging cut out thousands of emails, saving clients an average of four work hours per week. For a price, Flexport will even offset their carbon footprint. [...]

Shipping is a huge pond. Global spending on logistics reached $9 trillion in 2020, about 11% of the world’s gross domestic product [...] Third-party logistics, of which freight forwarding is a big part, amounts to nearly $1 trillion. In the U.S., it’s a $230 billion business, good for 1.1% of national GDP. [...]

[Flexport] Sales reached $3.3 billion in 2021, up from $1.3 billion in 2020 and $670 million the year before (Flexport passes about 80% of its revenue directly to its shipping partners). Last year, the San Francisco–based firm generated its first profit, posting net income of $37 million.

Petersen made an appearance recently in the first half of this podcast:

---

Science & Technology

‘Can Nuclear Powered Ships Clean Up Shipping?’

Interesting overview of using nuclear power for ships, military and civilian.

Speaking of splitting atoms, here’s highlights from a recent WSJ piece:

All of the nuclear waste produced in the U.S. since the 1950s adds up to about 85,000 tons of material. Compare that with the tens of billions of tons of carbon dioxide that would have been produced had that electricity come from fossil fuels instead.

The U.S. Department of Energy estimates that the nation’s total nuclear waste would cover a single football field, 10 yards high. By contrast, carbon dioxide, a colorless, odorless gas, is typically released into the atmosphere, affecting the climate of the entire globe. [...]

China, meanwhile, intends to build more than 150 new reactors in the next 15 years and will surpass the U.S. as the world’s largest generator of nuclear power within five years. [...] China is now building the first molten-salt reactor that uses thorium as fuel, instead of more radioactive plutonium or uranium.

Coal 🏭

Over the last decade, more than 600 coal-fired units generating nearly 100 gigawatts-electric (GWe) of electricity went offline. Another 25 GWe of coal-fired power is slated to go offline in the next four years, including 2.7 GWe in 2021 alone.

‘Nissan to end most development of new gasoline engines’

Nissan Motor will end development of new internal combustion engines in all its major markets except the U.S. and focus its resources on electric vehicles, becoming the first major Japanese automaker to make such a break, Nikkei has learned.

Volvo and Hyundai have made similar announcements in the past year. Expect more to follow.

---

The Arts & History

‘Spielberg’ (2017)

Interesting HBO documentary on the film director.

I didn’t know much about him personally, and it was fun to re-live the past 50 years of cinema through his films, while getting some of the back-story on how and why they were made, and which aspects of his personality and growth as an artist they reflected.

I saw some parallels with Stephen King when they show how he was dismissed as an artist for a long time, and considered too commercial and basically “cheap thrills”.

h/t David from Founders Podcast for the episode that made me watch it.

Video Essay on the Rhythms and Voices of ‘The French Dispatch’ by Wes Anderson

If like me you’ve seen the film and liked it, I think you’ll find this interesting.