SITALWeek

Stuff I Thought About Last Week Newsletter

SITALWeek #408

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: Fed policy is driving inflation higher in key parts of the economy, contradicting their goals; the potential for AI to solve the "big data" failings; the importance of workplace trust for driving innovation; Unreal Keanu; college financial pressures; a restaurant tipping point; and, much more below.

Stuff about Innovation and Technology
Escaping the Data Swamp
Last week, I touched on the failed promise of the “big data” era of IT spending. While most IT spending trends represent new technologies/functionalities meeting a demand in the market, the $5T industry is also susceptible to fads and other impractical forces. As it’s often said: software is sold, not bought. The rare piece of software may be so compelling that it sells itself, but everything else is largely sold to you. You can see some of the phases by surveying the various epochs of enterprise software starting with mainframe and then the client/server era, which ushered in a wave of corporate data centers filled with racks of servers, storage, networking, and security appliances. The client/server era, which ratified the corporate IT department as a key function in companies, powered productivity tools, like email, shared drives, intranets, and the beasts of all enterprise software: Enterprise Resource Planning (ERP; now mostly consolidated by Oracle and SAP) and the database market (dominated by Oracle and Microsoft in the 1990s and 2000s). Who could forget the dotcom boom, where every company built a website and was going to dominate digitally? That was perhaps one of the “shiny object” fads in IT spending. Then came SaaS applications, and, eventually, companies migrated their data centers to cloud infrastructure providers like AWS and Azure. Open-source software was another trend that grew up in the client/server era and then exploded with the cloud. Other trends in enterprise IT have included the shift from Ethernet to Wifi, PCs to laptops, office to remote working, etc. The “consumerization of IT” was a big theme when the iPhone landed – everything was going to be as simple as a tap on the phone. 

All the while, data accumulated by the petabyte, eventually birthing the era of “big data” and analytics in the 2010s. The bright promise was alluring: unlock the wealth of knowledge hidden in your organization, make better and faster decisions, get ahead of the competition, make your customers and employees happier, and so on. However, rather than being pristine reservoirs of knowledge, the reality of big data projects was closer to inaccessible Florida swampland. One of the keener examples was GE’s Predix Industrial IoT cloud platform, which promised corporations access to data from a network of zillions of connected sensors (it was largely viewed externally as a failure, but I see that GE still markets the platform). Often, as IT priorities shift to the next shiny object, the previous areas of spending don’t go away, they just become a smaller piece of the overall pie. I suspect that’s where we are now with SaaS and cloud migrations: while these mega platforms will be with us for as long as we can imagine, they are a lower priority that will, in aggregate, grow slower or possibly shrink for traditional workloads and especially “big data” projects. One of the reasons for the cloud slowdown is the arrival of the shiniest object to date: AI. Right now, there is a wild, global hoarding bubble for GPUs to theoretically train large language models. This will predictably burst in the most spectacular fashion in the next few years as many of those GPUs lay dormant thanks to the overbuild (e.g., as with EDFAs in the optical boom/bust) and AI efficiency gains.

I discussed some of these IT spending waves when I wrote AI is the New Dotcom, and That’s OK nearly two years ago: 
Back in the late 1990s, every business was either appending “.com” to their existing name or touting their dotcom strategy and how it was going to transform them or their industry. A lot of hyped-up ideas ended up being right, just twenty years too early. But, for the many legacy companies that put on dotcom lipstick at the turn of the century, the Internet was ultimately a negative disruption of their business. For some industries, such as media and retail, we’ve seen the near completion of the disruptive, Internet-enabled transformation. For more highly regulated businesses, such as the banking and healthcare sectors, which have successfully lobbied to keep disruption at bay, it’s unknown if/how they will ultimately be affected by the Internet Age. And, for a large bucket of companies that have harnessed the Internet to improve their products, supply chain, and/or customer interactions without significant disruption to their business model, dotcomization has been more subtle. For all industries, the Internet enabled an accelerated pace of change, and dotcom simply became shorthand for digital transformation. The biggest winners of the Information Age have been the new companies – those that were built by the Internet, for the Internet, in the late 1990s and early 2000s.
Further down in that post, I expressed skepticism about the reality and timeframe for AI to arrive, but, just a few months later, in early 2022, I changed my views completely as chatbots and transformer models began to emerge. While the timeframe may have accelerated, the sentiment from that post still holds: AI represents yet another incremental step in the long arc of IT, which we can think of as the digitalization of the analog economy. We are still early in this process of digital transformation, but AI is likely to become the biggest accelerant to date (by several orders of magnitude). Much like the touchscreen, AI represents an entirely new user interfaceIn conversing with our new AI chatbots, what we are actually doing is having a conversation with data, that heretofore impenetrable swamp of accumulated 1s and 0s. And, as I noted about the dotcom era, while every company will initially adopt AI to help their business, it will ultimately threaten most of them by enabling new competitors. While I expect early adoption to go much faster than for the Internet/cloud (since the infrastructure and data are already in place to enable AI), this new era of digitalization will still be plagued by frustrating fits and starts. 

The above rambling digression stemmed from an article I read in the WSJ about how farmers were struggling with “agtech”, swimming in a sea of data without practical ways to implement it. The article notes massive gains for some farmers that do adopt new technologies (e.g., increasing winter wheat yields by 49% by leveraging digital soil maps). But, by and large, the solutions remain complex. Here is where LLMs might save the day: rather than leaving the decisions up to the humans, LLMs can act on their own, especially when embodied in field robots (precision weed sprayers, harvesters, tractors, etc.) capable of collecting their own data. For example, Solinftec, which makes autonomous farming robots that use AI to precision apply herbicide and pesticide (resulting in a 95% reduction in usage), predicts their robot deployment will go from 20 to 250 among US corn farmers by 2025. 

