Welcome to this week’s edition of Cape May Wealth Weekly. If you’re new here, subscribe to ensure you receive my next piece in your inbox. If you want to read more of my posts, check out my archive.

Ever wrote a blog post that crashed the stock market?

If you’re reading any financial publication on a regular basis, chances are that The 2028 Global Intelligence Crisis (or an article mentioning it) made it across your desk. In the article, Citrini Research outlines a scenario in which the AI bull case of a rapid increase in agent capabilities turns out to be true. But rather than bringing the economy to new heights, it turns out to have fatal consequences on the wider economy. The article rattled the market, with some companies mentioned in the market (such as DoorDash, American Express, or Uber) falling as much as 10%.

More than one client sent the article to us, asking us about our view on the potential impact of AI, and of course, the subsequent impact on portfolios. But even before the article, few things have been more ‘top of mind’ than AI agents. One client shared how he was planning to hire ten developers for his new company, but made it work with two AI-savvy ones and a number of agents. And at an investor lunch that we hosted last Friday in Stuttgart, more than one family officer - who are typically not the most tech-savvy individuals, even if they invest in tech - lauded the substantial productivity gains from the latest version of Anthropic’s Claude.

So in today’s newsletter, let’s talk about the potential impact of AI on investor portfolios. First, we will take a look at the Citrini article, sharing our thoughts on what we think is realistic, and where we disagree. Secondly, we will think about how investors should think about AI impact to their portfolios, and how they should think about risk and asset class positioning.

Let’s dive in!

If you’ve already read the Citrini article, you can skip from here…

Let’s begin by taking a look at the original article. If you’ve already read it, or as one client said, ‘had AI summarize it for you’, feel free to skip to the end of this section. I’ve still proudly summarized this for you manually - no AI involved.

The Citrini article is not a research report, but rather a hypothetical Macro Memo from the future (June 2028), detailing the progression and fallout of the Global Intelligence Crisis. It is broken down into various sections, beginning with a hypothetical job numbers print from June 2028, before looking back on how this hypothetical scenario came out to be, and what more to expect.

How It Started

When Friction Went to Zero

From Sector Risk to Systemic Risk

The Intelligence Displacement Spiral

The Daisy Chain of Correlated Bets

The Battle Against Time

The Intelligence Premium Unwind

… to here.

So let’s take a closer look at a number of the thesis outlined in this hypothetical scenario. Where do we agree? Where are we sceptical? And are there many cases where the reaction outlined here doesn’t go far enough?

Let’s begin with the obvious: I think that the AI productivity ‘risk’ to white-collar jobs is absolutely real. As the article says, there will be no going back once a Claude agent can do the work of a $180,000 product manager for $200/month. There’s two main categories where I personally see real risk of jobs going away:

However, where I am more sceptical about the expected job losses is more complex work. That comes two-fold:

So while I do think that we will see some job losses (or at least a lack of new job postings) from AI, I think that maybe we will not be as widely affected as outlined.

With my humble technical expertise, I do think that there is some risk to AI software firms - in the future, firms might indeed be able to build software themselves more easily. But does that mean that they will all go away? Once again, I don’t fully agree, and see a few winners that will persist despite AI:

While the consequences of the shift away from large-scale, undifferentiated enterprise software might have big economic implications, once again I don’t find it bad from a ‘societal’ perspective. Just imagine how much more efficient we all could be if we had software actually tailored to our particular needs rather than struggling with SAP or Salesforce - also fueling the small businesses behind these bespoke software solutions rather than massive corporates with bloated overhead.

Moving on now to the second group of endangered companies, the ‘friction beneficiaries’. Think the business equivalents of the white collar jobs benefitting from the laziness of individuals and companies to not want to search the internet for cheap flights or cheap printing paper.

The winners are those that aggregate data, and build additional benefits at a fair price. Think Airbnb and Amazon before enshittification - companies that give small businesses the chance to build a better business by drawing on services and software-enabled efficiency from a platform. Or businesses that make it their relentless goal to make a profit not by maximizing revenue but by minimizing costs: While not a software business, one firm that comes to mind is asset manager Vanguard. They have chosen to relentlessly drive down prices of their products, keeping their competitors on their toes - and perhaps also making it AI-resilient by simply offering such products much cheaper than even AI could build it for you.

