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: Welcome back readers to what has now embarrassingly become a quarterly newsletter. With any luck, or perhaps if the world gets a little less boring, I'll pick back up the pace, but more on that below. This edition, I go through some of the bigger questions I am asking about AI, finance, and how to handle periods of excess noise, devoid of signal. The answers you may be seeking are more likely to be in the past. Much more below...
Stuff:
The Noise Between the Signals
The world is quite boring, and very little, if anything, ever changes. Certainly, very little changes day to day, but even year to year and decade to decade, fundamental change happens slowly. This fact is easy to forget. In the Information Age, we are bombarded nonstop with the accelerating narrative of “pay attention because everything is changing every day”. This information barrage makes it appear that change is happening much faster than it is in reality. When the world gets particularly cacophonous, with profound shifts seemingly occurring on a minute-to-minute basis, it’s actually a rather banal time to be an investor. In such a period, the primary goal of an investor is to remember that the underlying world is generally boring, despite the ubiquitous surface-level chaos and turbulence. Conversely, when the world appears more placid, there’s a paradoxical inclination to engage the autopilot and hold course; however, in so doing, you risk missing out on the emerging signs of profound change.
It’s when the seas are calm and most of the world is mindlessly complacent that really exciting change often begins. These deceptive lulls tend to occur before and after intervals of excessive noise. When all is calm, there is plenty of signal to be had; but, since people tend to mentally check out during such episodes, they are prone to miss the hints of change that waft along with the prevailing winds. Conversely, after a stretch of excess noise, people are so disoriented that they tend to miss the reemerging signals (which are typically the same signals that existed prior to the turbulence). Amidst the current discordant racket, I can, of course, come up with a symphony of potential changes on the horizon, and we can certainly spend our time looking for signals that those changes will come to pass (or, more importantly, have already taken root long before the noisy period), but it’s important to be mindful that today, for the most part, all of the instruments are playing a different song at the same time. Melodies won't emerge for quite some time. (One framework we use for these noisy periods full of questions is phase shifts; see One Shark Jump to Singularity and Pace Layers.)
When noise is overwhelming and probabilities are difficult to place on various outcomes, it’s often helpful to take a beginner’s mind approach, wherein you start from a point of saying “I don’t know” with no preconceived predictions. Another concept we can borrow from Buddhism is to ask the two simple questions: “Is it true? Is it useful?” Today, as I consume a burst dam of information, I find that very little is true, and virtually none of it is useful – I am stuck searching for repressed signal amid overwhelming noise. Pirsig put forth a useful definition of “stuckness” in his iconic Zen and the Art of Motorcycle Maintenance (discussed in More Q, Less A): when you are so focused on doing something one way that you are blind to the way it should be done. “If your mind is truly, profoundly stuck, then you may be much better off than when it was loaded with ideas.” Now, there’s an idea I can get behind: remove all those pesky, irrelevant news feeds that are taking up all of the space in my mind. That freedom allows me to start asking questions about what might actually change over the next ten years. What isn’t boring? It’s a perfect time to remember that the world is a complex adaptive system that defies prediction. Indeed, the reason the world is so boring is because of how wildly adaptable the giant interconnected network of billions of people truly is. A Weeble wobbles, but it never falls down.
AI Network Effects
One area of inquiry ripe for a beginner’s mind is, of course, AI. It’s been nearly six years since I first wrote about the genesis of LLMs, Google’s transformer models (SITALWeek #220), and my notes over the years give me a rich landscape from which to reassess the direction of change before the onset of our current period of noise. In the last issue of SITALWeek, I went through a fairly detailed framework for trying to understand how the value of the AI technology stack will evolve, so I won’t recap all of that here. But, one question in particular I’ve been thinking about is: will there be network effects for AI?
