SITALWeek #231
Welcome to Stuff I Thought About Last Week, a collection of topics on tech, innovation, science, the digital economic transition, the finance industry, graphene amplifiers, and whatever else made me think last week. Please grab me on Twitter with any thoughts or feedback.
Click HERE to SIGN UP for SITALWeek’s Sunday EMAIL (please note some ad blocking software may disrupt the signup form; if you have any issues or questions please email sitalweek@nzscapital.com)
In today’s post: how we think about Tesla (and why we still own it); video game publishers are in the driver’s seat on cloud shift; AI and energy are driving changes and efficiencies as cloud growth continues; linear TV and radio ad dollars holding up; $5B in funding is driving virtual kitchens in obsolete retail, but the math looks challenging without vertical integration; and, lots more below...
Stuff about Innovation and Technology
The Art of Virtual Traffic Jamming
A performance artist created a virtual traffic jam by pulling a wagon full of 99 smartphones running Google Maps to trick the algorithm into thinking traffic was backed up. Google took it all in great spirits, responding: “Whether via car or cart or camel, we love seeing creative uses of Google Maps as it helps us make maps work better over time.”
Avoiding Autoplay Trailers
PSA: You can now turn off the autoplay trailers in Netflix as you scroll endlessly looking for something to watch. From a desktop browser, click on your profile picture, then edit profiles, and then uncheck autoplay for previews. I wish there were an easy in-app on/off toggle; or, an even better UI would be a longer hover/simple click for autoplay preview. (After making the change, you may need to sign out and back into your apps to get it to take effect – so far, I’ve struggled to get it to work on Roku). The obvious issue this highlights is that we still need a better universal search UI in the world of streaming.
Traditional TV and Radio Ad Revenues Holding Up
Despite the shift to streaming, linear TV advertising revenue was up 2% in 2019 when you include digital, according to a report from SMI on Mediapost. Viewer declines were more than offset by price increases as TV ads remain a scarce asset, particularly for live sports. Meanwhile, digital ad sales for terrestrial radio crossed $1B in 2019, accounting for 10% of total radio ad sales. The report, citing data from Borrell Associates, also showed local radio stations grew revenues 25% in 2019. (Note, I have seen diverging estimates on local radio, but the source here seems respectable.)
Profitless Ghost Kitchens Haunt Abandoned Retail Space
SBE Entertainment Group is partnering with mall-operator Simon Properties, hotel-owner Accor, Tavis Kalanick’s CloudKitchens, and others to open ghost kitchens for food delivery in obsolete real estate including mall retail, parking lots, etc. The group plans to have 185 kitchens opened by the end of 2021 at a cost of $60,000 upfront. According to the CEO of SBE in the WSJ, the kitchens will break even in six months if they can hit 125 orders a day of around $30 each ($3750/day in revenues, or over $1.3M/year). The WSJ article also states that $5B of VC money has gone into these virtual kitchen startups since 2018. This editorial in QSR Magazine by a VC focused on this emerging restaurant brands indicates a restaurant needs $650,000 in sales to break even in a ghost kitchen. Rent and labor are said to account for around 25% of sales at that breakeven, and figure another 25% commission to delivery platforms. I have to admit to being challenged by the math a bit here, and I still tend to think complete vertical integration only will win in food delivery (including grocery), and there may not be enough margin for marketplace models (similar to my views in The Evolution of the Meal last Fall). More color on local impact of ghost kitchens in this LA Magazine article.
Cloud AI Architectures and Semiconductors
Azure, AWS, and Alibaba clouds are all using SmartNICs (network interface cards) to offload AI processing to FPGAs or other accelerated computing chips. According to Paul Teich writing for The Next Platform, this is in part because neural nets and algorithms are constantly changing, and this offloading from X86 allows for adaptability. This trend is one reason behind GPU- and AI-giant NVIDIA’s purchase of Mellanox. Mellanox uses Xilinx FPGAs, but Paul suggests they may be likely to switch to a different solution under NVIDIA. NVIDIA’s march to add more custom and programmable features to GPUs and their CUDA programming language could keep them in the driver’s seat for a while. Intel, meanwhile, owns FPGA-maker Altera, but has several fits and starts ongoing in AI acceleration – the x86 giant weighed in on this topic in a blog post this week. Also don’t miss Jim Keller – currently at Intel, but formerly at Tesla designing their self driving car chip – discussing chip architectures, the state of semiconductors, and autonomous driving on Lex Fridman’s podcast this week.