And Citrini might even be missing a winner - all other, non-technical businesses. While Citrini talks about contagion beyond the software industry, and even mentions a few winners (AI hardware producers or energy-related companies), there’s little talk of most of the ‘real economy’. And those firms could actually be winners as well: Your old-timey manufacturing firm is likely not who you would look to for software innovation, but could now have a few savvy engineers rapidly rework bottlenecks across white-collar functions, or perhaps even allow for more rapid iterations on the hardware side. It’s also those companies who are often struggling to find talent, but AI-enabled efficiency might simply allow them to do more with the same employees rather than the same with fewer. A win all around for shareholders and employees.

If certain software is no longer needed at the same level as before, software companies selling that certain software will suffer - that’s simple business. But the idea that software companies can increase their recurring revenues per seat into perpetuity, as outlined by the large software PE firms such as Vista or Thoma Bravo, didn’t just stop being true with AI - it already started happening in 2022: After the first ~10-15 years of most large-cap, SaaS enterprise businesses, the economy first saw its first rough patch since the financial crisis, and many companies looked to cutting back or at least stabilize their software spend. I don’t think that AI will make all “off the shelf” software obsolete, as outlined above, but it will likely further increase competition and price pressures.

I also don’t want to ‘beat the dead horse’ of private credit - enough people are doing that these days. But as with other industries in the past (Citrini draws a comparison to the energy sector in the 2010s), it is not surprising that the very high valuation expectations of software should one day come down. Paying 25x EBITDA for an enterprise software business, partially paid with 5-7x in debt, simply is a lofty valuation, requiring confidence into substantial growth or stability into the future. And the key word here is confidence: Even if those businesses continue to be growing, profitable firms (which I think some will be, further fueled by AI), AI is simply adding a degree of uncertainty into the 10+ year growth timeline underlying the DCF. And that uncertainty will be reflected by investors in valuations. Nothing unhealthy - more of an overdue repricing.

What is more troubling is the potential effect on mortgages, especially in software hubs. If lots of white-collar workers lose their jobs, that might indeed affect their ability to pay mortgages on million-dollar homes. That, in return, might need to fire sales, driving down prices, or even larger-scale defaults, also affecting the financial industry. But once again, maybe that is also an overdue repricing of a very overheated housing market: There are countless ideas out there on what’s wrong with the global housing market, and how you can fix that. I like the thesis outlined in Abundance the best, which says that the housing markets were brought to such lofty levels because the government wrongly incentivized on the demand side (by providing very cheap long-term mortgages) rather than the supply side (i.e. enabling more housing constructions). More demand simply drove up prices over time, which over the ZIRP time simply reached unsustainable limits - crowding out those without high-paying white-collar jobs. Of course, such a repricing would be painful, but might’ve otherwise also happened one day if we simply managed to eventually build more housing.

Just as I am not a technologist, I am also not really an economist. But it goes without saying that the prospects of AI, even if less dire than outlined here, will have an impact on not just individuals, but countries as a whole.

I think there is merit in the idea that AI might reshape the world away from the relentless globalization of the recent years. Firms will no longer look for an ever-cheaper destination for software engineers and other white-collar support stuff (think China to India to Philippines to wherever it might be next), but actually reverse that step towards a technological reshoring as ‘local’ AI becomes cheaper, and more efficient, than even the cheapest worker somewhere around the world. Those prior winners of globalization, like India mentioned in the article, will likely suffer. But it might not also just be them: I think it could also bring the danger of further enhancing the threat of unfavorable demographic distributions (think Western Europe or Japan) as older employees, often the least tech-savvy employees, are laid off or simply retired earlier.

