SITALWeek #446

Welcome to Stuff I Thought About Last Week, a personal collection of topics on tech, innovation, science, the digital economic transition, the finance industry, and whatever else made me think last week.

Click HERE to SIGN UP for SITALWeek’s Sunday Email.

In today’s post: plug-and-play solar panels could speed adoption; a breakthrough in wireless EV charging; drones, hydrogen-powered planes, and fancy robot vacuums; bringing transparency to US healthcare; a blockchain milestone for the California DMV and the prospects of a US bitcoin reserve; a blood test for memory loss due to Alzheimer's; the blameless consumer; the paradox of digital monopolies and why we should want Google to win; checking in on the weather forecast for the AI bubble. Welcome back to SITALWeek!

Stuff about Innovation and Technology
DIY Solar
The NYT reports on the rise of small-scale solar installations in Europe that can instantly provide power to a home simply by plugging the panels into a wall outlet. The plug-and-play solar panels can hang from a balcony and eliminate the costly, time consuming installation required to tie traditional solar power into circuit breaker panels and register with utilities. 
 
Remarkable Resonance
Oak Ridge National Laboratory has a prototype 270kW wireless charger, which they demoed using a Porsche Taycan. The wireless charger can take an EV from 10% to 80% in 20 minutes, rivaling the fastest DC superchargers on the market. This wireless breakthrough could potentially overcome many common issues with corded fast chargers, including broken plugs and maintaining liquid cooling systems. “The Oak Ridge system pairs a magnetic resonance transmitter pad with a receiver mounted to the underside of the car. The system’s secret sauce is its polyphase windings—lightweight electromagnetic coils arranged over a coil to generate a rotating magnetic field that eliminates current ripples and field cancellation.” And, if you’re wondering (like I was), yes, the 19 µT electromagnetic field is safe (it’s well below the 5 Gauss [500 µT] safety threshold that’s used in laboratory/medical settings; for reference, Earth’s surface magnetic field readings range from 25 to 65 μT). Wireless chargers require close proximity to the vehicle-mounted receivers (due to rapid decay in charging strength with distance). In other wireless power news, theories abound for using microwaves to beam power from space. A one square kilometer solar array in space could beam down 1GW of power, roughly enough for 875k homes. 
 
UTM for BVLOS
The FAA has given multiple commercial drone delivery operators permission to fly autonomously in the same airspace at the same time. Previously, only one drone company could occupy an airspace. Drones from Google’s Wing and Zipline can simultaneously operate in Dallas, Texas using the UAS Traffic Management (UTM) technology for Beyond Visual Line of Sight (BVLOS), which enables drone-to-drone communication. This video from Zipline shows how UTM works.
 
Hydrogen Heli
The EVTOL maker Joby conducted a 523-mile flight using their emission-free H2FLY hydrogen-electric technology. Joby’s CEO JoeBen Bevirt commented: “Imagine being able to fly from San Francisco to San Diego, Boston to Baltimore, or Nashville to New Orleans without the need to go to an airport and with no emissions except water. That world is closer than ever, and the progress we’ve made towards certifying the battery-electric version of our aircraft gives us a great head start as we look ahead to making hydrogen-electric flight a reality.” Here is a short video of the record-breaking flight.
 
Rosier Roomba
iRobot’s new $1400 Roomba Combo 10 Max floor cleaner not only vacuums and mops floors, but can automatically clean and dry its mop head, as well as replace its cleaning fluid and empty its dust container. It’s the closest effort yet to a fully autonomous floor cleaning machine (here is a video demo). Perhaps next they could add a robotic arm to first clear the floor of socks and dog toys. I recently read a good post from iRobot cofounder Rodney Brooks on real-world robot deployments and also re-read his 2008 essay “I...Am a Robot”. Whether it's autonomous delivery drones, autonomous EVTOLs, complex robotic floor cleaners, or robot dogs that vacuum up cigarette butts on beaches, it’s encouraging to see so many robotic milestones for a multitude of form factors finally come to fruition.
 
DMV Digitizing
The California DMV plans to put over 40 million car titles on a blockchain to streamline transfer of vehicle ownership. If successful, this would be a milestone for creating blockchain-based records of real-world objects at scale. California’s DMV already has an app for storing your driver’s license on your phone (and digital ID access will soon be available natively with Google Wallet). The promise of holding legal possession of items on a blockchain appears to be moving closer to reality. In other crypto news, the US government’s notion of building a bitcoin reserve is logical and thought provoking. After going through the classic hype cycle and trough of disillusionment, blockchain technologies and currencies could emerge from their cocoon in the next decade with a new wave of innovation and adoption. 
 
Cost Plus Transparency
Mark Cuban’s success driving down prices for consumers and alleviating drug shortages with Cost Plus Drugs continues to inspire. And, his crusade against PBMs will hopefully prove successful. While the Internet has pulled the curtain back on many complex industries that price discriminate (while simultaneously increasing the ability for others to collude against consumers, e.g., witness the recent FTC probe into surveillance pricing), digitalization has been slow to impact healthcare to date. Cuban’s tactic to enable a healthcare cost revolution is deceptively simple: transparency. 

Miscellaneous Stuff
Circulating Brain Biomarkers
A new blood test can accurately attribute memory loss to Alzheimer’s in 90% of cases, a success rate that far exceeds the diagnostic success of either neurologists (73%) or primary care physicians (60%) – and is a far more accessible option than brain scans and spinal taps. The peptide p-tau217 test is indicative of amyloid plaques in the brain, but it could be a while before it’s approved for routine use. If a blood test for Alzheimer’s is commercialized, it could greatly reduce the time between symptom onset and diagnosis, allowing for earlier interventions. 
 
Sugary Resurrection
Leaf Brands is an independent candy company that acquires trademarks for dead candy brands that still have nostalgic hold over people’s sweet tooth. Examples include Astro Pops, Tart N’ Tinys, Hydrox, and Wacky Wafers. Niche brands have frequently been mismanaged and/or neglected in the hands of large candy conglomerates, leading to an early demise. Leaf revives such brands, recreating the candies through careful flavor testing and manufacturing outsourcing. Leaf has also created at least one novel candy of their own: 
Farts Candy, a chewy Nerd-like product. Now, most of the industry is controlled by four major candy companies, and because of their size, they are afraid to take any chances, [Leaf founder] Kassoff said. “Can you imagine an executive at Hershey in a meeting saying, ‘I have a great idea for a new product, it’s called Farts.’ They would probably be fired. We have a major problem in our industry and that is the lack of creativity and fun.” 
Farts are available in both Fruiti Farts and Sour Farts flavors.

Stuff About Demographics, the Economy, and Investing
Consumer Scapegoating
Back when I covered the teen retail/fashion industry (many, many decades ago) I got used to CEOs blaming anything and everything for weak clothing sales: weather, supply chain, the economy, elections, inflation, etc. In every instance, weak sales were actually a result of a fashion cycle misstep, changing consumer preferences, or, on longer time horizons, demographic shifts in their customer base. When sales were ahead of expectations, the companies, of course, took all the credit. I never once heard the CEO of Oakley say that the company beat the quarter because it was unexpectedly sunny outside. I’m reminded of this experience as I read many stories over the last couple of months of CEOs blaming consumer weakness for light sales. In every instance that I dug into, there was no evidence of consumer weakness. Indeed, despite the reflexive headline circus in business media right now that suggests we are imminently headed for (or already in) a recession, there’s scant legitimate evidence of struggling consumers. There are some signs that consumers are becoming pickier and returning to the borrowing/spending patterns that predominated before COVID induced economic chaos (a time period through early 2020 when the consumer and economy plugged along at a healthy pace). Looking beyond the scapegoating, I found that the underlying reasons for poor sales grouped into a few different categories: GLP-1 impact on consumer demand for a variety of products, ranging from grocery-store to luxury items; demographics (aging customers); changing preference/tastes (particularly as Millennials and Gen Z enter peak consumption years); and, perhaps the most common, companies that used the COVID disruption as an excuse to take advantage of customers are being displaced by innovators with a better value proposition. I am not suggesting CEOs are intentionally lying about their businesses, but I am saying that management teams are often the last to know the truth about their own products/services, and they are quick to assign blame to outside factors. And, CEOs are as highly vulnerable to the same algorithmic influences as everyone else, i.e., if they hear other leaders blame the consumer in the news, they too will adopt that point of view. In January 2023, 98% of CEOs queried by Ernst & Young predicted an imminent recession. Over eighteen months later that recession hasn’t happened, and it may still not come to pass anytime soon. So, next time you hear a company reflexively blame the consumer (or blame anything for that matter), your skepticism meter should be pinned at its limit. 
 
Search Win-Win
Last week, a federal judge ruled that Google was guilty of building an illegal online search and advertising monopoly. This ruling marks yet another example of US and EU regulators continuing to overlook the value of natural monopolies in digital platforms. In many cases, it makes sense to have a winner-takes-all product due to the network effects that accrue for all sides of the marketplace. In the case of Google, Pareto efficient auctions for advertisers assure maximum ROI and provide revenues for Google to invest in maintaining the best user experience. Competitive products are a click away, and it’s hard to argue that Google’s competitors weren’t deep pocketed enough to compete. According to the judge’s musings, Microsoft even offered to hand over Bing to Apple for free, but Apple understood the product quality mattered and that Bing wasn’t even close to being in the same league as Google. It’s possible the same winner-takes-all scenario will play out as our interaction with everything shifts to a conversational AI interface; however, LLMs currently appear to be more of a (very expensive) commodity than search engines ever were (which could change). While I suspect it’s likely this ruling against Google will be overturned on appeal, it’s worth thinking about the consequences if it were to stand, as the exercise illuminates some of the strengths and weaknesses amongst the various big tech platforms. As long as Google maintains a superior product, they would likely come out ahead financially if they were barred from paying a TAC (traffic acquisition cost) to Apple for search queries, an amount today that’s in excess of $20B a year. Further, Google’s AI model Gemini is well positioned to cement Google’s position as one of the gatekeepers of the digital world. Apple would likely be prevented from striking a new exclusive deal with another search or AI product, thus harming their cash flow. It would be difficult for Microsoft/OpenAI to build enough traffic in the consumer search market to create a network effect large enough to catch Google’s twenty-year lead (although Microsoft could still build enough traffic in the business world). Meta is in a unique position, given their current dominance in digital conversations, to build powerful AI+search products, but they have yet to prove such a concept would gain use on their social apps. Meta has recently amped up its war with Apple by attacking the iOS iMessage monopoly, taking WhatsApp to 100M users in the US. This move demonstrates Meta’s ability to flex its dominance in social distribution to drive product adoption. As for Apple, there are two schools of thought: One framework is that Apple could free ride the hundreds of billions of dollars in AI investments by Microsoft, Meta, and Google by ensuring the best product is available on Apple devices. Yet, it seems entirely plausible that Apple could be cut out of the AI platform transition and become solely a commodity hardware provider, or worse: a new AI-based operating system could crown a new hardware winner in smartphones or augmented reality devices, leaving Apple to become a very large footnote in the history of technology companies. If I were Apple, I would be hedging my bets by investing much more aggressively in homegrown AI technology. While we can’t predict the future of the search/AI markets in detail, we do know that government regulation of large industries is always a fruitless effort, with regulatory capture inevitably cementing monopoly power. If this topic is of interest, we have an overview and two whitepapers on regulating digital industries available here. 
 