Speaking of the growing attempts to solve AI on silicon, the always insightful co-founder of chip design maker Synopsis had this to say in an interesting interview covering many topics: “We are supporting a radically large number of companies right now that all claim to have the best AI chip...the minute it looks like something is promising, is the next generation available from someone who watched this evolution and executed fast? We’re going to see this for quite awhile.” De Geus continued: “So if semiconductors are $500 billion, and electronics are $2 trillion, the worldwide GDP is $85 trillion. Even a small percentage of change is a huge opportunity. If you just improve crops by 5%, that’s an enormous gain. If you do medical diagnostics radically better, that is very big money. The push on more advanced chips or systems of chips will continue unabated, and regardless of whether a chip costs $1 or $2, if it has impact it doesn’t matter that much.”
Rounding out semi news, here is an insightful interview with the head of the RISC-V coalition, Calista Redmond, ranging across numerous topics, including the European Processor Initiative’s plan to have high-performance RISC-V chips in the data center by 2021 – with an aim to significantly lower the power needs in order to battle global warming.
Gas Company Buys ERP Software Business
Industrial commodities conglomerate Koch Industries has acquired legacy software rollup Infor for $13B. Infor is a hodgepodge rollup of legacy software business from decades past, but the deal surely proves the value of long-lived software FCFs. Koch was a customer, then an investor, and now owner of Infor. What’s next? Berkshire Hathaway acquires Oracle!?
Cloud Rankings
Google provided some new segment disclosures this week in a classic Three-card Monte trick of distracting investors from the metrics it stopped disclosing (like cost per clicks) with the new metrics (PDF quarterly release). YouTube posted $4.7B in Q42019 ad revenues, up 31% and $15B for the year (smaller than I thought it was at <10% of Alphabet revenues). Google Cloud was disclosed for the first time, but it includes G Suite, so the $2.6B headline number in Q4, when compared to AWS’ $10B, needs to be adjusted down to perhaps ~$2B (my semi-informed, but incomplete guess). Microsoft also does not disclose the relevant Azure number for comparison, but many estimates would put it ~$4B in Q4. That gives us a public cloud (ex Alibaba) share from the three US-based platforms of ~62% (AWS), 25% (Azure), and 13% (Google). By all accounts, Azure is growing the fastest, followed by Google, with Amazon bringing up the rear at only 34% y/y in Q419 (this is another example of de-powerlawing). Interestingly, Eric Jhonsa at TheStreet reports that capex efficiency is rising for all three cloud platforms, thus requiring fewer hardware purchases to fuel growth.
Game Publishers Will Drive Shift to Cloud Gaming
Video games remain one of the biggest, growing, global media businesses at $150B with 2.5B active players, according to this rundown on the big gaming platforms in Protocol. Around half of that $150B is on mobile phones, and there is a theory that the other half, largely on PCs and consoles, will transition to a streaming model. This streaming theory assumes games don’t become more locally data and processing intensive (with AR/VR and other immersive experiences), which in my mind makes it more of a guess at this point. However, clearly more and more elements of games will be cloud based, and the terms of this transition will be controlled by the game publishers, not the tech platforms. Microsoft has a lot of game content and obviously the cloud scale with Azure to have a go. Last year, Microsoft also announced an Azure partnership with Sony for gaming, but it’s not clear if it’s actually going to come to fruition. Microsoft’s Mixer social network pales in comparison to Amazon’s Twitch, YouTube, or Discord. Google launched Stadia game streaming and has the cloud and streaming video share, but it has no internal gaming IP (and, notably, Activision’s big cloud gaming deal with Google didn’t seem to include streaming on Stadia). Amazon has cloud and Twitch, which is losing some share of late, but no gaming IP, and no (as of yet) announced game streaming platform. NVIDIA, which dominates PC gaming, launched their own GPU cloud-based game streaming service GeForce Now. Taking a step back, no one has everything it would take – cloud+games+social – to win if game streaming becomes a big business; however, if Amazon or Google were to acquire a major game publisher, that could change things overnight. Regardless, with all the competition, I think publishers will get the revenue shares they want, far in excess of 70%, and in contrast to China, where Tencent, with their dominant position in gaming distribution, pays out far less to publishers.
Miscellaneous Stuff
AlphaFly Doesn't Fly with Sports Regulators
Nike’s Vaporfly uses energy-returning foam combined with a carbon fiber plate that resulted in ~4% faster running times for marathoners. Less fit runners see an even bigger percentage improvement. The current model has been deemed legal, but the upcoming AlphaFly appears to exceed regulations against >40mm soles. Maybe I’m in the minority, but I’m all for letting technology enhance human performance as long as the playing field can be level, which, in the case of shoes, seems possible. Maybe it’s a slippery slope when we have AI-enhanced robotic shoes connected directly to our neurons, or if and when someone finally invents Flubber.
