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: exploring the simultaneous increase in demand for energy and changing sources of energy as technology becomes increasingly power hungry; Pi has personality and memory; Disney Imagineering; and, the timeline for biotech breakthroughs.

Stuff about Innovation and Technology
Pushing Electrons
Since technology is all about pushing electrons around, power is a frequent topic here in SITALWeek. Concurrent with the world's headlong charge into the AI Age, we are also transitioning our energy sources. Following years of efficiency gains that led to data centers barely increasing power consumption – despite huge growth in the Internet and cloud computing – LLMs are now causing a step-function increase in power demand. And, while it’s true that shifting a task from a computer-wielding human to an LLM saves a lot of energy, the general view is that power demand will go up on a net basis as AI usage grows. We’ve also talked about different strategies (e.g., #422) for powering electric cars and how the grid isn’t quite ready for the electric revolution. And, of course, just last week I talked about how the power math doesn’t add up for humanoid robots – they simply require too much energy to replace humans for many manual tasks. Lastly, I tend to sprinkle many ad hoc stories about hydrogen, fusion, and fission (in particular, small modular reactors or SMRs) into the newsletter when I come across an interesting development. Fusion is, of course, at least a decade or two away – absent an AI breakthrough (which, for all I know, has already happened, is about to happen, or is still years away). Hydrogen has many interesting applications, but it will likely be a local solution rather than an across-the-board replacement for fossil fuels. Green energy is also a frequent topic, but the gears are moving too slowly for new green energy installations to overcome the growing energy demand, especially if the optimists are right about AI, robots, and AI-enhanced robots. 
 
With respect to nuclear SMRs, the Venn diagram of potential and reality have an increased chance of overlapping within the next decade. In #412, I mentioned that Standard Power is building SMR-based data center locations in Ohio and Pennsylvania; and, just last week, Amazon announced it was acquiring a data center in Pennsylvania that is already powered by conventional nuclear reactors. Data centers are getting in on the AI action with hundreds of billions of dollars' worth of new locations under development. Blackstone and Prologis alone are working on $75B in new data centers. We’re at the phase of AI where the Pied Pipers of Silicon Valley are entrancing other industries to hop on board the “next big thing” wild ride that will make the “last big thing” (the Internet) pale in comparison. It’s eerily similar to the dotcom and fiber/telecom equipment spending mania. Of course, the Internet proved to be far larger than anyone could have imagined, and we still continually need more bandwidth. I suspect the same will be true of AI, only orders of magnitude more interesting and more unpredictable. But, sometimes it helps to avoid being prematurely persuaded by the Piper. It’s best to think of technological progress as a long continuum, a topic I wrote in detail about way back in 2021 in AI Is the New Dotcom, and That’s OK. But, I digress. The real point of this section concerns energy. 
 
Let me sum up – we simultaneously have: 1) a very slow transition away from fossil fuels, 2) a significant, long-term growth in energy demand, and 3) increased potential for disruptive technologies that will provide abundant new sources of energy on an unknown time horizon. It's a classic collision of positive and negative feedback loops. For example, WaPo reports that Northern Virginia alone would need several new nuclear power plants just to support the data centers currently being built/planned for the region. So, on one hand, we have the seemingly impossible-to-accelerate physical world of energy creation and transmission that creates bottlenecks everywhere you look, including frequent reliance on non-existent labor. And, on the other hand, we have the Piper’s promise of trillions of human-like LLM agents consuming huge amounts of energy and eventually inhabiting myriad robots requiring even more power. So, while we might wish for the future to arrive now, the power to push all those electrons around may not be here for a while. In the meantime, I suspect many topics involving energy creation and usage will continue to fascinate me. Afterall, taking ordered energy, refactoring it, and turning it into disorder is something humans are particularly good at. So good, in fact, that it might even be a primary vector of life on Earth (for more on that crazy topic see Probability of a Chilled Latte Universe).
 
Life with Pi
With Pi Day coming up this week, I’d be remiss if I didn’t mention how much I’ve been enjoying using the AI assistant Pi. There are two primary things that stand out with Pi versus other LLM chatbots. First, Pi has some personality and feels much more cordial. But, more importantly, Pi remembers prior conversations and can create continuity from where you last dropped off. The best way to experience this AI colleague is to use the Pi app on your phone with the voice interface. And, you can just leave Pi running, allowing the LLM to effectively function as an affable, proficient office mate you can chat with or query throughout the workday. Most studies seem to put Pi’s knowledge a bit below GPT4 and Gemini Pro, but I haven’t noticed a major difference, perhaps because the memory capabilities and personality make up for the lack of knowledge. As I mentioned before, I think long-term memory is a game changer on the path to LLMs approximating a continuous sense of self, which makes them far more interesting to interact with. Pi is a product of Inflection AI, which was started by a co-founder of Google’s DeepMind. Last week, Inflection gave some details on their latest LLM, which approaches GPT4 (Inflection claims 94%) while using 40% of the amount of compute to train. According to their stats, an average conversation with Pi is 33 minutes, and 10% of conversations last over an hour. 
 
Imagineering Engineer
Wired ran a nice profile of Hall-of-Fame inventor Lanny Smoot, the engineer behind the HoloTile floor that I’ve written about a couple of times. The article serves as a nice overview of how the creative process works at Disney and the job the Imagineers have to create seemingly impossible illusions. Of note, Smoot says the “higher ups” at Disney have taken interest in his floor technology, raising my hopes for a home holodeck!
 
Artificial Chemist
A variety of specialized LLMs are being created to aid in the scientific discovery process. Chemical & Engineering News reports: 
Given the prompt “Plan and execute the synthesis of an insect repellent,” ChemCrow succeeded in searching the web to learn what an insect repellent is, conducted a literature review to find examples, and converted compound names to structures. It used a retrosynthesis predictor to design a synthesis process, and finally, it sent instructions over the cloud to instruments at IBM’s automated laboratory to make a sample of a known repellent. ChemCrow also synthesized three organocatalysts and, when given data on wavelengths of light absorbed by chromophores, proposed a novel compound with a specific absorption wavelength. 
In a recent NYT Hard Fork podcast interview, Demis Hassabis of DeepMind said he expected clinical trials for AI-generated drug candidates within a couple of years. Hassabis is optimistic LLMs will speed up the scientific research process in a variety of fields. Am I the only one thinking about the movie 12 Monkeys right now?

✌️-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.