The fiscal impact on countries is harder to assess. There will likely be some sort of effect on labor policy as employees are laid off or if companies simply hire fewer new employees, making due with the same, more efficient workforce. In return, this requires more educational measures to make unemployed individuals “AI-ready”, or to simply send them to another, perhaps blue-collar profession. In terms of budgets, it remains to be seen if the impact is large enough to actually make a dent in the already stretched budgets of major economies amid rising costs of healthcare, unfunded pension obligations, and of course, newly-required defense budgets.

But I think it does make it necessary for society to talk about how AI winners give back. I am a cautious supporter of capitalism, although with a clear responsibility of the government and the ‘economic winners’ (think large corporations and their owners) to do their part. If AI really results in a shift of human labor to ‘agent labor’, we maybe need to consider if it would be fair to also treat digital workers as taxable individuals. But of course, there’s clear challenges in that: One, it would be an unfavorable treatment to the most efficient firms, effectively putting a tax on winners while not (or less) taxing those less efficient in rolling out AI. Two, and perhaps more importantly, there might simply not be someone to tax at the prior level: A 15.000€/month product manager might’ve paid almost half of that in payroll taxes and social charges. But what if their work is replaced by a 200€/month Claude agent? Would we put a 7.500€/month ‘tax’ on that AI worker? That also doesn’t quite seem straightforward. More clear paths of course are continued fair taxation of large enterprises and their various tax avoidance strategies (tax havens, IP transfers, and so on). But even then, those gains might ‘only’ be taxed at ~20-30% rather than the prior 40-50%, also leaving a fiscal budget hole. Either way, it gives stretched government budgets another funding gap to address.

We’ve taken a look at the Citrini paper’s hypothetical future, and already touched upon a few winners, losers, opportunities, and risks that we see. But if we see this hypothetical future to be somewhat likely - how should we prepare ourselves? How can we protect our portfolios against the looming threat of superintelligence?

Before we talk about specific asset classes, we should always take a step back and assess our risk tolerance as an investor. Whether it’s the looming threat of AI reaching new levels, geopolitical crises (as we’re seeing right now in the Middle East), or other events, we should first think about how much risk we are willing, and able, to take in our investments - and also consider what the easiest ways are to reduce our risks.

This is particularly true in the current, more turbulent market environment. Amid geopolitical troubles and high valuations in many asset classes (especially US equities, a major building block in most portfolios), we actually prefer to run client portfolios at lower risk levels than intended over the long term. We would consider rebalancing to a higher risk level (i.e. higher expected risk, higher expected return) during a market correction, or if we feel like the market sentiment is calming down a little bit - hard to imagine in a time of ‘polycrises’, but not impossible.

To make this a bit more tangible: Many of the tech entrepreneurs we work and talk with come to us with portfolios that consist mostly of equities, and at best, of a little bit of gold and crypto. They spend a lot of time thinking about how they should position within those asset classes (esp. equities) to protect themselves from AI-driven threats - but forget that the much easier way to reduce their overall risk is actually by simply reducing equities in favor of less risky asset classes, in particular, bonds. Or in other words: If the threat of a 50-70% drawdown in an equity-heavy portfolio is too much for your risk tolerance, maybe you should just have less equities. It sounds simple, but after many years of strong equity markets, it can sound almost counterintuitive.

Perhaps the ‘portfolio talk’ is not enough for you - you want specific examples. Let’s run through the different asset classes:

First, fixed income. Amid an environment of likely stable interest rates that are above historic inflation rates, we think that bonds can be a great diversifier for equity-heavy portfolios. We would be mindful to diversify widely, but also pay particular attention to tight credit spreads and duration risks. We also continue to like inflation-linked bonds to hedge against higher inflation rates amid increased government spending, and continue to see opportunities in more ‘complex’ but investment-grade structured credit, such as AAA CLOs. Lastly, if you think that the AI crisis could cause severe consequences, you can also think about adding some duration to your fixed income portfolio in expectation of interest rate cuts.

We do share some of the scepticism around private credit, which might turn out to not be as stable as promised amid AI disruption and tighter credit provision, especially to software companies. However, I personally think that there might be opportunities in certain oversold vehicles that primarily invest in software debt (otherwise certain smart hedge fund investors wouldn’t offer to buy out illiquid stakes, I think).