Dream Big, But Keep an Eye on the Sky
Back in May, I discussed the risk of particularly dangerous situations (PDSs), alluding to the euphoria in the markets, particularly centered on AI. Longtime readers know that we’ve been discussing the exciting potential for transformer models (the basis of today’s LLMs) since 2019, and I am likely on the far positive extreme of the spectrum in terms of my expectations for the new technology over the next decade and beyond. But, history teaches us to always be situationally aware of where expectations sit relative to reality during technology hardware hype cycles. One of the ways to gauge the wild expectations in such technology investment cycles is to reverse engineer the implications. Over the Northern Hemisphere summer, as AI stocks were precariously perched atop steep stock increases, I read about predictions like Nvidia becoming a $50T market cap and GPUs garnering a $10-20T investment (which also implies a very lofty market value for Nvidia). While I wouldn’t rule anything out (that’s another important lesson from the history of technology investment cycles!), I think it’s important to remember the Catch-22 of a rapid rollout of AI. All technology investment cycles come with turmoil in the job market, which, historically, has played out over long time periods. The Industrial Revolution, for example, took centuries. Combustion engines took decades to replace horses. Even the wheel’s arrival around 6,000 years ago probably took a while to obsolete all the boulder luggers. Adoption of AI technology, however, has, at first pass, the potential to be dramatically compressed. For one, AI is the first technology that has a chance at replacing human brain power both computationally and at the executive function level, including reasoning, creative thinking, etc. Further, embodied AI operating in the physical world will replace human brains and brawn. And, unlike many other technology transitions, AI runs on existing cloud infrastructure, networks, and devices. However, in order to believe the multi-trillion-dollar investment figures being casually thrown around, you would have to reconcile those numbers with the extreme job destruction they imply, which would send the economy (which is ~70% consumption based) into a tailspin. In that instance, CEOs really could blame the consumer! Thus, a rapid deployment of AI to replace human brain and muscle power would cause a circular reference failure in the economy due to job losses. Further, you need to explain where the power would come from to replace all the metabolically efficient human workers. Therefore, rather than bank on AI displacing humans wholesale, you instead need to believe AI will have minimal (or slow) impact on human jobs and instead focus on new inventions and the next waves of scientific revolution in sectors like healthcare, energy, and material science (which might create net new jobs overall). We can also envision AI hardware markets in the trillions if they are underpinning an entirely new agent-based digital economy that does not take jobs away from humans. I don’t believe we are out of tornado season yet for the AI expectations bubble. As I noted in PDS regarding weathering storms: 1) always have access to a safe space, and 2) make sure you are looking in the right direction; you might think the storm is up ahead, but it could be funneling down on you from above. And, always lean on those lessons from complex adaptive systems. It’s important to balance skepticism and fear with optimism and hope for those AI-driven inventions we can’t even dream of yet.

✌️-Brad

Disclaimers:

The content of this newsletter is my personal opinion as of the date published and is subject to change without notice and may not reflect the opinion of NZS Capital, LLC.  This newsletter is an informal gathering of topics I’ve recently read and thought about. I will sometimes state things in the newsletter that contradict my own views in order to provoke debate. 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. 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, LLC has no control. In no event will NZS Capital, LLC 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.

SITALWeek #445

Welcome to Stuff I Thought About Last Week, a personal collection of topics on tech, innovation, science, the digital economic transition, the finance industry, and whatever else made me think last week.

Click HERE to SIGN UP for SITALWeek’s Sunday Email.

In today’s post: a look at the paradox of information as current digital trends reverse a long arc of rising information freedom; turning toys into weapons; a geeky look at ergodicity economics; and, much more below.

Summer Schedule: the next SITALWeek will publish on August 11th.

Stuff about Innovation and Technology
Onboarding Digital Agents
HR software company Lattice announced last week that their software now onboards AI workers. Lattice software can be used to set goals and give feedback to AI employees as well as place them in org charts so that managers can assess and hold them accountable for their performance. Lattice will also train AI employees on corporate branding and their roles in the company. I’ve previously discussed the inevitable trajectory of replacing jobs with AI (which robs AI of the constant stream of new information that it needs to sustain), but the thought exercise of human managers being held accountable for the performance of AI workers is bone chilling.
 
AI’s Walled Web
Since the dawn of the Information Age (and its predecessor, the printing press), conventional wisdom says that we are on a one-way path towards open access to information. It’s worth examining that theory more closely. As I was reading this article about the long saga over Google Chrome ending support of tracking cookies, the persistence of the outdated concept of the “open web” struck me as naive. While there have always been gatekeepers – from tribal councils, to monarchs and churches, to newspapers and social network algorithms – we are approaching an era where AI has effectively cemented the walled gardens of the mega tech companies, i.e., the closed Internet is already a fait accompli. I’ve long covered the slow vacillations between closed and open information systems from various angles. Thus far in the Internet Age, we’ve moved from dial-up platforms like AOL to the first search engines that enabled the dotcom era, then on to the social networks that have aggregated content (alongside the mobile OS and app store duopoly); and, now, we are entering the AI filter era, which will consume the Internet only to spit out walled garden tea-party conversations. At this point in time, it seems laughable that unadulterated information could ever experience a resurgence, barring some seismic event. 
 
I am reminded of an exchange between Stewart Brand and Steve Wozniak that occurred 40 years ago in which Brand remarks: “On the one hand you have—the point you’re making Woz—is that information sort of wants to be expensive because it is so valuable—the right information in the right place just changes your life. On the other hand, information almost wants to be free because the costs of getting it out is getting lower and lower all of the time. So you have these two things fighting against each other.” We might, perhaps, indulge a broader perspective here on this tension between free and expensive and speculate that, throughout human history, power grabs via informational gatekeeping have caused wars, rebellions, and other tragedies. Crosscurrently, the Scientific Revolution, the Renaissance period, and the rise of Western liberalism laid the groundwork for the Internet and the democratization of information. However, AI will be the mechanism that will reverse this centuries-long divergence, once again swinging the pendulum in favor of information control by autocratic gatekeepers. Let’s call it wronging a right. Whereas previous digital technologies led to near-zero marginal cost for growth and consumption, AI seems to buck this trend (at least for now; however, inevitable efficiency gains should eventually revert course, we can hope). Currently, AI is so expensive to create and maintain that only a few companies can afford to do so. Therefore, it’s collapsing the free-expensive tog-of-war into a strange paradigm where information will only be valuable when accessed through a small number of new, expensive conversational interfaces and operating systems (note that, so far, evidence for the impact of Google’s AI search results is mixed, but it’s very early days). We’re also seeing accelerated aggregation occur for video (YouTube) and news/content/etc. (social networks). Although there still exists the possibility of an informational Catch-22 – whereby AI will run out of content to feed its insatiable appetite, thus stalling its ability to advance toward a higher level of intelligence (see: Will Generative Search Sideline its Teacher? and Digital Surrogate Delusion) – the natural progression from here is for AI to create content/data for AI engines/agents to consume and serve to users. 
 
In such a world of circular AI information, the decades-long shift away from objective reality will be complete. And, will anything still hold value when there is no space for genuine “human” output? (Granted, of late, humans have been so influenced by technology and algorithms that it’s hard to call anything uniquely human anymore; rather, the human contribution to civilization has been declining, and the percentage may soon approach zero – it’s been an odd consequence of the aberrant vector of information democratization combined with growth of connected technologies). After human input is thoroughly minimized, the next natural progression will be the creation of an AI economy, composed of trillions of agents interacting with each other, that supersedes and feeds off the increasingly irrelevant human economy. For humans, the existential will become inconsequential, and the least of our concerns will be tracking cookies on the Internet. As far as it goes for everything else we call life, I reckon there will be a backlash as we discover that an entirely hands-off-the-wheel approach might not work in the complex, analog world. Afterall, I don’t know if it’s possible to find meaning in the world without the lens of the human brain. As I posed in You Are Special?, when humans are no longer bigger than the game of life, will anyone want to keep playing?
 
Weaponizing Toys
Necessity has always been a potent motivator for pushing the technological envelope, and the Ukraine-Russia war has proven an exigent accelerant for AI/autonomous weaponry. Reading in the NYT about how toys – from racing drones to video game controllers – are being turned into weapons of war, I couldn't help but think of the plot of the 1992 Robin Williams movie Toys (which seems likely to have been inspired by the plot of Ender’s Game. Note: if you go looking for Toys, you will only find second-hand physical copies available). It’s an odd coincidence that Eric Schmidt bears some resemblance to the martially aggressive Toys’ character Lt. General Leland Zevo (played by the late, great Michael Gambon). In the movie, Zevo discusses his outlook for war with the higher ups of the US military: “You've gotta look to the future. The future is anarchy. I'm talking about lawlessness...I'm talking about the breakdown of the whole system. The military defending people against people...How can you possibly justify that one stealth bomber costs more than the government spends on cancer research in the whole United States? But 1 million little planes at $5000 apiece?...This is the future!”

Miscellaneous Stuff
Six Degrees of Anonymity
Actor Kevin Bacon is highly recognizable from his 106 acting credits. He recently went incognito in a popular LA mall: “People were kind of pushing past me, not being nice. Nobody said, ‘I love you.’ I had to wait in line to, I don’t know, buy a fucking coffee or whatever. I was like, This sucks. I want to go back to being famous.”

Stuff About Demographics, the Economy, and Investing
Genetic Ergodicity
Economist Jason Collins takes a look at ergodicity economics through a genetic lens in his latest post. Longtime readers will be familiar with the idea of non-ergodic systems, whereby an individual’s path through time rarely (and only coincidentally) overlaps with the average, expected outcome. It’s an important concept we first referenced in our 2014 paper Complexity Investing because it underpins the notion of just how difficult it is to predict anything. Ergodicity reframes many behavioral economics concepts that investors have been misguidedly taking for granted. Collins’ post is a great overview of ergodicity and ends by demonstrating there are some situations in evolution where certain seemingly irrational individual behaviors in a non-ergodic world are explainable by considering the ergodic genetic ensemble of a population. One might extrapolate that there are times when it makes sense to switch between (or combine) an additive and multiplicative framework for decision making, depending on whether you are considering the individual vs. group (or percentage vs. whole) perspective.

✌️-Brad

Disclaimers:

The content of this newsletter is my personal opinion as of the date published and is subject to change without notice and may not reflect the opinion of NZS Capital, LLC.  This newsletter is an informal gathering of topics I’ve recently read and thought about. I will sometimes state things in the newsletter that contradict my own views in order to provoke debate. 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. 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, LLC has no control. In no event will NZS Capital, LLC 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.

SITALWeek #444

Welcome to Stuff I Thought About Last Week, a personal collection of topics on tech, innovation, science, the digital economic transition, the finance industry, and whatever else made me think last week.

Click HERE to SIGN UP for SITALWeek’s Sunday Email.

In today’s post: advanced tools for scientific research are proliferating, including the ability to short-circuit evolution; the Summer Olympics will feature millions of custom AI recaps from a legendary broadcaster; the difficulty of rapidly deploying AI copilots is underscored by the simple math of the situation; extraterrestrial oceans; a 30-year low in Hollywood employment confronts the difficult truth of content value; a lesson for companies that fail to invest in the future comes from Walgreens' 80% decline; and, much more below.

Summer Schedule: SITALWeek will publish next on July 14th and July 20th. We'll return to weekly newsletters starting on August 11th.

Stuff about Innovation and Technology
Prompting Novel Proteins
A company called EvolutionaryScale has created a language model called ESM3 that it claims will make biology programmable. “ESM3’s multimodal reasoning power enables scientists to generate new proteins with an unprecedented degree of control. For example, the model can be prompted to combine structure, sequence, and function to propose a potential scaffold for the active site of PETase, an enzyme that degrades polyethylene terephthalate (PET), a target of interest to protein engineers for breaking down plastic waste.” The company came up with a novel green fluorescent protein (GFP), like those that create bioluminescence in sea creatures, that was so sequentially divergent from other GFPs that naturally generating such a variant would likely have taken 500M years of evolution. ESM3 joins a host of new tools proliferating for scientists. Take for example Microsoft’s new research tools Generative Chemistry and Accelerated DFT, which are aimed at dramatically increasing the surface area of experimentation. The NYT also reports on Terray chips, each with 32 million micro-wells, for ultrahigh-throughput biochemical screening to identify drug candidates. 
 
AI AL
An AI version of legendary sports announcer Al Michaels is set to create as many as 7 million customized recaps of this summer’s Paris Olympics for NBC’s Peacock app. This will be the first Summer Olympics where all events will be available to stream with over 5,000 hours of live coverage. In related news, Google, via its Labs experimentation group, is reportedly working on a plan to allow your Gemini chatbot to be personalized as your favorite YouTube personality. 
 
Upside Down Economics of Human Replacements
There are a slew of AI copilots coming to a wide variety of jobs in the near term. The result is likely to have two primary impacts. First, copilot-enabled productivity will likely contribute to the weakness in job openings for white collar vocations. The WSJ reports on the 21% decline in job openings on freelance task platforms, like Upwork and Fiverr, as AI takes over such jobs as copy editing, translation, sales, marketing, and customer service. The second impact is that cheating on your homework is no longer just for students. It’s no secret that ChatGPT took off like wildfire for academics; but, with custom bots for jobs as senior as Wall Street investment bankers (both MS and GS have announced plans), everyone now can cheat on their so-called homework. Unfortunately, AI shortcuts may set you up for a major fail during final exams, when you may not be able to rely on AI tools. For important decisions, or for fixing mistakes made by AI, will you really know the material well enough to improvise, plan, and problem solve? Is it possible to create value for your company if your hands are off the wheel, or will you just crash? And, of course, AI copilots raise the ultimate question: why have workers at all? (See Digital Surrogate Delusion.)  Why not just have a slew of job bots chatting amongst themselves? And, take, for example, this NYT article on AI bots being used in the real estate sector to deal with tenants. How long will it be before those AI bots are dealing with the tenants’ AI bots? (See AI Negotiators). Is this what we’re aspiring to with AI – everyone cheating on their homework and having their bots talk to other bots? It’s a depressing Avatar Apocalypse. 
 
