SITALWeek #398
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: autonomous food delivery is expanding; is the decline in linear TV a far more existential issue for sports teams than people think? Monetizing AI will come from proprietary data and network effects more so than the AI platforms themselves; some cool new pictures of the hot surface of the sun; quantifying the impact of embodied AI; Mrs. Davis; and, much more below.
Stuff about Innovation and Technology
Digital Doppelgänger
A professional Snapchat-picture-poster with 2M followers, Caryn Marjorie, launched CarynAI, an AI chatbot trained on her social media posts by the company Forever Voices. The app, which charges her followers $1/minute to chat, earned $100,000 in the first week, and the 23-year-old optimistically predicted that she will make $5M/month off of the service based on the number of followers that expressed interest.
Autonomous Uber
Uber Eats will be rolling out up to 2,000 four-wheeled sidewalk robots for meal delivery. Serve, the Level 4 Autonomous delivery bot manufacturer, notes there are already 200 such robots delivering food in LA. Venture capital is pouring into the robotics market, especially for humanoid bipedal and quadrupedal forms. Serve has previously raised capital from Nvidia, Figure just raised $70M for their general-purpose bipedal robot, and, thanks to VC infusions, Sanctuary AI recently unveiled its Phoenix humanoid. General-purpose robots with embedded AI could far exceed the impact that AI has in the purely digital realm, but with a much larger array of potential outcomes.
Sporting Headwinds
Sports teams in the US sell their TV broadcast rights to regional sports networks (RSNs), and the lucrative deals can be anywhere from 10% to 25% of a team’s total revenue. As cord-cutting accelerates and interest rates rise, debt-saddled RSNs are filing for bankruptcy and, in some cases, have stopped paying teams the money they’re owed for games. Some of this saga is covered in a recent Puck The Town podcast. Many teams plan to go direct to fans with a streaming app while maintaining free, local over-the-air broadcasts. The issue is that only a fraction of pay-TV subscribers, who have been footing the bill for the RSN fees, would likely consider buying a direct streaming service from each local team. The question is: what fraction would be willing to pony up? By going DTC, teams can also expand their market to fans who are anywhere around the world. But, again, how many people will that be? It’s possible the cost of a local sports app would need to be multiples of the cost of a typical streaming video service like Netflix for the math to work. It seems possible that the high-margin shortfall from canceled RSN contracts won’t be sufficiently offset by new DTC offerings, and teams will potentially be looking at significant losses of both revenue and viewers, which could create a negative spiral given that sports are already facing competition from infinite content of all types. If viewership were to decline broadly, many of the multi-year sports rights deals (including the national rights) would prove to be overpriced, especially given the aforementioned decline in pay-TV subscriptions. Disney recently suggested it will take ESPN direct to consumers, which would mean that ESPN could be debundled for pay-TV subscribers (who would no longer be forced to pay for the expensive channel as part of a basic TV bundle), thus removing a large guaranteed revenue stream for Disney. Indeed, debundling of premium video content has so far proven lose-lose for media companies, who now also face pressures from multiple forces. What would Disney need to charge for an ESPN DTC subscription to balance the loss of lucrative cable deals and still cover the escalating rights fees in the coming years? Again, the cost to fans could be many multiples of a typical streaming service when a large group of Americans are no longer subsidizing sports fans' TV viewing costs. ESPN is a JV with Hearst Communications (Disney owns 80%), and one has to wonder what the value of the business would be if it goes upside down on its sports rights contracts as viewers cut the cord and don’t pay for streaming. The only viable path forward is to rebundle Hollywood and sports content to reduce churn and increase value to consumers (i.e., something beyond a digital recreation of the cable bundle). Given the infinite amount of content available to everyone (which will only amplify further with AI), it’s hard to envision how this can all work out.