Second, equities. Frequent readers know that we don’t make bets on individual companies. However, it goes without saying that if the aforementioned scenario comes true, there will be a much wider dispersion of return outcomes than in recent years, which could make active management attractive. We would continue to be mindful of concentration risks in large global indices, esp. when it comes to US tech stocks. While some of the heavily-weighted ‘Mag 7’ might continue to be winners, their fate could also change rapidly. On a pure valuation basis, we see opportunities elsewhere, i.e. in Europe or Emerging Markets. Ironically, ‘boring’ Europe and its asset-heavy companies could actually benefit more if asset-light software and service businesses (which make up a lot of the US indices) struggle due to an AI disruption. Lower valuations also provide a natural buffer when the outlook for those macroeconomies worsens, rather than the “priced-to-perfection” US stock market. Once again, if you feel uncomfortable with the level of equity risk in your portfolio, consider simply cutting back - but be careful around making active bets, i.e. pivoting your portfolio entirely towards perceived ‘AI winners’.

Third, commodities. This is where we see the biggest opportunity for diversification: One, in terms of gold, which could see a continued price rally as investors stay cautious around increased government debt and uncertainty (for further reading, check out Gold: A Primer). But it’s also broader industrial metals and other commodities that could profit - while intelligence will no longer be scarce, and while capital seems to rally relentlessly towards anything that has an AI sticker, one thing that will likely be scare for the foreseeable future is physical materials: Even with most white-collar workers obsolete, we can only mine so many materials required to build more chips and data centers. And especially the huge demand increase for energy infrastructure and more advanced chips will be hard to match by the mining capacity, which takes years, if not decades, to be ramped up after well over a decade of underinvestment.

Fourth and last, alternatives. While readers know me as cautiously bullish around private equity, I would understand if investors want to wait out the next months to see where AI will really go. I continue to be optimistic around smaller (non-tech) businesses, which are less prone to being disrupted by global trends and have fewer workers and tasks, making AI enablement perhaps easier than for a large, established firm. The same scepticism applies to venture, where I would go as far as to wonder whether an AI future as outlined in the model might be a death blow to investing in conventional high-tech, asset-light start-ups - they simply might not need external funding anymore.

However, where we see a big opportunity are real assets: As with commodities, we only have, and can only build so many houses. There is the outlined risk of asset price depreciation, there likely will be opportunities in scarce assets (as one AI-savvy client told me, we can build many data centers, but we can’t just build another quaint Tuscan old town). The same applies for energy infrastructure, which might see almost limitless demand in the foreseeable future - think renewable energy and energy storage.

Finally, you might wonder: Those are all fairly generic bets. What about the specific bets, like trying to find the next OpenAI, the next NVIDIA? 

That is a valid thought, and one that we have as well (as another client said: Jan, let me know if you find a small-cap Chinese chip producer that can be the next NVIDIA). If I knew which companies could be the next NVIDIA, I likely wouldn’t be writing this newsletter. As a wealth manager, we don’t quite see it as our job to find the next outlier stock, rather than making sure that clients can be on track without ever having to ‘hit it big’ beyond their initial liquidity event. But we definitely see the attraction of why you’d want to find such an opportunity. So if you want to make a few bets, make sure to size them correctly: Properly size your Aspirational Bucket that can hold such bets, and properly size them as individual, tactical trades.

Unfortunately, none of us have a crystal ball. Perhaps AI will fizzle out and stay at the current level. Perhaps we will all be obsolete workers next year and live according to the rules of Fully Automated Luxury Communism. Or more likely, somewhere in the middle. AI is without a doubt one of the biggest opportunities and risks of our lifetime. But we should not lose sleep until things are clearer - and be informed investors, with a clear plan on how to react in different scenarios.

Hopefully, it won’t be an Intelligence Crisis, but rather an Intelligence Boom.

Liked what you read? If you enjoyed this piece, make sure to subscribe by adding your email below. I write about topics covering the world of family offices, asset allocation, and alternative investments.