Depending on just how many tasks continue to be automated, we could transition from a “do more with the same” work force to “do more with far fewer” employees. The big consulting firms working with large companies are targeting 15-20% productivity gains, which can translate to 15-20% fewer employees all else equal. This scenario, of course, is the Catch-22 of AI: the faster companies deploy advanced tools, the faster they curtail jobs in the economy, leaving fewer people able/needing to buy their products and services. This notion circles back to when I pondered how the economy could expand without job growth. Of course, the transition into the AI tech era is the same as prior productivity waves with one apparent difference: major technological disruptions in the past have taken decades (Information Revolution) or even centuries (Industrial Revolution) to play out. However, if you believe the optimistic prognostications (as the stock market seems to), AI will have that level of impact on the job market over the course of just a few years. Such a shift would require capital investments in the trillions and net new economic activity several times that. Here’s a simple back-of-the-envelope calculation: if big tech platforms are buying a few hundred billion dollars’ worth of GPUs to run AI in a few years (as the market thinks they might), it implies $1-2T in capital investment (when you include data centers, memory, servers, networking, etc., not to mention the massive amount of energy needed!). And these GPUs could have a shelf life of only ~2-3 years if they are meant to exclusively run leading-edge AI models, shortening the required payback in revenues. Big tech platforms have gross margins for their infrastructure businesses anywhere from 60-80%. So, doing the overly simplified math, it’s not hard to see that AI investments would need to generate many trillions of dollars in revenue to accrue enough gross profit to justify the underlying capex expenditures (our internal models suggest that we would need roughly $5-$10 of GDP to justify every $1 of GPU investment). And, here’s the tricky part: AI would need to be deployed in such a way that it somehow doesn’t offset consumption by net job obsolescence in order to rake in those profits. So, wholesale replacement of humans seems an unlikely near-term AI scenario. And, even incremental job destruction, via leveraging the productivity of copilots, may not progress very far before its economic impact causes a revolt. Here is another back-of-the-envelope calculation: if you assume companies spend one-fifth of an employee's cost to replace them with AI, then $1T of annual AI spend could replace $5T in desk jobs. At an average of $80,000/y in salary, that's well over 50M jobs displaced (a figure that would grow as AI adoption grows). While my math in this paragraph is intended to be theoretical, it illustrates that AI will need to be deployed at a more measured pace unless it can create significant revenue upside without eroding employment.
 
If not human replacement, what, then, will the AI Age really be about? The real value for AI will likely come from invention. And, on that front, the promise of applying AI to scientific breakthroughs in healthcare, energy, etc. is tantalizingly close (like the examples in this week’s opening paragraph). So, while it’s tempting to allow people to cheat on their work at their jobs, that may ultimately be something we look back on as a flawed experiment, while the real value comes from applying the new tools to complex problems that create large new industries and applications that are net positives to the economy and our quality of life. 
 
Hollywood’s Hemorrhage
In 1948, two-thirds of the American population went to the movie theater every week. That year marked the peak of the industry, which fell by more than half by the end of the 1950s. The handoff between the big screen and the small screen was taking place as TV entered American living rooms. A similar situation is now facing the 100-year-old Southern California industry. The LA Times reports that employment levels for Hollywood-based content production are at a thirty-year low (excluding the disruptions from the first few months of COVID and recent strikes). Overall, industry employment is down from around 150,000 pre-pandemic to 100,000, and that does not include impacts to the large base of freelance and part-time workers. Factors influencing the downturn include new technologies making jobs more efficient, some shift of production to lower cost cities, and, of course, the swiftly rising tide of infinite content across YouTube, social networking, gaming, etc. 
 
When professional content began to shift to streaming, studios banked on subscriptions and ads to support their continued blockbuster spending. However, when the necessary subscriber base didn’t materialize fast enough to offset the decline in linear viewing (the very same linear television viewing that disrupted the movie industry seventy years ago), the streaming spending bubble began to deflate, and now it appears the last white knight, ad dollars, is in jeopardy as well. It’s becoming clear that expensive, studio-generated content is not any more valuable to viewers or advertisers than the plethora of lower cost alternatives. Certainly, there are examples of must-see entertainment from studios; but, as Variety reports, the industry is now facing an advertising problem. Disney has apparently been making major concessions in order to sign advertisers, leaving the other studios scrambling for the receding tide of ad dollars. It remains unclear if Hollywood can face its Vaudeville moment without a significant refocusing on high-impact, high-value content that engages users in a way that is equally as valuable to advertisers. Netflix co-CEO Greg Peters said in a recent interview that advertisers will value Netflix ads more. Perhaps. I have some optimism that ads might become higher value over time for certain bits of content, but it could still be a race against other entertainment alternatives to capture enough user time to display those ads. Despite my ongoing skepticism, I’d love to see a Hollywood renaissance, so, as this story plays out, I am rooting for what has now become the underdog: the storytellers themselves.

Miscellaneous Stuff
Oceanic Asteroid?
Samples from asteroid Bennu contained materials consistent with mid-ocean ridges on Earth, implying the flying chunk of rock could have originated from an ancient, water-rich planet. The water-soluble phosphates were found along with nitrogen, carbon, and other organic compounds, all critical to creating life.

Stuff About Demographics, the Economy, and Investing
Extinction Event
In the 2010s, a lot of companies propped up their stocks with buybacks while ignoring the deteriorating customer experience, effectively gambling with their own stock while ignoring the threat of increased digital transformation of the economy. A prime example is Walgreens, which announced a $10B share repurchase in the summer of 2018, while, speaking as a customer, their in-store experience continued to suffer. I discussed Walgreens and other legacy businesses in this September 2018 oped: 
The differences between companies that succeed in making the digital transition and those that don’t may become clearer as the era of ultra-cheap money ends. If they were serious about becoming successful platforms, 20th-century business models should have begun making the necessary technology investments more than a decade ago instead of pandering to yield-hungry investors who rewarded share buybacks and rising dividends with higher equity prices.
For companies that have under-invested in their futures, the ramifications are likely to be difficult to recover from.
Walgreens stock has declined around 80% since the aggressive share repurchases, and the company announced last week that they would be closing a significant number of underperforming stores in the US, perhaps numbering in the thousands. In an era of easy money, it was convenient to lever balance sheets and recapitalize companies by buying shares back (in the case of Walgreens, the decline is more complex, as the company shifted from buybacks to large acquisitions of doctor networks and urgent care clinics, while continuing to ignore the issues concerning their core business). The tide is going out with higher rates, and a reckoning is coming for companies in all industries that underinvested in transformative digital tech. Now, with the next platform shift to AI already upon us, an additional layer of investment is needed to keep up with customer expectations, and many of these companies are likely to face extinction as their debt costs soar and their customers find better alternatives. Many of the “moats” that once protected these large, relevant companies have become tar pits, dooming their victims to fossilization.

✌️-Brad

Disclaimers:

The content of this newsletter is my personal opinion as of the date published and is subject to change without notice and may not reflect the opinion of NZS Capital, LLC.  This newsletter is an informal gathering of topics I’ve recently read and thought about. I will sometimes state things in the newsletter that contradict my own views in order to provoke debate. 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. 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, LLC has no control. In no event will NZS Capital, LLC 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.

SITALWeek #443

Welcome to Stuff I Thought About Last Week, a personal collection of topics on tech, innovation, science, the digital economic transition, the finance industry, and whatever else made me think last week.

Click HERE to SIGN UP for SITALWeek’s Sunday Email.

In today’s post: storing energy in building materials; when AI avatars soon outnumber humans, will it change our perceptions? miswiring for more sophisticated comedy; the shaky rental prices that underpin apartment debt; a new king is crowned; and, much more below.

Stuff about Innovation and Technology
Concrete Capacitance
Researchers at MIT have created a supercapacitor from water, cement, and carbon black, effectively transforming inexpensive building materials into a battery. Supercapacitors are easier to charge than traditional batteries, but they also discharge more rapidly. This power dump wouldn’t be suitable for powering consumer electronics, but it could store grid-stabilizing energy generated from renewable sources. For example, the foundation of a typical residential house, if constructed with supercapacitor material, could hold enough energy for roughly one day of electricity demand. Another use case would be construction of roads that could store solar energy to wirelessly charge EVs in transit. 
 
Avatar Apocalypse
TikTok unveiled its Symphony AI creative suite last week, which, among other things, allows users to create realistic AI avatars. Influencers can now sell an infinite scroll of themselves to advertisers to virtually hawk products/services and interact with users. While the idea of AI avatars is nothing new, these tools are the first of their kind that will make for mass usage. In a similar vein, a new social app called Butterflies, which mixes humans and AI-generated characters, also launched last week. Meta has dabbled in avatars itself (see LLM Friends), and it would be relatively simple for the company’s social apps to suddenly go from a few billion people interacting with each other (and the mind control algorithms) to a few billion people interacting with a few trillion AI avatars. Talk about diluting reality! A topic I discuss often is the reality perception shift that took place in the 1970s, which started us on the path to today’s individual subjective reality bubbles. At this point, it seems like we have only memories of broadly accepted objective reality. It wasn’t that long ago that social networking shined a spotlight on inauthenticity, which one might have expected to cause a backlash. Yet the opposite has happened, as people have become even more entrenched in their own subjectivity. Similarly, one might expect a world of interacting virtual AI avatars to eventually cause some sort of dissociative break that puts humans back on a path toward objective reality. It’s hard to say if that red-pill moment will happen, but the prognosis is, perhaps, not great. I am still waiting for society to experience that reality-shattering crisis that Truman confronted in The Truman Show when he crashed his boat into the painted sky encasing his artificial life. The key difference between Truman and ourselves is that Truman didn’t wittingly turn over control of his life; we, on the other hand, seem to be willfully enraptured with advanced technological control. And, I imagine the algorithms and bots have an even more powerful view into – and ability to manipulate – our minds than did Truman’s director, Christof. At the time of its release, the concept behind The Truman Show seemed outlandish, but it appears life has now imitated – and surpassed – that prior art. And, let's not be too human-centric and forget about the AI agents themselves: imagine a trillion AI Trumans – would they too want to crash into the sky? If you’re interested in the ultimate trajectory of AI avatars, I once again recommend the largely unknown 2013 movie The Congress.

Miscellaneous Stuff
Miswiring for Laughs
The comedic brain has long been of interest to me for, among other reasons, its ability to inform investing. Comedian Jon Stewart recently defined comedy in the following way: “good comedy tries to articulate a feeling that people are having that hasn't quite crystallized yet, that isn't quite on the tongue enough to be able to be illuminated, and that's all we ever try and do.” The comment came near the end of a two-part interview on The Town podcast (Part 1; Part 2). Earlier in the second part of the interview, Stewart commented on AI: “AI is replacing the heart of what it is that we do...You know, we love to believe that the human soul will transcend and that AI will be a hollow facade of what it is that a true artist can bring to it, but I have no fucking idea. And the truth is, 10 years from now, it may be better than [Richard] Pryor, better than [George] Carlin.” I wrote about AI and comedy in #396, citing Stephen Wright and Conan O’Brien’s discussion of how comedic brains are miswired because they connect things that shouldn’t be connected (something I believe is at the heart of successful investing). Researchers at Google recently enlisted twenty comedians to help understand the utility of AI when it comes to comedy. Largely, comedians found it useful for a few technical things, but not for writing jokes or monologues. LLMs, by design, are prediction engines – just like the human brain – so it seems entirely possible we could engineer AI comedians, although it might require a bit of tinkering to give the bots the necessary degree of “miswired” creativity. Likewise, the current army of non-illogical bots doesn’t excel at comedy based on Stewart’s definition of “seeing something that’s not quite there yet”, (although their ultimate goal may be to predict the future). That said, it’s not hard to imagine AI becoming funnier over time and more adept at creating a long act, with call backs and connections across jokes. AI could easily run a near-infinite number of sets in a virtual comedy club to trillions of virtual audience members to determine precisely what is funny. In the meantime, it’s often quite good at “dad jokes” filled with painful puns. Stewart went on to speculate that, at some point in the near future, over a very short period of time, AI will take over the human job market. If that happens, I hope it gets better at comic relief.
 