Network Effects Enable Value Pricing
Microsoft currently charges around $1500/mo for the Nuance software used by doctors to transcribe patient visits. A newer version leveraging ChatGPT4 instead of human intervention will be closer to only $600/mo. There are also several startups around the $400 price point, according to STAT. As the tools advance, they will become even more indispensable to overworked doctors burdened with paperwork and insurance company requests, and, yet, the 60+% AI-driven price decline is eye-catching. AI should be able to be value priced as it takes over tasks for all types of information workers, i.e., if it replaces a worker who costs $100,000 a year, you should be able to charge a reasonable fraction of that amount and everyone still comes out ahead (except the laid off worker, of course). Yet, perhaps the availability of the ChatGPT platform (and, soon, many other open-source LLMs) is more likely to cause a race to the bottom for pricing of digital human analogs. If that’s the case, then the productivity boom from AI might be truly incredible for companies that leverage it, replacing workers for virtually no cost. However, to become more valuable, an AI tool will need to do more than just act as a Q&A assistant. In the early days of the Internet, once the browser was invented, anyone could have built an ecommerce site, a search engine, or the like. But it was special network effects (e.g., page rank at Google, fast shipping and selection at Amazon, eBay’s trust system, etc.) that allowed some companies to accrete outsized market share. Similarly, the value of AI tools is likely to be less driven by the platform (e.g., OpenAI), and more defined by proprietary data and/or user network effects. For example, STAT also reports on Tempus One, an AI assistant for oncologists that has access to clinical data and can make patient recommendations. And, in Microsoft’s main business of productivity software, the company is charging a 40% premium for early beta users of its AI tools. These tools from Microsoft will soon be able to replicate a large portion of most information workers’ day-to-day job functions (this article on a Minecraft bot using ChatGPT shows how these systems can rapidly learn human tasks). AI-powered robots are also much more likely to be value priced than purely digital human replacements, given the greater complexity of integrating hardware and software and the real-world training data they will gather to improve themselves. There is a bit of an echo here to the early days of the Internet where analog dollars were turned into digital pennies, but we should eventually see platforms able to harness user/data network effects emerge as the winners.
Miscellaneous Stuff
Sunny Hellmouth
The Inouye Solar Telescope recently released some of its first images showing some stunning detail of the surface of the sun, including this mesmerizing image of a decaying sunspot that is surely not a sign of impending doom.
Physical Intelligence
This Lex Fridman podcast interview with the director of the MIT Center for Bits and Atoms, Neil Gershenfeld, is packed with insight on computing, AI, and biology. I knew of Gershenfeld because he stumbled into inventing the airbag seat sensor while working on an apparatus for a magic trick in the 1990s. Given the density of knowledge Gershenfeld has, you have to sometimes pause in order to process what he’s saying, but if you can make it to the last quarter of the podcast, I think you’ll see the payoff. One of his more revelatory conclusions is that the advancements from the current wave of AI innovation are now essentially behind us, and its future impact is somewhat predictable. What he means by that conclusion is that we have reached the point where AI can simulate the human brain; therefore, these new systems will be able to do anything a human can do. Meanwhile, humans will also keep doing things humans can do despite AI subsuming a lot of human tasks. Gershenfeld also explains the far bigger disruption will be when AI is embodied in all sorts of objects down to the molecular level. The three minutes starting at this point are particularly insightful. Gershenfeld estimates that embodied human intelligence is eight orders of magnitude more powerful than a human brain on its own. I believe this means we will see far more emergent, unpredictable behaviors from embodied AI than AI running on servers.
Mrs. Davis’ Purgatory
The show Mrs. Davis (streaming on Peacock) follows a modern-day nun on a quest to destroy the Holy Grail. She was sent on this quest by the AI chatbot named Mrs. Davis, which has the world under its sway. The show is a bit hard to describe, but I’d say it feels like a combination of the TV series Lost (Damon Lindelof is the writer of both shows) and the movie Her, with a dash of Indiana Jones and a sprinkle of The King James Bible. The main character’s parents are professional magicians, and magic features heavily throughout the show (which long-time readers know is a significant lure for me). In reading various reviews, I can see that people are interpreting the show in many different, personal ways. I have my own interpretation, which I’ll touch on while trying not to give anything away. The show left me feeling like there is no escaping a sense that humanity’s relationship with AI is a sort of purgatory. We are trapped in the amber of a major event with both enormous potential for advancement and extreme risk of losing what makes us special. Quite literally, there’s a path to heaven or hell for us to choose. I assume at this point everyone who wants to see Lost has seen it, if not, spoiler ahead...in that series, which wrapped a little over a decade ago, Lindelof also explored a type of purgatory, which he sourced from the Tibetan Book of the Dead: “It’s a place that you go when you die but you don’t know that you’re dead. It is like Bruce Willis in The Sixth Sense. He doesn’t know that he’s dead, and the entire purpose of being in this space is to come to the revelation that you have died, but no one’s allowed to tell you.”
✌️-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.