Autonomous AI will eventually create a “do it for me” virtuous circle, which, of course, will come with its own perils, monitoring requirements, and withering human meaning. Today, I think most companies in most industries are sitting in the big data pit of despair, but, with the right amount of adaptability, some companies will leverage the next phase of AI-driven IT spending to their benefit. However, as with the dotcom boom, most enterprises will fail to catch the next digital wave (e.g., recall all the major retail chains that launched a website, only to go bankrupt as Amazon and others gobbled up their customers; or, consider the Hollywood Studios that launched streaming services, only to lose viewing time to YouTube, TikTok, video gaming, etc.). The best way to prepare for AI is to 1) make sure your organization is collecting every bit of data possible, and 2) develop and refine chatbot interfaces so you can begin conversing with it (e.g., Ethan Mollick’s “Now is the Time for Grimoires” is a good guide).

Virtual Circle of Trust?
One of the reasons the founder of Zoom cited for the company’s new partial return-to-office mandate (yep, you read that correctly) is that people are too friendly over Zoom meetings. And, this overfriendliness stifles debate, which stifles innovation. I was initially baffled by this finding because my assumption would be that the physical insulation provided by remote communication would tend to make people less empathetic/friendly. That people are more likely to debate in person vs. remotely certainly contradicts the entire cesspool of social media “discourse”. The other reason cited was that it’s harder to build trust when fully remote, which makes more sense to me. Brinton connected the dots here: in order to constructively argue with someone, you have to have built trust to begin with, otherwise criticism/questions are easily construed as a personal attack. The importance of psychological safety and trust is covered in Brinton’s excellent whitepaper on how companies can slow down time. Even though this logic makes sense, the hypocrisy of Zoom calling employees back to the office feels like a real failure to innovate. Can’t trust be built between two people who never meet in person via new features and technologies like spatial computing?

Miscellaneous Stuff
Unreal Keanu Reeves 
This deep-fake, short-form video account has over 9M followers on TikTok (and over 1M subscribers on YouTube). We’ve looked at similar face-swapping AI in the past, but this is a good example of how real the unreal feels, and also it’s just funny to see the normally understated actor in a series of social media tropes. 

Collegiate Resource Drain
Over the last decade, the number of adults that said college is “not worth it” has risen from 40% to 56%. Mirroring that sentiment, overall college attendance has dropped 15% from 2010 to 2021. Bloomberg also notes that wages for college-educated workers have risen more slowly than for non-college-educated workers for the last 30 months. College apathy is something we’ve pondered in the past in Giving Up on the Old College Try. In that piece, I wondered if there was a connection between an apparent loss of hope for the future and the diminishing role of humans in an increasingly automated world. Regardless of the causes (which are also heavily demographic in nature rather than just philosophical), colleges are facing increasing expenses and decreasing revenues. One example is West Virginia University, which is shutting down entire departments (who needs to learn languages when you have Google Translate!?). If the trend continues, I suspect endowments will be increasingly focused on supporting operating expenses for schools rather than growing or maintaining their capital base. If, in the extreme, college budgets come under more pressure and endowments are called on to sell more assets, the recent focus on allocating to more illiquid assets could pose an issue.

Less Give, More Take(out)?
As Americans have become increasingly stingy with tips, restaurants in some areas are under pressure to do away with the practice of paying below minimum wage (which they can do assuming tips will make up the rest). In Chicago, servers are paid a minimum of $9.48/hr, but they would make $15.80/hr with the changes. If paying servers minimum wage becomes widespread (it’s been implemented in places like LA since the 1970s), the cost transfer to customers would likely negatively impact sit-down dining, which could lead to even more emphasis on pickup and delivery.

Stuff About Demographics, the Economy, and Investing
Solve Inflation with Lower Rates
Fed Chair Jerome Powell, speaking last week, noted that rents are slow to move, but are showing signs of falling:
Measured housing services inflation lagged these changes, as is typical, but has recently begun to fall. This inflation metric reflects rents paid by all tenants, as well as estimates of the equivalent rents that could be earned from homes that are owner occupied. Because leases turn over slowly, it takes time for a decline in market rent growth to work its way into the overall inflation measure. The market rent slowdown has only recently begun to show through to that measure. The slowing growth in rents for new leases over roughly the past year can be thought of as “in the pipeline” and will affect measured housing services inflation over the coming year. Going forward, if market rent growth settles near pre-pandemic levels, housing services inflation should decline toward its pre-pandemic level as well.
This slowdown in rent inflation appeared to be the case a while ago (I noted the drop in rental demand and increase in supply last Fall here). However, since the Fed tends to live in the past when it comes to data, Powell is now missing the fact that rents are on the rise. Why are rental rates going up? Chiefly because the Fed’s supposed inflation-fighting policies are causing rent inflation! As John Burns Research explains here on X, mortgage rates keep ticking higher, faster than Fed rate increases, because the Fed is no longer purchasing mortgage-backed securities. So, higher rates and reduced Fed mortgage purchasing are making homes particularly unaffordable, which means more people are staying in their rentals longer, which causes rents to increase. I’ve previously written about how rental rates influence Fed policy when discussing the algorithmic manipulation of rents during the pandemic, as well as the rearview-mirror problems at the Fed. Further, I’ve noted how, given the massive amount of leverage still in the system, higher rates are causing inflation as companies pass on higher interest expenses to customers. We appear to be in an ouroboros moment for the Fed, whereby their data delay is blinding them to the fact that their inflation-fighting policies are causing inflation. But, don’t hold your breath for the Fed to realize that, instead of catching the inflation bogeyman, they've sunk their fangs into their own tail end.

✌️-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.

jason slingerlend