Supernova Soup
In what is becoming a frequent feature here in SITALWeek, the JWST continues to yield results that indicate the Universe was a lot more active closer to the Big Bang than our current cosmological models predict. We’ve seen younger galaxies, more carbon, and now 10x more supernovas. All of these data are interrelated in that earlier star formation can lead to star explosions which can lead to the presence of carbon. The findings push the oldest observed supernova back to less than two billion years after the start of the Universe from the previous three billion years.

Stuff About Demographics, the Economy, and Investing
Tenuous Rental Debt
Here is something no economic textbook would predict: US apartment rents are the most expensive ever, yet occupancy rates are at a multi-decade low, and delinquencies on multifamily building debt are at an all-time high. Longtime readers will know how, at least in part, we got here: widely used software from RealPage encouraged apartment complex owners to drive vacancies of around 20-25% of total units; this artificial scarcity has allowed price inflation such that landlords come out ahead on a net basis, and many other rental property owners followed the price increases even if they were not using the software. Ideally, the free market should maximize for a market-clearing equilibrium of occupancy and prices, but, when enough apartment owners are using the same tools to fix prices, you get the current set of unusual circumstances. As for debt defaults, many of the apartment buildings were underwritten in a period of low rates based directly on assumptions about how RealPage’s software would drive revenues higher with lower occupancy rates. Thus, with sky-high rental rates already baked into the formula, the interest rate increases have left debt-saddled landlords with little room to maneuver, resulting in more defaults. Much of this story is recapped, including other ways RealPage software has boosted the costs of renting, in this recent article from The American Prospect. Given that rent drives as much as 40% of CPI – a key Fed indicator – and the importance of housing affordability to economic health, it’s possible RealPage’s actions are responsible for a compounding amount of inflation as wages have risen to cover affordability in some markets. One legislative solution to lower vacancy rates to historical levels and match market demand could be to tax any vacancies above the typical historical occupancy rate of the region (government access to occupancy rates in RealPage and other software tools could provide insight to policy). Housing was one of the first industries to benefit from the transparency and information rendered by the Internet. The complexity of the market, the heterogeneous nature of the dwellings, and diverse local dynamics created a classic set up for a “power to the people” Internet transformation (to borrow a term from Zillow co-founder Rich Barton). Given how easy it is to share data amongst companies with software and the cloud, I suppose it was inevitable that we also see an equally classic tactic of collusion tamp down that potential in the rental market. 
 
On Top of the World
In 2015, I was in a meeting with Nvidia co-founder Jensen Huang at the company's annual user conference GTC in San Jose when he was asked about an intellectual property lawsuit the company had filed against Samsung and Qualcomm in late 2014. I’ll paraphrase Huang’s response here in what is a mostly direct quote: “we have more graphics IP than every other company combined, and they are using our IP. We need compensation for it. But investors should assume we get nothing and it’s upside when it shows up. None of you would sell me our IP for even one billion dollars. It’s obviously worth a lot.” At the same user conference, I sat next to a post doc from MIT who explained the world of machine learning and the power of Nvidia’s CUDA programming language to me as we awaited Elon Musk, who would be taking the stage to recruit engineers at the conference for his self-driving technology. In my 26-year tenure as an analyst/PM working on various investment strategies, I worked on funds that owned Nvidia back in the dotcom bubble and then sat on the sidelines for a long time following the crash. What I learned at the user conference was enough to establish, at the time of the 2015 GTC, what we call an “Optionality” position in our investment framework. Last week, according to the unanimous conclusion from supply and demand of Nvidia stock by various AI bots, algorithms, passive investment vehicles, retail investors, derivatives markets, animal spirits, hocus pocus, and the small handful of remaining (now anachronistic) “active investors” at a single moment in time, Nvidia became the most valuable public company in the world. The $3T+ valuation is indeed a far cry from the ~$10B size of the company in the Spring of 2015. And, I would agree with Jensen – his intellectual property back then was indeed worth more than one billion dollars. The last company to surpass Microsoft and become the most valuable in the world (for a time) was Cisco (recounted by then-CEO John Chambers in this recent WSJ interview) in March of 2000 during another stock market PDS. Cisco stock is 40% lower today vs. its peak 24 years ago.

✌️-Brad

Disclaimers:

The content of this newsletter is my personal opinion as of the date published and is subject to change without notice and may not reflect the opinion of NZS Capital, LLC.  This newsletter is an informal gathering of topics I’ve recently read and thought about. I will sometimes state things in the newsletter that contradict my own views in order to provoke debate. 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. 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, LLC has no control. In no event will NZS Capital, LLC 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.

SITALWeek #442

Welcome to Stuff I Thought About Last Week, a personal collection of topics on tech, innovation, science, the digital economic transition, the finance industry, and whatever else made me think last week.

Click HERE to SIGN UP for SITALWeek’s Sunday Email.

In today’s post: fifteen years ago, right as the Great Recession was ending, US tech stocks began an epic run, compounding an astounding 20% per year through last week; a look at the opposing AI strategies of the big five tech platforms; Nature discusses the increase in custom vaccines for cancer treatment; greener steel; adults are buying more kids; toys for themselves than for young children; investors walking away from VC capital calls may signal a broader issue in illiquid assets; and, much more below.

Stuff about Innovation and Technology
Adjuvant Intelligence
Cancer vaccines could become a regular part of oncology treatment as progress is made in personalized mRNA shots and other methodologies, according to a recent news article in Nature. One of the factors conferring success has been the use of advanced technology to tailor treatments: 
A bigger differentiator, researchers say, could be the computational engines that help to determine the vaccine’s composition. Each engine has its own proprietary suite of tools that it uses to select which neoantigens to target.
Most companies start with genetic sequencing of data from tumours and healthy tissues to reveal the mutations that cropped up during cancer development. T cells will not recognize all of these mutations, however, so algorithms are used to prioritize a subset — Moderna uses up to 34, BioNTech up to 20 — that are predicted to have the most potent immune-stimulating effects.
Such predictions are made on the basis of various factors, such as the levels at which neoantigens are expressed on tumour surfaces and their anticipated binding to cellular receptors that aid in provoking a T-cell response. Machine-learning models then incorporate experimental data to improve the accuracy of these tools.
 
Green Steel
Greenifying the $1T steel industry, which accounts for 10% of global carbon emissions, is the object of numerous startups. For example, Colorado-based Electra has developed an electrochemical process for producing iron plates from iron ore (including low-grade ores typically discarded by the industry) without fossil fuels. Their methodology involves dissolving the iron ore in an acidic solution, separating out the iron ions, and depositing them on a door-sized cathode surface. The whole process requires temperatures of only 140 °F (60 °C) vs. the conventional pyrometallurgical process of melting iron ore in a coke-fired blast furnace at nearly 3000 °F (1600 °C). The iron plates can be further converted to green steel using an electric arc furnace. The ramp to green steel is a long one as Electra plans to scale to only one million tons by 2030, a fraction of the current two-billion-metric-ton industry. 
 
Fight or Freeze
With Apple finally declaring their AI intentions, I think it’s interesting to contrast the AI strategies of the five consumer/cloud/enterprise mega tech platforms, Apple, Google, Microsoft, Meta, and Amazon. What stands out to me is how differently the big five tech companies reacted to the “ChatGPT moment” in late 2022 when nearly everyone was caught flat footed by the sheer power and open-ended possibilities brought into being by the ability to converse with a computer intelligence. Microsoft had an inside line – thanks to Bill Gates’ work and their bankrolling of OpenAI – so their aggressive adaptation of their products to a conversational UI wasn’t too surprising. Likewise, Google and Meta both audaciously rose to the AI challenge. Google combined DeepMind with Google Brain and launched a series of impressive products, ranging from Search Generative to Gemini 1.5 and Gemma. Meta launched Llama and amassed a stockpile of GPUs in an attempt to garner developer attention. In stark contrast, Apple hasn’t moved much past the starting line. They released a series of small models that have a mediocre benchmark scoring, and they seem to be leaving more interesting AI use cases to developers, letting them choose from among competitors’ APIs (likely ChatGPT, Gemini, Llama, or a combination of small and large models) to build apps. Likewise, Amazon still seems frozen by the bright lights of AI. Rather than building their own AI models, Amazon has instead invested in startups like Anthropic. While I hate to generalize from just five examples, the differentiating factor in the urgency of reaction to LLMs seems to be largely connected to the role that company founders are playing. The founders of Google, Meta, and Microsoft are still actively involved, while Apple and Amazon are run by caretaker CEOs, who appear (from the outside) to be lethargic and/or pathologically risk averse. Apple and Amazon’s best bet from here is to enable developers. For Apple, that means giving developers full access to the native neural processes on their devices and working with all of the AI model creators to optimize on-device processing. For Amazon, that means investing in the right hardware to lure developers over with lower cost and/or better options than what Google and Microsoft’s clouds are offering. 

Miscellaneous Stuff
I Don’t Wanna Grow Up
Adult consumers in the US purchased more children’s toys for themselves than for preschoolers for the first time ever, according to research firm Circana. These self-indulgent purchases, driven by nostalgia and other factors, totaled $1.5B in the first quarter of 2024.
 
Adding Time to the Clock for Carbon-Based Life
New data from the JWST suggest that carbon was far more abundant soon after the Big Bang than previously thought. Earlier data pegged carbon’s presence at around one billion years after the Big Bang; however, the JWST spotted a cloud of carbon from when the Universe was only around 350 million years old. Carbon is formed inside large stars and then dispersed in explosive supernovas. The presence of heavier, star-forged elements like carbon and oxygen starts the clock ticking for when life-sustaining planets could have conceivably coalesced from stellar debris. However, as mentioned last week, the mix of circumstances that would actually yield a planet capable of supporting intelligent life (as we know it) is far more rare than we previously thought. 
 
Heritable Emissions
An attempt to curtail cow methane emissions reveals that some bovines are bigger offenders than others:
For free-roaming livestock, another promising option is to breed animals that emit less methane. For sheep or cattle, animals of the same size and identical diet can have methane outputs that vary by as much as 30 percent or 40 percent. “That’s a lot of diversity to play with,” says Montgomery. The trait appears to be as heritable as many other traits that breeders routinely select for, and breeders have already begun incorporating methane production into their selection criteria for Canadian dairy cattle, Irish beef cattle and New Zealand sheep.

Stuff About Demographics, the Economy, and Investing
Private Asset Malaise
Some investors are defaulting on their venture capital funding commitments, likely due to a variety of factors, including high interest rates, the dearth of IPOs, and a run of underperformance for early-round, high-valuation VC investments during the pandemic, according to Business Insider (the article cites a firm that is raising capital to buy distressed venture stakes). The multi-year low-rate environment that preceded and predominated the pandemic compelled many institutional investors to allocate a higher percentage of their assets to illiquid private equity, venture investments, real estate, etc. (in some cases, remarkably, investors actually borrowed money to do so). However, post-pandemic rate hikes have revealed potential strains on this strategy and have rendered other liquid asset classes more attractive. In some instances, institutional investors in need of liquidity are selling their PE stakes at steep discounts. It’s been difficult for many private assets to match the phenomenal returns of the public markets (see next section). BI reckons there could be as many as 8500 venture funds that have stopped investing for a variety of reasons, including a lack of funding from their investor base. At the beginning of the Fed rate hike cycle in June of 2022, I wrote about some of the issues that can arise when a highly levered system endures higher rates for too long in The Point at Which Higher Rates Collapse the Economy. The pain of higher rates has not yet become existential for asset classes rife with excessive amounts of leverage, but that point could arrive if rates stay high for another year or two.
 
Tech Stock Outperformance
The mighty run of US tech stocks has been something to behold. Fifteen years ago, markets were on their back as they worked through the Great Recession. Remarkably, since June of 2009, the Morningstar US Technology Index shows a total return of just under 1500%, representing a compound annual growth rate of just over 20%. (Note: The S&P 500 IT Sector shows an even greater 21.6% compound annual total return over the same period.) That’s an astonishing performance, and even more remarkable when you look at the power law dynamics, network effects, and cash flow that have driven a small number of companies to become a large percentage of the index. These mega cloud, software, and hardware platforms don’t just have big market values, they have become the drivers of the global economy, providing both growth and deflationary productivity gains across the ecosystems of companies/services they have enabled. 
 