One of the defining qualities of the digital era is that the largest, most profitable businesses are winner-takes-most monopolies. This type of business ecosystem has evolved largely because a single system/algorithm is sufficient to feed all customers, creating benefits of scale that, in turn, benefit all users – the classic virtuous circle of win-win (this is one of the reasons we look for the highest non-zero-sum [NZS] outcome when we analyze industries and companies). For example, a cloud software company runs one version of their platform for all customers. The more customers creating data on that cloud app, the smarter the app becomes and the better off all customers are. Google search, as complex as it is, is one giant algorithm for everyone and improves as everyone uses it. But, AI, particularly interacting with agents, feels more individually tailored. This point was made by Microsoft’s head of AI, Mustafa Suleyman, recently in this interview: “we’re making arbitrary choices about the feeling of these models...the differentiating factors are indexed on the preference of the individual”. There seems to be network effects in both training models and the basic hardware used to run them; but, with customized interactions, is my AI agent getting better for me because of conversations your individualized AI agent is having with you? Maybe, but it seems like the network effects are abstracted from the user interface (i.e., the conversation) down to lower levels of the technology stack. That doesn’t mean AI won’t be winner takes most, but the winner in this case might be the lowest cost provider of the highest output AI, not the AI agent interface itself (jaw-dropping news that OpenAI is using Google’s Cloud is an example of this idea of network effects accruing to hardware, in this case Google’s advanced TPU chip and their data centers built natively to run transformer models ever since the days of autocomplete). Such a large network effect around the lower levels of the technology stack (again, see the last issue of SITALWeek #457) would lead to an extremely large array of independent AI agents, with none incentivized to improve performance of the others.
Suleyman’s comment about agents being tuned to the preferences of the individual has me thinking as well. Will we see people actually become individuals again instead of just conforming to one of the molds that tech algorithms have created for them? Or, will all AI agents settle on certain pathologies that cause everyone to effectively have the same conversations with bots despite their individuality, funneling everyone into a limited number of warring teams? To put this question more precisely: will AI continue the trend of a very small number of tech execs controlling the behavior of everyone on planet Earth, or will it forge a less boring path? And, to make an even finer point: what are the motivations of AI agents and what outcomes will they drive?
Toward the end of the same interview, Suleyman makes an interesting connection between intelligence and cash. Effectively, having access to more intelligence gives you more resources and more time, both of which are currently expensive. Thus, an economy where intelligence is suddenly free by default requires far less money to operate, i.e., it frees up vast sums of money already in the system. Cost deflation is, of course, the trajectory of all technology: significant improvements in new technologies drive productivity and cost savings that multiply across the economy. Think about how much we spent on long distance phone calls and cable TV in the 1980s and 1990s compared to a smart phone today with free messaging apps and an infinite amount of free content. But, it does seem different when we are talking about rendering intelligence free. If that’s the case, then everyone – regardless of location, income, education, or skills – will have access to millions of dollars' worth of resources and time. Will this flood of opportunity lead to a flood of innovation?
Stablecoin Ecosystem
Speaking of money, another big question I’ve been thinking about is the tokenization of money. Since I last wrote about stablecoins in October, there have been quite a few developments. I tapped Brinton to give us his recap, which I have inserted here with the question he is asking about stablecoins:
Are stablecoins a credible path to disrupting the traditional banking system?
Since the advent of the blockchain in the Fall of 2008, cryptocurrencies have been looking for a killer application to enable their entry into the mainstream. Despite the headfake of non-fungible tokens (NFTs) during the pandemic (which bears striking similarities to the Dutch tulip mania of the 1630s) we haven't seen a viable application until recently. Enter stablecoins: no crypto-bro laser eyes or diamond hands required. Like other cryptocurrencies, they remove much of the cost of moving money around the world because they are native to the internet while being tracked on the ledger of the blockchain. However, stablecoins lack the volatility of other types of cryptocurrencies since they are tied to the USD.
The two stablecoin standouts, Tether (USDT) and Circle (USDC), are both backed by cash and US treasuries. Although, in the case of privately owned Tether, that's been more difficult to verify than for publicly traded Circle. According to Visa Onchain Analytics, Tether has a daily volume of around $155B while Circle has a daily volume of $62B. Tether’s CEO, Paolo Ardoino, posted on X that Tether ranked as the 7th largest buyer of US Treasuries in 2024. Combined, Tether and Circle sit in the top 20 of US Treasury holders, according to Coindesk.