Tapping into THz
A graphene-based amplifier may allow us to tap into the elusive Terahertz frequency of the electromagnetic spectrum for the first time. Terahertz waves are between microwaves and infrared waves down at the long end of the spectrum. Previously, sources of THz waves were too weak to be useful. Applications would include a safer version of X-rays and, potentially, communications.
Stuff about Geopolitics, Economics, and the Finance Industry
Securitizing SaaS Revenue
Alex Danco posited an interesting inversion of logic for startup investing this week, predicting that debt will supplant or supplement equity VC financing. I tend to agree that there is opportunity to use different financing structures, especially in light of the extreme abundance of capital in the world that is lowering the cost of money for everything. Danco suggests that startups with recurring revenues could securitize those streams to generate funding to go after more customers. The risk I see is the conscious and subconscious impact that leverage has on decision making and behavior. Many once-great tech companies have taken on debt to maximize short-term shareholder returns and failed to invest enough in innovation and risk taking to the point where they face an existential crisis (examples include Oracle, Cisco, IBM, and to a lesser extent Apple, which has simply missed opportunities to be much better than they are).
For Returns, PE on Par with Public Markets
Since 2006, private equity returns have been about the same as public markets, according to data in this FT article; yet, there remains a perception that PE's added leverage and illiquidity result in higher returns for pensions and other institutional investors – perhaps because being locked up saves us from the fear-induced decision to sell when things are down (and buy when things are up). Those inline results are not stopping passive-fund-giant Vanguard from moving into the PE game. Initially, Vanguard will partner to target institutional investors; but, with efforts underway to open the investment funds up to mom and pop investors, I suspect they might eventually have broader ambitions.
-------------------------------------
Tesla: Auto Industry Races/Crashes into the Information Age
I get a lot of questions about Tesla, and I’m afraid that short-term predictions are almost pure guesswork. However, I do think it’s worthwhile to look at what is happening over the next decade with autos. Cars are shifting from a 1900s Industrial Age business to an Information Age platform business. As we have seen in many other similar transitions, vertical integration and network effects are winning features of successful Information Age platforms. Also, NZS, or non-zero sumness – the degree to which a business provides the most win-win for all constituents including customers, employees, suppliers, shareholders, society, and the environment – is a key factor in success during this global economic transition. There are three main components of businesses in the Information Age to consider: hardware, software, and data (and, in some cases a fourth: services). Most car makers today are overly focused on the hardware engineering of internal combustion engine (ICE) vehicles, under focused on software, and outsourcing many of the more advanced components and sensors in the vehicle.
There are two technological shifts, one potential business-model shift, and one distribution change happening in the car industry. The technological shifts are: 1) ICE to EV requiring significant battery knowhow, and 2) human driver to autonomous (including the interim steps of various driver-enhanced safety features like auto braking, lane keeping, etc.). The business model shift is a potentially-significant move from car ownership to fleet-based, on-demand car services. Lastly, the distribution shift involves a move towards direct-to-consumer sales, avoidance of the legacy dealer network, and decreased reliance on financing, service, and spare parts. One quick note on the distribution change before we address the other points: such a shift could significantly impact profit pools for legacy auto makers (who tend to make a lot of their money on financing and service/parts). There is one other business model shift that I won’t dive into here related to the union workforce of legacy automakers: EV drive trains require 40% less labor creating another wrinkle for legacy auto makers to grapple with.
These changes all require a significant integration of software, data, and AI that is daunting to say the least. The network effects as a car becomes a platform are typical of other industries, creating an ongoing gap of advantage for the initial leaders over the competition. In other words, the virtuous flywheel of innovation is constantly extending the lead of the leader in the Information Age. Novel technology development and vertical integration (including software and user interface) also mean that there is not an established supply chain to enable incumbents.
I imagine you can sense where this narrative is heading: in today's auto market, it's Tesla who has created a lead that is widening every day. Tesla’s internally-developed neural network semiconductor – which powers autonomous features and is integrated with all of the vehicle's hardware and software – is just one example of the incredible level of innovation they've achieved on a short time scale. Tesla is soon to have 1M cars on the road, all feeding data into their AI platform, providing an informational advantage that will be difficult to catch. And, their pace of R&D is only increasing.