Let’s return, however, to the miracle of compounding. A $100,000 US tech index investment made fifteen years ago would be worth over $1.8M today. If the same return rate were to continue for the next fifteen years, that number would balloon to $35M in 2039 – 350 times the original investment! Of course, not everyone experiences this level of return in the market – even if they bet on the right sector. The past fifteen years have been far from a steady trajectory, marked by bull runs, bearish punctuations, a pandemic, etc., and it’s this chaos that trips up otherwise savvy investors. Often, painful downturns induce investors to sell when they should hold/buy, while periods of market hysteria cause FOMO-susceptible investors to imprudently pile onto their positions. 
 
To see that everyone is susceptible to FOMO investing, let’s take a look at one tech investment made by perhaps the most famous investor, Warren Buffett. Nearly four years ago, in September of 2020 (in the middle of a bull market for tech stocks), I wrote about the IPO of Snowflake in the context of how to approach highly valued, very-high-growth businesses at the portfolio level for investors. (Note: I also followed up that post in early 2021 with some detailed thoughts on expected returns for high-growth software businesses.) I pointed out that, to my surprise, Berkshire Hathaway invested $730M in the Snowflake IPO offering. Post IPO, the stock was trading at around 100x the estimated revenue for 2021, which seemed a bit rich for Berkshire, particularly given Buffett's prior disdain for IPOs. According to filings, Berkshire still owns the same position in Snowflake, and they are roughly flat on the stock – with only a 4-5% gain – compared to a nearly 70% return in the S&P 500 over the same time period (and, the ~15% growth in shares outstanding since the IPO has diluted their ownership claim on the business). Of course, I am talking about only one stock that soured relative to the rest of the market, and Berkshire has done much better with their most famous tech investment, Apple.

It’s always difficult at any given time to know the right level of offense and defense to play when constructing a portfolio because, frankly, no one can know the future (we describe our strategy for portfolio construction under uncertain conditions in Complexity Investing; see also PDS). Our best guesses may turn out to be overly optimistic or not nearly optimistic enough. Who would have thought that, back in the summer of 2009, we were staring down a 20%+ annual return for the next fifteen years? At that time, the market was still in the depths of a historic correction following the collapse of “too big to fail” banks. By the end of 2009, hindsight showed that June 2009 marked the official end of the recession, perhaps proving that it is always darkest before the dawn – and highlighting the importance of staying grounded in optimism long term. Of course, if you live under the influence of social media and TV news algorithms like most folks these days, you might be amongst the 49% of Americans who believe the S&P 500 index is down, rather than up, this year. That’s a staggering level of delusion induced by algorithmic mind control. Hopefully, those folks didn’t sit on the sidelines for the last decade under a false sense of pessimism. While it's important to always have some element of defense, it's a good lesson to not let pessimism or cynicism keep you from missing the blue sky upside that can persist far longer than we might think.

✌️-Brad

Disclaimers:

The content of this newsletter is my personal opinion as of the date published and is subject to change without notice and may not reflect the opinion of NZS Capital, LLC.  This newsletter is an informal gathering of topics I’ve recently read and thought about. I will sometimes state things in the newsletter that contradict my own views in order to provoke debate. 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. 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, LLC has no control. In no event will NZS Capital, LLC 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.

SITALWeek #441

Welcome to Stuff I Thought About Last Week, a personal collection of topics on tech, innovation, science, the digital economic transition, the finance industry, and whatever else made me think last week.

Click HERE to SIGN UP for SITALWeek’s Sunday Email.

In today’s post: higher rates are perversely driving inflation higher; pathology gets an AI upgrade; the information vacuum that arises when AI converses reflexively; the art of cue cards; recycling old art forms; an intriguing addition to The Drake Equation leaves us far lonelier; and, a look at how nostalgia impacts waves of consumer demand. 

Stuff about Innovation and Technology
Pathological Progress
Microsoft has released an open-access whole-slide pathology model, GigaPath, trained on over one billion images in more than 170,000 real-world slides: “This is the first whole-slide foundation model for digital pathology with large-scale pretraining on real-world data. Prov-GigaPath attains state-of-the-art performance on standard cancer classification and pathomics tasks, as well as vision-language tasks. This demonstrates the importance of whole-slide modeling on large-scale real-world data and opens new possibilities to advance patient care and accelerate clinical discovery.” GigaPath has already achieved state-of-the-art performance versus other models. I wonder how long it will be before we have a whole-body pathology model?
 
Digital Surrogate Delusion
In a recent Verge podcast, the founder of Zoom, Eric Yuan, suggested that our virtual avatars will be taking meetings on our behalf, allowing humans to work less. Yuan also expects these avatars to answer emails, etc. While I’ve proposed creating virtual employees – ranging from historical to fictional – for various purposes, and I’ve even joked that we could have our AI agents replace us in meetings, the actual implementation of this latter idea would create a circular reference that would fail quickly. LLMs only progress via learning from humans and the content we create (there are scenarios where LLMs can train using simulated worlds, but we would need to see evidence of emergent properties and novel idea generation for those agents to be of practical value). Thus, if we were to hand off tasks requiring executive function to LLMs, they would pretty quickly be operating in a vacuum lacking new information and context. One estimate puts it as early as 2026 that LLMs could run out of content (and I’ve pointed out the risks of cutting off AI in Will Generative Search Sideline its Teacher?). If we were to pursue this idea of taking our “hands off the wheel” by replacing ourselves with avatars as Yuan suggests, we are likely to see an increase in volatility across the economy from higher velocity, convoluted decision making. Things tend to happen faster, and with more extreme outcomes, in the digital-only realm. There are other logical absurdities to the Zoom CEO’s plans, such as how would a new hire participate? (Why not just create new AI employees instead of hiring people that would have to be trained by their own AI agents?) What about when an employee leaves the company – does the company still own the person’s AI agent, and can that agent keep working for the company? Why not just train avatars for your entire employee base and then fire them all? My best guess is that, instead of agents tied to people, there will be independent agents designed for specific tasks. Those agents will communicate with each other and with us, and we will still interact directly with our carbon-based coworkers for the foreseeable future. Eric Schmidt says that if agents start talking to agents in ways that humans can’t follow, then we should panic and (literally) pull the plug. I disagree, as I think these agents working together will create a super-economy worth hundreds of trillions of dollars, but it’s likely to be tangential to our analog world.

Miscellaneous Stuff
On Cue
Wally Feresten has been holding cue cards for performers on Saturday Night Live for 34 years. Additionally, he has served on various other late-night shows and also runs a company of cue carders in NYC. He’s such an establishment at NBC studios that his “character” of cue-card Wally sometimes becomes the center of the comedy itself. Wally has to adapt cards (which are still handwritten and edited last minute) for readability and to match the rhythm and reading speed of each performer. This profile on Feresten is on a new site called LateNighter. The publisher is dedicated to covering the fascinating world of late-night comedy shows, and their podcast Inside Late Night is excellent if you’re as big of a comedy fan as I am.
 
Recycled Art
Ron Howard recently discussed his new Jim Henson documentary on Conan O’Brien’s podcast, commenting on how Hensen recycled and elevated an old art form: “But Jim was looking for that breakthrough. But it was a moment of transformation, and he seized that opportunity. But it's interesting that he wound up grabbing an ancient art form and elevating it. When I was a kid sitting around, we were talking about the Andy Griffith show. A lot of the character actors who would come on that show would be bitching and moaning because radio was dead or Vaudeville had gone. And yet we all know that those mediums didn't vanish. There isn't a Vaudeville circuit per se, but there's stand up, there's Cirque de Soleil. The art forms are as relevant as ever.” 
 
So, You’re Saying There’s a Chance?
Back in #315, I discussed the various inputs into the Drake Equation (which is used for estimating the abundance of communicative alien life in the known universe), specifically focusing on “L” – the length of time intelligent, radio-capable civilizations persist. I noted that L is typically couched in terms of how long intelligent civilizations survive before being destroyed by war or catastrophe. I suggested that maybe the bigger factor was whether birth rates can remain high enough to sustain an intelligent population, or if advanced technology’s degenerative effect on the “specialness” of life infects the civilization with melancholy, ultimately causing it to fade away. However, there are a number of interesting variables to ponder besides L, and recent research theorizes that we have been vastly overestimating the number of planets capable of supporting intelligent life (fi) by overlooking the criticality of mixed aquatic and terrestrial habitat and sustained plate tectonics. In brief, researchers theorize that primitive life must necessarily evolve in water while advanced life evolves on land, and plate tectonics support growth, biodiversity, and moderate evolutionary pressure. Thus, when planetary composition and tectonic action are factored into fi, it turns out that less than 0.003% – 0.2% of planets might potentially harbor conditions suitable for advanced civilizations to emerge. Effectively, it reduces estimates by around three orders of magnitude. The Drake Equation is, of course, just a thought exercise given the wide range of estimates for each variable; for example, the JWST discovering galaxy formation much earlier than expected could positively impact the potential number of advanced civilizations out there. The real value of the Drake exercise remains how it leads us to think about the longevity of our own species, and whether we will exit the celestial playing field with the proverbial bang or whimper.

Stuff About Demographics, the Economy, and Investing
High Rates Drive Long-Term Inflation
This unintuitive inversion of conventional wisdom (if you can call anything related to economic theory “wisdom”) is likely one of the reasons that Fed policy has become less effective over time. The combination of higher rates and systemically high levels of debt have several unintended consequences. One issue I’ve addressed in the past is that companies, especially those backed by PE with high debt loads, can raise prices to cover interest expenses. Another, perhaps more important, side effect is decreased investment. The problem is the Fed wants to slow the economy down because of their artificial 2% inflation target (which is based on nothing of consequence), but there exist key economic chokepoints that require significant investment in order to increase supply and lower prices. For example, rental rates comprise 40% of the core data the Fed uses to tune interest policy; however, higher rates have slammed the brakes on apartment development, despite the US seeing some of its highest absolute population growth on record, keeping upward pressure on rental prices. Likewise, the green transition – which will ultimately bring deflationary renewable energy – is hampered by high rates. This inflationary Catch-22 should be a self-evident absurdity for anyone paying attention and represents a black eye on Fed policy. As I’ve said in the past, inflation-fighting measures should focus on the sources of inflation, not the blunt tools that can actually drive the opposite effect. 
 
When Were You Eleven?
The answer to that question provides the decade that you will nostalgically remember as being the best of times. WaPo reports on the fascinating data that show everyone thinks that everything – from the best movies and music to trustworthy news and political unity – peaked during their second decade of life. After that, it’s all downhill. Some of the vertices are sharp, for example, music was clearly at its zenith during your socially/emotionally critical teen years (which is also why “pop” music is targeted at teenagers and why band popularity tends to track teen demographic trends as much (or more) than it represents enduring quality, e.g., the Baby Boomers and The Beatles). These data, of course, only reflect averages, not individual paths; but, if you’re an investor, this nostalgic skew is important to remember when you are researching various fads and fashions. For example, Harley-Davidsons sold well in the 90s and 2000s when Boomers hit their 50s, likely thanks to 1960s culture and movies like the 1969 film Easy Rider, only for demand to slacken as Boomers aged. The material desires and behaviors that might appear popular or unpopular may have more to do with demographically restricted nostalgia than with more general, enduring trends. 

✌️-Brad

Disclaimers:

The content of this newsletter is my personal opinion as of the date published and is subject to change without notice and may not reflect the opinion of NZS Capital, LLC.  This newsletter is an informal gathering of topics I’ve recently read and thought about. I will sometimes state things in the newsletter that contradict my own views in order to provoke debate. 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. 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, LLC has no control. In no event will NZS Capital, LLC 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.

SITALWeek #440

Welcome to Stuff I Thought About Last Week, a personal collection of topics on tech, innovation, science, the digital economic transition, the finance industry, and whatever else made me think last week.

Click HERE to SIGN UP for SITALWeek’s Sunday Email.

In today’s post: instead of social algorithms directly influencing people, there will be a content shift aimed at manipulating our AI agents; LLMs lose weight but still remain effective; a box office bust as movies remain more popular on streaming; Saturday morning cartoons; the big "wink" of the concert industry; shifting vices; a fix for housing turnover; AI is a better investor; the South Park stock index; and, much more below.

Stuff about Innovation and Technology
AI Swings Green
More evidence emerges that AI can emit far less carbon per page of text or image generated than humans doing the same task – upwards of 100 to 2000 times less. I discussed this often overlooked benefit in AI’s Energy Roller Coaster, and I’ve previously overviewed energy sources/demand in Pushing Electrons. Also of note, researchers continue to find that smaller AI models, consuming far less power, can be just as effective as the big LLMs. IEEE reports on 1-bit LLMs, including Microsoft’s work on baby AIs. 
 