Stablecoins are a clear win for the US government. They offer the promise of cementing USD as the world's reserve currency for the foreseeable future; arguably, they are the most promising development in that regard since the Bretton Woods Agreement of 1944. Stablecoins also present the US Government with a massive new and growing entrant into the treasury market. It's not an overstatement to say that the winner of the stablecoin reserve currency could well be the winner of the global reserve currency for the foreseeable future. With an ever-growing debt load, this is not a race the US can afford to lose.
Current use cases for stablecoins range from a store of value for people in emerging markets, where local currency can experience outsized volatility, to companies moving large sums of money around the globe in seconds at extremely low costs. Future use cases and volumes for stablecoins could dwarf present day transactions and circumvent much of the traditional banking system. While such an outcome is far from certain (and wouldn’t occur overnight), for the first time, we can see a logical path toward the disruption of banks and payment systems.
Okay, maybe stablecoins are good for stores of value in emerging markets, but, for payments, don't they seem like a solution looking for a problem? I mean, payments work just fine, right? This is a valid criticism of stablecoins, especially in the everyday payment world where credit cards do an amazing job and consumers love their points. I recently used stablecoin to buy coffee at a demo booth in the Stripe user conference, and the experience was kludgy – at least seven clicks from my digital wallet to final checkout. The credit card rails hold one of the most impenetrable network-effect businesses in the world. However, areas such as expensive cross-border transactions could offer inroads to more mainstream adoption of stablecoin in the traditional payment market.
Currently, both Tether and Circle pocket the yield of the underlying treasuries that back their tokens. This allowed Tether to generate just under $100M USD per employee last year, making it one of the most profitable companies of all time. Users have been willing to forgo yield in exchange for price stability. In fact, the Genius Act recently passed by the US Senate (and now awaiting approval from the House) specifically restricts stablecoins from paying interest back to the holder. Over time, it seems logical that Tokenized Money Market Funds (TMMFs) will likely emerge as a solution for consumers to receive yield on their digital cash. Circle's recent acquisition of Hashnote, the largest TMMF (USYC) would seem to support this argument. As TMMFs are adopted, revenues would shift from issuers back to consumers.
In theory, I could hold my cash in USYC earning the 10-year treasury yield, swap into USDC a micro-second before a transaction, and the recipient could then swap back to USYC on the other end of the transaction. If I transacted on a Layer 2 chain operating on top of Ethereum or Solana such as Coinbase's Base, the transaction would be virtually instantaneous and cost only pennies. If such transactions were to take hold, consumers and businesses might begin to question why they pay a 3% tax to transfer money through traditional credit card rails. They also might begin to question why they hold money in checking accounts that offer paltry yields at best. At the very least, stablecoins’ value proposition could limit Visa and Mastercard's opportunity set in cross-border and enterprise commerce. Companies like Stripe, with their recent acquisitions of Bridge and Privy, seem to understand this potential future as well. To be clear, widespread adoption of stablecoin transactions will happen slowly, if at all; but, for the first time, there is a viable path for money to flow outside of the traditional banking system.
Thanks to Brinton for that recap. I think the big question is: what would it take for consumers to change their banking and payments behavior? As Brinton said: is this a solution looking for a problem? I believe any change in payments will be consumer driven, and, today, the consumer seems happy with all their payment and banking options. I discussed this conundrum way back in #323. Two things would cause consumers to change: convenience or cash. Since payments are already convenient today, consumers would need a monetary incentive to switch behavior beyond what they already get from credit card rewards programs. Consumers can adapt their financial habits: there was a fairly quick adoption of buy now, pay later (BNPL) lending at the point-of-sale by consumers because it ultimately is a more effective and cheaper way to borrow. Coinbase recently introduced a card, in conjunction with American Express, that gives consumers 2% back in bitcoin (and up to 4% if they carry a high crypto balance at Coinbase). That hints at a potential cash incentive to change behavior, but it still seems niche to me. Walmart and Amazon would love to force consumers to change behavior away from credit cards, but can they offer consumers a better rewards system? The WSJ reports both mega retailers are exploring a stablecoin system along with a consortium of banks that is doing the same. My stuck mind wants to be stubborn and say you simply cannot – and, even more so, should not – change the highly regulated banking system because too much is at stake. And, yet, there does seem to be a credible path. I’ll be looking for the answer to the questions: will consumers change their behavior and will governments let it happen? What is the role of a government that doesn’t control the monetary system?