However, given the diversity of use cases for both consumer and commercial vehicles, it seems unlikely that Tesla will end up with winner-takes-all share. In the Information Age, we sometimes witness the 'Apple vs. Android' phenomenon, whereby you have one highly vertically-integrated platform (Apple), geared toward a certain segment of the market, in competition with a widely horizontal solution provider (Android) enabling the rest of the market. For example, compare the iPhone, which has a minority share of units but majority share of profits, to the Android phone, which is in near-ubiquitous use around the world (excluding China, which cultivates its own phone ecosystem). Today, the same enabler of Android (Alphabet) also owns the most likely 'Android of Autos' – Waymo. Waymo is ahead of Tesla on the autonomous elements, but is neither in the business of EV drive trains nor cars themselves. So, we could see Tesla becoming the Apple of cars, and Waymo, or some other technological enabler of competitors, playing the Android role (at least $16B has gone into autonomous development at startups and established companies!). And, as usual, China will likely have an independent ecosystem.
There is an additional wrinkle in this particular Industrial-to-Information product transition that complicates things greatly: climate change. Society-wide cultural values can experience punctuated equilibrium, and it seems likely that environmental concerns could drive a relatively rapid adoption of EV technology. Legitimate economic savings from operating EVs over ICE cars (for example, OPEC could dramatically raise oil prices as demand drops) could potentially factor in as well. And, legacy ICE car makers could have an even more limited window for advancing their own EV technology, as governments around the world set their sights on banning ICE vehicle sales, in some cases by 2035, as was announced in the UK this week (however, note that regulation can cut both ways, as nationalism may lead to wide financial support for many car makers on their home turf).
Putting this all together, legacy automakers need to become best in class in EV drive trains (including batteries, sensors, and software), figure out how to amass/acquire data for autonomous safety features, and provide comparable or better value and user experience – AND accomplish this all while potentially facing a business model transition from individual ownership to fleets, a direct-to-consumer distribution change, and a cultural values shift...all within the next 10-15 years.
In most technological transitions, the legacy platform has a very long tail of profits as disruption looms. For example, IBM is still minting money on 40-year-old mainframe platforms despite the shift to client/server and then the rise of cloud computing. But, mainframes were never made illegal by governments, and they don’t have an environmental or cultural problem. So, this situation in cars appears unique in the history of tech paradigm shifts. However, none of this hands victory in perpetuity to Tesla. The car industry is regulated, it turns over slowly, there will be supply constraints for EVs (powerful negative feedback loops governing the pace of change); therefore, a complete transition to EVs in 10-15 years seems very difficult to achieve. But, given the high debt loads and inertia that are keeping legacy auto makers from investing in the dozens of complex technologies they need to in order to transition, a rapid meltdown could force a phase shift to occur. Ford, facing down this existential crisis, announced (after missing expectations this week) that its autonomous investments will be “modestly higher” this year. To borrow a quote from our Redefining Margin of Safety paper (in which we also discuss Tesla and the auto industry), "hoping the world will cycle back to a bygone era is not a productive business model".
The auto industry is a classic, complex adaptive system with a wide range of unpredictable paths through time. As such, analysis requires an objective, Bayesian approach to each new data point that surfaces. The scenario I describe here may or may not continue, but for now the range of outcomes is slowly narrowing in favor of Tesla, which drives us to continue to own Tesla as an Optionality position.
✌
Disclaimers:
The content of this newsletter is my personal opinion as of the date published and are subject to change without notice and may not reflect the opinion of NZS Capital, LLC (“NZS”). This newsletter is simply an informal gathering of topics I’ve recently read and thought about. It generally covers topics related to the digitization of the global economy, technology and innovation, macro and geopolitics, as well as scientific progress, especially in the fields of cosmology and the brain. I will frequently state things in the newsletter that contradict my own views in order to be provocative. I often I try to make jokes, and they aren’t very funny – sorry.
I may include links to third-party websites as a convenience, and the inclusion of such links does not imply any endorsement, approval, investigation, verification or monitoring by NZS Capital, LLC (“NZS”). If you choose to visit the linked sites, you do so at your own risk, and you will be subject to such sites' terms of use and privacy policies, over which NZS Capital has no control. In no event will NZS be responsible for any information or content within the linked sites or your use of the linked sites.
Nothing in this newsletter should be construed as investment advice. The information contained herein is only as current as of the date indicated and may be superseded by subsequent market events or for other reasons. There is no guarantee that the information supplied is accurate, complete, or timely. Past performance is not a guarantee of future results.
Investing involves risk, including the possible loss of principal and fluctuation of value. Nothing contained in this newsletter is an offer to sell or solicit any investment services or securities. Initial Public Offerings (IPOs) are highly speculative investments and may be subject to lower liquidity and greater volatility. Special risks associated with IPOs include limited operating history, unseasoned trading, high turnover and non-repeatable performance.