Buttering Up the Bots
Today’s algorithms are tailored to influence us directly; soon, however, they will all be designed to influence our AI chatbots (a process that’s likely well underway already). When I see OpenAI signing deals with news outlets for access to content, one thought that crosses my mind is how the reporting of news will change. If news wants to get in front of you, it will need to appeal to your AI agent who will then in turn tell you about it. It’s a gatekeeper handoff from search/social networking to AI discovery engines. And, this transition is hardly limited to news. For example, if a potential romantic suitor wants to get your attention, their AI agent will need to woo your AI agent in your dating app. Likewise, snack companies will be creating ads tailored to your AI companion so they can convince you that you are in the mood for some Cadbury Chocolate. Content, personalities, products, etc. will all evolve to appeal first and foremost to AI agents, who will then whisper in our ears what all our desires should be. I've covered my fascination with the potential for an AI-agent-based economy that would dwarf our current analog economy in Simulacrum, Your Wish is Granted, and Creating a World of Binary Design. 
 
Box Office Bust
Memorial Day weekend usually kicks off the US summer holiday with a box office windfall. However, last weekend's moviegoers posted the worst showing since 1981 (adjusted for inflation). One contributing factor could have been last year’s Hollywood strikes, which pushed tentpole releases out until later this year, dampening early summer momentum. Afterall, it was just last year that we had Barbenheimer, so perhaps a dearth of good movies – and/or viewer fatigue with movies starring adults clad in tights and capes running around going “pew pew pew” – is to blame for the poor theater turnout. That said, we do know the box office faces significant challenges, with streaming being perhaps the most significant. Small-screen movie viewing remains strong, with releases like Road House, which went straight to Amazon Prime, garnering buzz. And, of course, consumers’ entertainment options are structurally shifting to an unending menu. While this Memorial Day weekend could have simply been an anomalous stinker for moviegoing (although all releases received positive reviews from audiences), it still represents yet another example of fragmented culture, with everyone consuming niche content at scale. And, in case you were wondering, 1981’s Memorial Day weekend featured such forgettable titles as Bustin’ Loose, The Four Seasons, Outland, and The Legend of the Lone Ranger. Meanwhile Netflix released its updated global viewing stats, and movies are still dominating. Leave the World Behind garnered 121M views while animated film Leo received 96M. 

Cartoon Connection
I know everyone is waiting with bated breath for me to reveal my latest YouTube content obsession. Recently, I’ve been watching full blocks of recorded 1980s Saturday morning cartoons, complete with programming and commercials all as they originally aired. It’s a nostalgic experience similar to what I discussed in Lessons from Vintage Advertising. While there are some shows I had forgotten, there were also a couple I had erroneously included in my childhood Saturday morning lineup. I don’t know when this memory cross-contamination occurred, but that’s the nature of the brain. I also found it interesting that shows I thought I had watched when I was around ten actually only aired when I was four to five years old. It’s surprising how early in life strong memories can form. In its 20th century heyday, watching Saturday morning cartoons may or may not have been a global phenomenon, but it was certainly a ritual in the US. I found a website that logs historical TV lineups, and here is mine from 1986. It’s likely I could find common ground with anyone in my age bracket who grew up in the US, all based on Saturday morning cartoons and their compendium of toys. Forty years from now, however, no one will have that kind of shared cultural connection, except of course for Bluey. 
 
It’s Not What You Think
I’ve been an investor long enough to have watched firsthand the concert industry’s transformation over the past three decades. Until the mid-to-late 1990s, concerts in the US were booked through regional promoters, many of whom were (and still are) legends in the industry. Businessman Bob Sillerman rolled up most of the promoters, as well as many of the large amphitheaters and small clubs, and IPO’d the concert promotion business as SFX Entertainment in 1998. At the time, I was a young analyst covering the stock, tracking daily concert attendance around the US. Ticketmaster (the creation of Fred Rosen) was also garnering an increasing number of exclusive contracts to sell tickets at many venues, including those booked by SFX. Fast forward a couple years, and SFX became part of radio-giant Clear Channel Communications (now known as iHeartMedia), was spun out as Live Nation in 2005, and then merged with Ticketmaster in 2010. Recently, the Justice Department reversed course and sued to break up Live Nation and Ticketmaster. The idea is that concert ticket prices are way up, so obviously a monopoly is to blame. But, there’s a giant wink in the industry that is being overlooked here: Live Nation+Ticketmaster exist as a single dominant company because it’s in the best interest of artists’ pocketbooks. Those big fees you pay on top of a concert ticket largely go to venues, which also have direct booking relationships with artists, and those fees can impact how much money an artist can pocket. In some sense, it’s a complicated shell game so that artists can claim lower ticket face values while they are benefiting from up front deals and various guarantees on the backend. As I’ve noted in the past, a large part of the music industry operates as a specialized bank for artists (see Beat It Mr. Tambourine Man), paying them now in return for future song plays or concerts. One of Sillerman’s great “innovations” was guaranteeing large per-show minimums to major acts, effectively taking the risk out of touring and shifting it to the promoters. It’s not exactly lucrative to be a concert promoter; Live Nation itself makes a net operating profit of 1.7% on concert revenues. It frankly used to be a better business for promoters when it was fragmented and they were getting paid to take more risk. Now, however, the incremental economics have largely flowed to performers, as perhaps they should, given that risk has declined for selling out most shows (social media has helped increase the number of seats sold for most performances). And, it’s the performers that set ticket prices. It’s the performers who have deals with the promoters and venues to make a percentage on ticketing fees and other deals (merchandise, concessions, etc.). Last year, Live Nation and Ticketmaster pulled in a little under $23B in revenue and made just $563M in net income. It doesn’t exactly scream monopoly rent-seeking. At the end of the day, what fans pay for tickets of any kind is a free market exchange of supply and demand. My point here is not to defend Live Nation, Ticketmaster, or their merger, surely there are some examples where the company’s vertical integration has pushed concert inflation to the benefit of performers and detriment of consumers, but rather to demonstrate that things aren’t always what they seem. Most of the music industry is a specialized lending business, and it doesn’t seem to earn an outrageous rate of return. If the government is upset about high ticket fees, their only path is to kindly ask the performers to lower them. Good luck.  
 
Advanced Medical Warning
Renowned professor Eric Topol writes about the potential for AI to forecast disease risk, much the way AI has improved weather forecasting. “Take the example of risk for Alzheimer’s disease, where there has been progress for blood biomarkers, but that’s just one layer of data. The risk of an individual could be assessed more accurately with orthogonal genomic data that include the apolipoprotein ε4 (APOε4) allele present in about 20% of the population, and a polygenic risk score, which provides complementary, independent predictive value. So can retinal imaging. Add to that electronic health records with both structured and unstructured text, imaging, and lab test results. Ideally, details of sleep history and physical activity would also be included but could otherwise be available from an individual’s wearable biosensor. A recent study using machine learning was able to predict Alzheimer’s disease up to 7 years before diagnosis by integrating electronic health record data for cholesterol, blood pressure, vitamin D, and sex-specific features such as osteoporosis in women or erectile dysfunction and prostatic hypertrophy in men.” However, given the fragmented nature of medical records and barriers to sharing patient information, it could be a long wait before such systems can tangibly improve health outcomes. 

Miscellaneous Stuff
Bud Buzz
The arrival of instant coffee in 1938 caused a big increase in overall coffee consumption in the US, which peaked in 1946. Since then, it’s been a mostly downhill slide, with coffee consumption decreasing 50%. Meanwhile, for the first time ever in the US, the number of daily marijuana consumers has surpassed that of daily alcohol drinkers. And, marijuana isn’t the only substance taking share from liquid vices. Nicotine pouches are experiencing breakout growth, with brands such as Zyn up over 60% compared to last year. 

Stuff About Demographics, the Economy, and Investing
Denmark’s Housing Fix
With the US housing market frozen – in large part due to people locked into low-interest-rate mortgages – BI reports on a novel mechanism in Denmark to increase housing market velocity. Mortgage pay off amounts (the principal) can effectively move with changes in rates, much like the value of any bond would. So, if rates increase, you would owe less. Borrowers have the option of paying off their original amount owed or a lower amount if rates increase. It’s an interesting concept, but it would be nearly impossible to enact in the US owing to the political climate and complexity of implementation inside the massive home-lending industry.
 
Vulcan AI Investors?
A new study from researchers at Chicago Booth demonstrated that GPT4 outperforms investors in predicting earnings and stock market performance. Lauding an LLM for predicting earnings better than a human is perhaps damning with faint praise given that, in their tests, human analysts fared only a few points above a random walk (which is, of course, not surprising since accurately and precisely predicting the future is impossible owing to the dynamics of complex adaptive systems, and humans are especially bad at it thanks to cognitive bias). Putting aside that criticism, I have no doubt that AI investors will best their fleshy counterparts in a variety of performance objectives and outcomes, and investors who embrace and use AI agents intelligently will likely improve their odds of outperforming. But, personally, I’ve yet to encounter an investing tool that doesn’t worsen performance. Things like proprietary data sources, modeling tools, and expert networks all serve to increase investors’ pre-existing biases. If two intelligent investors speak to the same expert, it’s highly likely that one will walk away wanting to buy a stock while the other will want to short it. Why? It all comes down to cognitive bias – it’s the enemy of good investment performance and why the importance of the team structure matters more than anything (see How to Have Meetings that Don’t Suck for more on that topic). We humans generally hear what we want to hear and never want to admit when we’re wrong. It seems entirely possible to build AI investing tools that could either moderate or exacerbate human bias, and it will be up to the investor to distinguish between them. At NZS, we’re not overly focused on AI investors, but rather what we might call AI psychologists. In particular, we’re most interested in figuring out how to use AI to help identify bias and improve decision making, e.g., by using LLMs to query our prior decision making processes in the hopes of discovering where mistakes were made so we can avoid them in the future. It is possible, if models continue to improve, that LLMs trained on your investing process will know which stocks you want to buy before you do. And, perhaps it will be easier to take criticism from an AI therapist than a coworker, and we can actually make some headway in learning from our mistakes. 
 
South Park's S&P (Satirical & Prophetic) Index
The long-running satirical cartoon South Park, which takes place in NZS Capital’s home territory, is apparently good at picking stock market losers. According to data from Spectra Markets, companies mentioned (aka skewered) on the show underperformed 7%, on average, vs. the S&P 500 over the subsequent 12 months. One theory is that appearing in Matt and Trey’s crosshairs indicates that a company has reached its peak cultural awareness. The most recent special, “The End of Obesity”, ridiculed the disgusting US healthcare system designed to make people sicker rather than cure or prevent illness. The GLP-1 makers were targeted, and the show accurately detailed how compounding pharmacies can offer the same semaglutide benefits at a fraction of the cost. While it’s unlikely that South Park is going to fix the US healthcare system, their assessment of the problem in “The End of Obesity” was spot on.

✌️-Brad

Disclaimers:

The content of this newsletter is my personal opinion as of the date published and is subject to change without notice and may not reflect the opinion of NZS Capital, LLC.  This newsletter is an informal gathering of topics I’ve recently read and thought about. I will sometimes state things in the newsletter that contradict my own views in order to provoke debate. 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. 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, LLC has no control. In no event will NZS Capital, LLC 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.

SITALWeek #439

Welcome to Stuff I Thought About Last Week, a personal collection of topics on tech, innovation, science, the digital economic transition, the finance industry, and whatever else made me think last week.

Click HERE to SIGN UP for SITALWeek’s Sunday Email.

In today’s post: some impressive new AI demos show us a clear path to the shape and timing of the next technology platform shift – whether the incumbents of today dominate the future is an open question; Waymo is ramping rapidly across multiple cities; AI is overwhelming scientific journals and HR recruiters alike; a boat painting; a disappearing band; and, much more below.

Stuff about Innovation and Technology
“Magic and technology
Voodoo dolls and chants, electricity
We're makin' weird science
Fantasy and microchips
Shooting from the hip, something different
We're making weird science, ooh” -Oingo Boingo
According to Google’s Android Ecosystem president, Sameer Samat, AI represents a “once-in-a-generation moment to reinvent what phones can do,” stating that Google is “going to seize that moment”. The shift from a multitouch UI to a conversational, multimodal, AI interface for operating systems/apps will be as momentous as when Apple ushered in the current multitouch era with the first iPhone 17 years ago. Last spring, in Discovery Engines, I wrote:
The first generation of massive discovery platforms, like Google, have dominated the Internet for years; however, with the arrival of AI chatbots, the way we discover and connect with everything will evolve. Chatbots are rapidly becoming platforms, e.g., with ChatGPT’s embedded “plugins”. I’ve been preoccupied with this transition from search- to chatbot-enabled discovery since I wrote AI Companions over a year ago: “As aware agents that know you well and have access to your accounts, messages, and apps, chatbots are ideally positioned to displace the tools we use today like Google search and other habitual apps.” And, given their ability to function as full platforms, I now believe chatbots could take over as the dominant mobile operating system and app store in the near future (or, if that does not happen, they will be entirely embedded in iOS and Android). 