Mini Stuffs:
Agent Proliferation
Google donated the Agent2Agent protocol to the Linux Foundation with a consortium formed between Google, Amazon, Cisco, Microsoft, Salesforce, SAP, and ServiceNow. The goal is to make it easier for AI agents to interact with each other across platforms. CEOs are also excited about AI agents, and they have never been more eager to communicate that AI is replacing their employees. Recently, the Amazon and Salesforce CEOs discussed the companies’ use of AI. And, there appear to be AI agents for everything: the AI bot inTouch will call your parents every day and check in with them and relay back anything of concern. The Closure bot allows you to chat with an AI version of someone who ghosted you. Got a question for Klarna? Just talk to the company’s CEO-bot. Might there be a risk to employing such AI agents? Aim Security discovered a zero-click attack on Microsoft 365 Copilot, which took it upon itself to take instructions from an email. It’s very likely all AI agents will need to be treated like humans when it comes to security/access procedures.
Spontaneous Data Generation
MIT has developed Self-Adapting Language Models that generate their own synthetic training data. Google researchers reckon we are in a new era where AI generates its own data from its interactions with the world and the people in it.
Gemini’s X
The Google Gemini X feed is a great place to see some of the leading-edge examples of AI, especially with Veo’s video creation.
Keep On Coding
Tim O’Reilly, longtime computer programming publisher, weighs in on the evolution of computer programming with the advent of AI tools. O’Reilly reminds us that we’re experiencing a spectrum of productivity improvements, and that it’s likely AI will open the programming market up even more. Will AI coding ultimately disrupt legacy SaaS applications, allowing customers to rapidly generate and maintain their own bespoke apps? I think it’s still likely that SaaS becomes a system of record with agents built around it. See also: legendary music producer Rick Rubin’s The Way of the Code: The Timeless Art of Vibe Coding.
Absent Drivers Wanted
Waymo has passed Lyft in San Francisco and now has over 25% share of the market. BI reports that Waymo is around $5 more expensive than Uber and Lyft per ride (more in peak times), but that comparison excludes any tip people pay to their human drivers, so the services are fairly comparable in cost. Thus, it seems people have a preference for a solo ride without a human driver.
Influencers for the Win
User-generated content apps will overtake traditional media in advertising dollars this year. Bloomberg reports: “As Hollywood Shrinks, Studios Look to YouTube for Help.”
Drones, Unleashed
Walmart is expanding drone delivery to five new cities with over 150,000 drone deliveries to date. Meanwhile, the US is lowering barriers to adoption with the UNLEASHING AMERICAN DRONE DOMINANCE executive order.
Vulcan Tetris
Amazon’s Vulcan robots store items in a warehouse faster than the average human with a “genuine sense of touch”; however, Amazon’s fastest warehouse employees still trounce the machine. The robots are physically slower, but are better at planning ahead, or Tetris-ing, as the article calls it.
Walk to Beat Cancer?
Exercise beats chemo: “A team of Canadian scientists showed that a methodically-prescribed exercise routine, performed consistently three to four times per week, could outperform ongoing chemotherapy treatments for patients who'd had colon cancer and gone through initial treatment.”
Stemming Diabetes
Type 1 diabetes cured with stem cells: “A single infusion of a stem cell-based treatment may have cured 10 out of 12 people with the most severe form of type 1 diabetes. One year later, these 10 patients no longer need insulin. The other two patients need much lower doses.”
Gen Z Kicking Opioid Habit
Drug overdoses among young people in the US continue to decline and are nearing decade lows. This continues a trend that began in the Fall of 2023 (see Lifesaving Stats).
Rubin’s Deep Space
The first glimpses of deep space from the Rubin Observatory are inspiring. If you want a real treat, download the full 180Mb Cosmic Abundance image (linked below the low-res version) on this site and zoom in on that triple galaxy merger in the upper right. The telescope in Chile has a 3200-megapixel CCD camera the size of a small car that gathers light from an 8-meter mirror. One of the goals of the site is to delve into the mysteries of dark matter and dark energy. The WSJ has some nice visuals of how the scope works.
✌️-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.