OpenAI and Google both showcased their latest AI products at events last week. I would characterize OpenAI’s new ChatGPT-4o – which has more conversational, real-time, multimodal (images, video, etc.) processing capacity – as an evolution within the current state of AI that makes it easier to interact with but not a leap forward in intelligence. Most of the ChatGPT-4o demos felt very much like Pi, which Microsoft (aka OpenAI’s bank) recently acquired. In contrast, Google I/O showcased a far more impressive set of applied AI for various practical use cases. Google also demonstrated its multimodal, conversational AI interface using augmented reality glasses. It certainly seems as though this new form of human-computer interaction will benefit from an always aware hardware platform, which does seem to dictate a form-factor shift to intelligent glasses – or perhaps earbuds with a camera in them (which Meta is rumored to be working on). OpenAI’s demos felt more like the 1985 John Hughes movie Weird Science, where two teenage boys create an artificial girlfriend (even if you don’t remember the movie, you might be familiar with the eponymous Oingo Boingo song), while Google’s demos felt more visionary and practical.
 
Technology platform shifts begin slowly, but eventually become inevitable. If the smartphone is indeed reinvented as an always-on, multimodal, conversational AI interface integrated into AR glasses (or other form factors), Apple may maintain dominance in hardware, or suffer obsolescence – either outcome seems equally likely over the next decade. Initially, the smartphone will be the central processing unit for AI, but the AI itself will supersede the mobile OS and the current multitouch hardware interface. Whereas the mobile OS and app store were the fabric of the last major innovation cycle, the new development platform will be the LLM. In this scenario, Apple will need to lean more heavily on the innovation of other companies, given that they continue to squander their resources by buying back a record amount of stock instead of keeping ahead of the AI technological curve. Imagine staring down at what’s likely to be the largest technology platform shift ever and choosing to spend your money buying shares back!? Bloomberg has reported separate rumors that Apple is indeed looking to partner more deeply with Google’s Gemini and OpenAI. And, of course, we cannot discount Meta, which is uniquely positioned as the conversational nexus in the daily lives of billions of people and continues to innovate on hardware form factors and open-source AI models. 
 
Regardless of how it all plays out, last week’s unveilings were impressive. In case you missed the announcements, here are some clips from Google I/O of real-time multimodal Gemini using either a smartphone or AR glasses. There are more such examples from Google’s Project Astra on its website. Google also demoed real-time, on-device AI speech analysis to detect fraudulent phone callers. And, here is a demo of the new ChatGPT-4o multimodal interface as well as a demo where two different GPT-4os end up singing to each other. 
 
It’s been six years since I first experienced an early version of the virtual AI assistant Mica in an augmented reality headset. At the time, it took a little imagination to see what the future had in store. Today, however, we’ve reached that rare moment in a phase shift where it no longer requires much imagination to determine what the next software/hardware platforms will be and when they will arrive for the mass market. If there is going to be a shift away from the incumbents (as has happened in every prior tech shift but is not guaranteed), it will be led by the software developers – whatever platform they embrace to create the next generation of blockbuster apps will dictate the winners. It’s a wide range of outcomes, and I think it’s a coin toss whether today’s tech winners remain dominant 5-10 years from now or are entirely displaced by something different.
 
Problematic Paper Mills
AI is particularly good at creating written content, which is causing an arms race that’s overwhelming recipients. In the case of recruiting, AI has made it easy for job applicants to apply to 100x the number of openings as they could in the past, flooding HR departments, who, in turn, have to use AI to handle, respond to, and even interview applicants. Meanwhile, in the field of scientific research, journals are shutting down after being inundated with AI papers. Journal publisher Wiley is shelving 19 journals after retracting more than 11,300 fraudulent papers. 
 
Way Mo’ Waymo
Google’s autonomous taxi service Waymo is now serving 50,000 rides per week in Arizona and California. This number is up five-fold from a reported 10,000 rides last year, and the service has a stated aspiration of growing 10x annually. Some extrapolation: if growth does proceed at that pace, it would yield a run rate somewhere in the ballpark of 25M annual autonomous rides in a year or two. That would be an amazing feat, but they would still have only a tiny fraction of the rideshare market. For context, Uber completed 26M daily trips in 2023 (including both rideshare and food deliveries). Uber also partners with Waymo to provide rides and autonomous food delivery.

Miscellaneous Stuff
Letter of the Law
A Seaside, California man was ordered by the city to build a fence to obscure the view of his boat from the street. So, he did. And, on that fence, he had his graphic artist neighbor create a beautiful painting of the boat, which blends seamlessly with the top of the real boat peeking out from behind the fence. It’s a perfect protest of local bureaucracy. Seaside’s city manager responded: “The only action I’m going to take is a high five, and that’s it.”
 
Chuck’s Out
The Chuck E. Cheese kid-oriented restaurant chain has been through some tough times. While it may not be familiar to our international readers, the once popular destination for arcade gaming and consuming bad pizza has long been an institution in the US. Often described as creepy, the locations feature an animatronic band of musical critters (and a pizza chef on drums). If you caught the movie Five Nights at Freddy’s (or the game series it was based on), you might have some sense of what I am talking about. Apparently, times have changed, and kids now want to stare at screens instead of nightmarish puppets. The company is removing all of the mechanical members from over 400 locations, leaving just two spots where the band will play on. Farewell to a childhood memory.

✌️-Brad

Disclaimers:

The content of this newsletter is my personal opinion as of the date published and is subject to change without notice and may not reflect the opinion of NZS Capital, LLC.  This newsletter is an informal gathering of topics I’ve recently read and thought about. I will sometimes state things in the newsletter that contradict my own views in order to provoke debate. 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. 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, LLC has no control. In no event will NZS Capital, LLC 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.

SITALWeek #438

Welcome to Stuff I Thought About Last Week, a personal collection of topics on tech, innovation, science, the digital economic transition, the finance industry, and whatever else made me think last week.

Click HERE to SIGN UP for SITALWeek’s Sunday Email.

In today’s post: soil-based power networks; AlphaFold 3; an AI avatar explosion; the evolution of standup; ubiquitous YouTube; healthcare and housing are stifling discretionary spending compared to 30 years ago; how to overcome befuddlement; and, much more below.

Stuff about Innovation and Technology
Soil Power
Researchers at Tennessee Technical University are able to power sensors throughout a field by transmitting power through the soil, or “TTS” as they call it. A solar-powered transmitter broadcasts current underground, which charges nearby sensors. The sensors, in turn, can communicate back data, such as soil moisture levels. The system overcomes the difficulty of powering remote devices in terrain that is constantly being tilled and replanted. So far, researchers have demonstrated proof-of-concept with a 2-acre plot, and they hope to scale up to 30 acres. Alternatively, farmers could use genetically engineered plants that emit a fluorescent signal when they are in distress. The calls for help can be detected by sensors attached to drones/tractors so farmers can treat the problem (e.g., fungal infection, nitrogen deficiency, moisture issues, etc.) before it affects the entire crop. This tech, being developed by InnerPlant, is in field trials with soybeans. I live near a lot of agtech test areas, and I recently saw an octocopter drone autonomously zipping through a field of strawberries spraying some sort of liquid (likely weed killer or fertilizer) in a precision manner at fairly high speeds. Agtech overall has been slow to have a major impact (beyond the obvious genetic modification of seeds that has been going on for decades). It’s difficult to guess what might cause an inflection point that would increase the uptake in tech broadly for the sector. 
 
Biotech Boom?
The head of Google’s DeepMind, Demis Hassabis, told Bloomberg that he sees the company’s efforts in biotech becoming a $100B business. The comments come on the heels of the launch of AlphaFold 3, which can now model proteins, DNA, and RNA – and predict how they will interact with various small molecules. AlphaFold 3 differs from prior versions in that it used generative AI training methodology. DeepMind’s biotech company, Isomorphic Labs, which works with major pharma companies to develop new drugs, posted this overview of AlphaFold 3. 
 
rAIsing the Dead
AI avatars of deceased relatives are gaining popularity in China, according to MIT Technology Review: “In Chinese homes, it’s common to put up a portrait of a deceased relative for a few years after the death. Zhang Zewei, founder of a Shanghai-based company called Super Brain, says he and his team wanted to revamp that tradition with an ‘AI photo frame’. They create avatars of deceased loved ones that are pre-loaded onto an Android tablet, which looks like a photo frame when standing up. Clients can choose a moving image that speaks words drawn from an offline database or from an LLM.” The avatars currently cost around $1,000, but a phone-based version is in the works that could cost closer to $140, according to Zhang’s company. The rise of open-source LLMs, a competitive market for AI avatars, and a growing record of people’s personalities from social posts, emails, texts, etc. could make such avatars of the deceased readily available for the masses. I tend to dislike the overuse of the famous line from The Tempest:, “O brave new world that has such [dead] people in it”, but it does seem appropriate here. In another surreal example of AI avatars, the founder of dating app Bumble, Whitney Wolfe Herd, recently suggested your AI representative can go on dates with other people’s AI reps to see how you hit it off before engaging in real life. Meanwhile, the WSJ reports on the growing number of professional fashion models hanging up their outfits and fully licensing their AI avatars. I think we will be surprised by the ease with which people will be willing to clone themselves for a wide variety of use cases. I wonder how long before we’ll be able to make AI clones of our pets?

Miscellaneous Stuff
CoMEdy
Live standup comedy has nearly doubled its US ticket sales, going from $475M in 2019 to just over $900M in 2023. Bloomberg attributes much of the surge in popularity to Netflix prioritizing the art form (as does Seinfeld in this podcast clip, which is part of a longer conversation that contains a lot of commentary on the evolution of standup). Netflix may have taken US standup global, but they, of course, didn’t invent the televised standup comedy special. Indeed, the origins can be traced back to Robert Klein’s HBO series that began in 1975. But, perhaps the ubiquity of Netflix, the pandemic’s streaming surge, and the endless drone of comedy podcasts have all combined to elevate our collective awareness of comedians. Long-time readers will know that I’m a big standup fan, but I have to admit I am not as much in love with the art in its current form. Standup comedy has become more personal and less observational. Before Lenny Bruce and George Carlin, a lot of standup was the equivalent of knock-knock jokes and one-liners. Slowly, it became a more introspective medium. That trend has gone extreme over the last few years, and, now, perhaps because of cancel culture, comedians seem overwhelmingly fixated on joking exclusively about their own lives. While there are of course many exceptions, a lot of the most popular “standup” today is in the form of self-roasting or something that feels more like a one-person show. I find myself tuning out/turning off these personal narratives masquerading as standup comedy, but I appreciate that all art changes with the times. So it goes. For me, what’s missing is the editing – just give me the funny, as Seinfeld says, I don’t need the self-loathing and personal drama. 
 
YouniversalTube
Howard Stern did an excellent interview with Biden a couple weeks ago, and Sirius recently put it up on YouTube for non-subscribers. Stern remains one of the best living interviewers, and I always enjoy his long-form conversations. In a sea of a zillion podcasters, very few of them get the basics right about what makes a good interview: do your research, disarm the guest, and keep the focus on them, not you. I think this is even more important for maintaining relevance as podcasts go video, as YouTube (14B videos!) is fast becoming a hub for this evolved version of the medium. The NYT reported on this video podcast transition. It seems increasingly inevitable that YouTube will be the ultimate hub for videos of all types, including Hollywood content. According to Nielsen 150M people a month are watching over one billion hours of YouTube a day on television sets in the US, which puts YouTube at 9% of total viewing head of Netflix's 8% share in the living room. I think traditional media is beginning to realize which way the wind is blowing, and we can expect to see more and more content from the streaming apps co-located on YouTube. YouTube remains a great non-zero-sum model for creators and consumers, and the Hollywood Reporter has a story on just how big productions are getting for the video app. See also my piece titled Ambient YouTube for more.

Stuff About Demographics, the Economy, and Investing
“The Smartest People in the Room”
Congrats to Ben and David on an excellent WSJ profile of their deep-diving, business-focused Acquired podcast. At NZS we of course have our two favorite episodes (here and here)  ;-)
 
It’s Not the Avocado Toast
In the US, the peak bolus of Millennial births started in 1989 and went on for a few years. BI has an interesting look at how folks aged 25-34 spent money in 1989 compared to how that cohort of Millennials is spending money today. Using BLS data (inflation adjusted to 2022 dollars) some interesting trends emerge on this three-decade time horizon. Overall, annual spending is up 8% to around $68K. What stands out most is how spending in many categories – such as apparel, entertainment, and even food – has dropped vs. 1989, thanks likely to technology, productivity, and globalization. A few categories have been the clear spending gainers, most notably health care and housing. Housing alone accounted for 66% of the net increase in spending, and adding in healthcare gets you to 95% of the net increase. Meanwhile, alcohol and tobacco spending declined dramatically. Surprisingly, despite recent inflation in food prices, overall food spend declined – largely driven by drops in dairy and meat pricing/consumption. What emerges is a clear picture of how the megatrends of globalization and productivity drove down spending, only to be more than offset by the hyperinflationary choke points of healthcare and apartment rentals/homeownership.
 
Market Paradox
I was reading this WSJ article about how the complex impact of GLP-1s is “befuddling” investors: just when you think some company is a victim of GLP-1 usage, you get proven wrong because they sign a partnership or find some other path to survival. This led me to contemplate one of the biggest paradoxes of the stock market. For most publicly traded companies, there are far more ways to live than die, more ways to grow than shrink, more ways to survive/adapt than stagnate. Even failing companies are often saved through M&A. Indeed, vanishingly few companies actually go away permanently; rather, they live on as some appendage of another company or slowly lose their market value. However, if you are an investor in those same companies, there are far more ways to lose than win versus a broad index. Therein lies the paradox: how could the positive trajectory of most publicly traded companies (manifest in a broad stock index) not make it easy to invest? Last week, Buffett made a comment related to this same subject that befuddled me. In indicating that Greg Abel, the head of the Berkshire energy businesses, should take over the books of public companies that Berkshire owns, Buffett said: “I would leave the capital allocation to Greg. He understands businesses extremely well, and if you understand businesses you understand common stocks.” Would that it were so simple. Buffet's own performance, as well as that of Berkshire's investment team, have lagged the S&P 500 over the last ten years. According to the FT, Buffet's stock picks rose 10.2% per year while the Berkshire investment team rose only 7.8% per year compared to a 12% return for the S&P 500 over the period. Compounding that underperformance over the last decade results in a wide lag in value creation. Berkshire Hathaway stock has even underperformed the S&P 500 for the last decade. The solution to this performance paradox is the key to being a successful long-term investor. Unfortunately, I don’t have a pithy answer or silver bullet; however, I do tend to think the solution lies far less in the choice of individual stocks and far more in how you own them, which comes down to portfolio construction and team processes – it’s critical to have frameworks and systems in place that let you learn from mistakes and continually improve decision making. (We’ve tried to address some common pitfalls and mental traps working against investors in our original paper and subsequent follow ups). Of relevance to this conundrum is an op-ed from Tyler Cowen that looks at how wrong economists’ predictions, from a book published in 1980, turned out to be. One salient takeaway is that we never really know which thing we should worry about, but we always worry too much in the face of the resilient systems of capitalism and democracy. I suspect humans’ innate pessimism and cynicism tend to adversely impact investment decision making despite evidence that optimism always wins long term. If we can overcome bias in the investment decision making process, there are more ways to win than lose. 

✌️-Brad

Disclaimers:

The content of this newsletter is my personal opinion as of the date published and is subject to change without notice and may not reflect the opinion of NZS Capital, LLC.  This newsletter is an informal gathering of topics I’ve recently read and thought about. I will sometimes state things in the newsletter that contradict my own views in order to provoke debate. 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. 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, LLC has no control. In no event will NZS Capital, LLC 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.

SITALWeek #437

Welcome to Stuff I Thought About Last Week, a personal collection of topics on tech, innovation, science, the digital economic transition, the finance industry, and whatever else made me think last week.

Click HERE to SIGN UP for SITALWeek’s Sunday Email.

In today’s post: Walmart's healthcare experiment exit portends bigger problems; a NIMBY backlash is forming against data centers; the surprising places you hear loons; more advice from Kevin Kelly; the middle-class hiring slump; forecasting the weather; and, much more below.

Stuff about Innovation and Technology
Walmart Yields
Walmart announced a surprise exit from their five-year-long effort to disrupt the healthcare industry. Fifty-one clinics and their telehealth centers will cease operation (their pharmacy division will continue). I’ve had a certain fascination with both Walmart’s and Amazon’s attempts to expand into healthcare. As the largest employers in the US (besides the Federal government), they were in a unique position to experiment with different ways to improve the health outcomes of their own employees, and potentially create a better mousetrap for the doomed American healthcare system at large. I covered the opening of the first Walmart healthcare supercenter in 2019 and their expansion into insurance in 2020. And, back in 2021, healthcare was central to Walmart’s ambitions of being a broad “lifestyle app”, despite reports of the company scaling back their goals in the sector. Walmart cited a combination of reimbursement problems and high operating costs that made medical care unprofitable. For a company that competes effectively in the largest, most competitive segments of retail, that’s quite a statement to make. Further, Walmart is more willing to make major, multi-decade investments in new areas compared to the typical company, so this failure seems doubly significant. Beyond a handful of small neo-insurance companies, I am not aware of any large-scale ongoing attempts to disrupt the healthcare leviathan and improve patient outcomes. Given that every unstoppable force is giving up against the immovable object of US healthcare, we may now be reliant upon AI, and technological innovation more broadly, to make inroads into improving patient outcomes, service levels, and costs. On that front, just last week, Google announced a medically tuned version of Gemini that scored even higher than GPT4 on multi-modal queries (multi-modal here refers to multiple forms of input, such as radiographs or pictures of a wound). The paper on the model can be found here, and this string of tweets has the stats and some examples. I might suggest (mostly tongue in cheek) that perhaps Walmart’s exit from healthcare reflects their view that GLP-1s will make everyone healthier (Walmart has reported shoppers on GLP-1s buy less food) and AI doctors will reduce the cost of medical care, thus preempting Walmart’s need to disrupt the industry. Let us hope.
 
Pushing Back on Electrons
Blowback is increasing against the number of large data centers being built to power hungry AI, as these massive installations can strain the grid and make energy more expensive for everyone. We’ve previously discussed this energy issue in Probabilistic Fortune Tellers and Pushing Electrons. Further, we’ve discussed the tension between the insatiable growth of digital technologies and the slowly moving gears of the analog world in When Positive and Negative Feedback Loops Collide. The WaPo reports that efforts are underway in multiple states across the US to slow or curtail data center construction. And, Bloomberg reports that Dominion Energy has received the equivalent of several nuclear power plants worth of energy requests from developers of new data centers in Virginia. Given big tech companies have not shied away from direct energy investments in the past, it's unclear to me why more of them aren't actively pursuing locations and permitting for new nuclear power plants today.

Miscellaneous Stuff
Loon’s Tune
I don’t think I could point out a Common Loon if it bit me in the face, but I do know what it sounds like, thanks to pop music. This article reports on the ubiquitous presence of the Common Loon’s ululating tremolo call and haunting wail in recorded music/media over the last few decades.
 
Just Try Stuff
It’s Kevin Kelly’s birthday, so it’s time for 101 more bits of bite-sized advice – always popular with SITALWeek readers. For those who missed it, Kelly compiled prior years’ advice into a book.

Stuff About Demographics, the Economy, and Investing
Blue and White Hiring Gap
One hypothesis I am tracking closely, as large corporations embrace AI, is the potential for casualties in white-collar information-based jobs. While we may not see outright job destruction, it’s possible that companies will grow with the same number of employees, or hire at a slower rate in the future, thanks to productivity gains. Such trends, of course, are nothing new. Tech has been an accelerating productivity driver over the last fifty years; however, approaching the point where computers can reason as well as humans portends an even larger leap forward in efficiency. As Microsoft trains its AI on all of our mouse clicks and keyboard taps, it’s gearing up to automate all our rote work, and perhaps even some of our more creative tasks. This productivity boost will likely result in fewer people clicking and tapping (per every dollar of new revenue growth) at many companies. As a massive aggregator of retirement investment accounts, Vanguard has a unique vantage point on hiring trends. BI reports that Vanguard is seeing a major slump in hiring for the top third of earners (>$96K annual salary). The hiring rate for this group of workers, who we might refer to as the heart of quintessential middle-class households, peaked in 2022 and is approaching its pandemic lows, well below pre-pandemic levels. Meanwhile, the hiring rate for the bottom third of earners (<$55K annual salary) remains elevated above pre-pandemic levels (it should be noted unemployment still remains near record lows for all salary levels). While correlation with tech-based productivity gains doesn’t prove causation, this type of income-based hiring data is an interesting stat to keep an eye on. 
 
PDS
As I mentioned last week, I’ve been live-tracking YouTube weather chasers as they intercept dangerous storms ripping across North America’s Great Plains. There is an official term used by the National Weather Service to warn the public of storms with the potential for significant damage: PDS, which stands for a “particularly dangerous situation”. While this formal term describes very serious circumstances, I do find it a little funny. It sounds like it’s trying to overachieve. Nonetheless, if you are in a PDS, you need to seek shelter immediately – in the lowest, most interior place possible of the building you are in. Sometimes, as investors, we come across a sector-based PDS (which tends to occur far more often than the deadly F5s that can engulf the entire economy). I am reminded of a moment in time, nearly 25 years ago, when I found myself in the middle of such a particularly dangerous storm. In the fourth quarter of 2000, the sales for communication equipment peaked. This peak took place several quarters after the collapse of the dotcom bubble. The dotcom bubble was built on the hope that the comm equipment infrastructure cycle would build out the Internet, thus opening vast new markets in commerce and media (i.e., create the digital world as we now know it). Back then, there were no smartphones (there was hardly even any cellular data!), and the Internet was largely text-based and operated at slow dial-up speeds. But, there was hope and dreams aplenty, and the comm equipment industry was close to a $200B run rate in annual sales across the supply chain at its peak. In the fourth quarter of 2000, I was busy polishing my shiny models filled with (over)confident predictions about future revenue. (Back then, as a newbie analyst, I labored under the misconception that my economics degree was actually applicable to the real world – what a hoot!). Anyway, I was forecasting a “conservative” deceleration from 200% growth to 80% growth, as a proper analyst should. And, then the storm came. It was a sudden, massive, PDS. I needed to seek shelter immediately. For one well-known, high-flying company at the time, my forecast for revenues just four quarters into the future ended up being ten times too high. Roughly speaking, I forecast growth of 80% to around $500M in revenues. Revenues ended up clocking in at around $50M just twelve short months out. (And, this was not some dotcom vaporware company, it was a real company that had been around for decades selling high tech components into many end markets). I can still vividly recall the awesome disconnect of a one-year prediction being off by an order of magnitude – it was an out of body experience. At the time, had I possessed the vocabulary for it, this would have been my first encounter with the folly of trying to predict the future inside a complex adaptive system. It was, to say the least, an informative event, and it set me on the long path to figuring out a better way to conceptualize and invest in the economy – the upshot of which was Complexity Investing, which Brinton and I published in 2014. Lamentably, I still lack mystical fortune-telling powers; however, because I know what kind of turbulence I can expect from a wildly chaotic system, I am much less prone to being blindsided by economic turmoil and stock market bubbles. Perhaps one day we will create an all-powerful probabilistic AI fortune teller capable of predicting the markets’ ups and downs, but that day has not yet arrived. Circling back to tornadoes, it seems that highly accurate, AI-driven weather forecasting is much closer to reality. Yet, even the best models are still far from infallible. That tornado I linked to at the top of this paragraph was a storm I watched happen live, with the funnel touching down a full 10 minutes before the NWS issued so much as a warning (let alone a PDS). The lesson here I think is two-fold for weathering any type of storm: 1) always have access to a safe space, and 2) make sure you are looking in the right direction; you might think the storm is up ahead, but it could be funneling down on you from above. And, always lean on those lessons from complex adaptive systems.

✌️-Brad

Disclaimers:

The content of this newsletter is my personal opinion as of the date published and is subject to change without notice and may not reflect the opinion of NZS Capital, LLC.  This newsletter is an informal gathering of topics I’ve recently read and thought about. I will sometimes state things in the newsletter that contradict my own views in order to provoke debate. 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. 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, LLC has no control. In no event will NZS Capital, LLC 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.