Last month, Citrini Research published a thought piece on the potential impact of AI on the economy. They presented it as an article written in 2028 when the imagined unemployment rate exceeded 10%. The gist of the piece was that AI replaced the functionality of so many Software as a Service (SaaS) companies that those high-margin firms laid off developers and salespeople. In this hypothetical future there were so many firings of high-income people that the entire economy suffered as millions no longer were able to support their disproportionate consumption. People currently earning hundreds of thousands of dollars a year turned to driving for DoorDash to pay the mortgage. The article was powerful enough that SaaS stocks got crushed the day after it was published. I recommend reading it: The 2028 Global Intelligence Crisis.
While the stock market interpreted the article as Citrini’s prediction, the article clarifies that it’s intended to be a “scenario that’s been relatively underexplored”. A couple of weeks ago, Jack Dorsey’s Block laid off 40% of its staff due to AI so it’s clear Citrini can’t be completely wrong. Finally, Alex Macris at Contemplations on the Tree of Woe, a regular guest poster at DKI, wrote a similar article where he expressed concern that when the US economy loses the consumer spending of these soon-to-be-fired high-income people it will lead to a horrible recession. Alex is smart, thoughtful, and strategic. He’s seeing the same thing Citrini is.
I’m less concerned about this potential outcome. Losing expensive friction isn’t necessarily a problem. When I first started my finance career, individual pieces of software tended to cost $50 – $70 on average with some programs costing closer to the $150 range. That was for a purchase. Once you bought that software, you owned it. If you upgraded your computer, you could easily install the program on your new machine. Because internet upgrades weren’t possible, developers worked to ensure software shipped finished.
Let’s contrast that with what we see today. Many programs that used to be available for purchase are now only available for a monthly or annual licensing fee which can resemble the previous purchase price. You can’t own it. Because all programs connect to the internet, developers ship beta versions and figure they’ll fix problems with future updates. A developer can cancel your license if they don’t like the way you use “their” software or if you get cancelled for politically incorrect views. They can prevent you from transferring your license to another computer.
Companies claim that they charge a lot because they provide service. Has that been your experience? What happens when you call tech support? In my experience, you have to search for a long time to find a phone number. Then, you have to call and authenticate yourself. After a long wait, you get to speak with someone who isn’t a fluent English speaker who insists that you re-authenticate yourself as they ask the exact same questions the automated system did. Then, after you explain the problem to tech support, they spend the next two minutes reading a script explaining to you how important it is to them that they help you fix the problem…instead of fixing it. Finally, they put you on hold for ten minutes while they look up the answer to the question you originally asked. It’s not a great experience and the software companies are increasingly replacing their tech departments with AI ensuring fewer of us will feel badly when the job losses come.
Worse, many of these SaaS companies act as monopolies. As a result, they rarely update or improve their user interface and instead, rely on high switching costs and a lack of alternatives to keep the monthly revenue flowing. These companies also make it confusing and difficult to cancel. I read a report over the weekend that Adobe requires users to go through nine separate steps to cancel an account. The idea is that by creating friction in nine places, users will eventually give up and just continue paying. If true, these people make terminating a gym membership seem easy. If you have a high-quality product at a great price, you don’t need that many barriers to leaving.
I accept Citrini’s premise that lots of high-income people losing their jobs at one time can lead to economic disruption. However, I’m on record as noting the perverse way we calculate GDP for our “consumer economy”. It’s not consumption that creates value; but rather, production. Many of these SaaS companies have gone from producing quality software for a reasonable one-time fee and no license restrictions to a model featuring poor user interface, big monthly fees, no ownership, no control, and terrible customer service.
Most importantly, if AI-aided coders can replace these expensive seat licenses, do we really need companies to continue to overspend on monthly software licenses to protect the economy? It’s not like that money disappeared. Every million dollars that doesn’t go to Adobe or similar SaaS companies means that exact same million dollars stays in the pockets of other businesses and business owners. The transition from our current job environment and economy to whatever the new AI-powered one will be is going to be painful for a lot of people; just as previous technology-led transitions remade employment in the agriculture, manual and factory labor, blacksmithing, secretarial, and scroll-copying industries. Some of those changes disrupted the physical labor market while others disrupted white collar jobs. (People seem a lot more concerned about SaaS salespeople losing their jobs than they do about cab drivers and uber drivers losing their positions to self-driving cars.) My point is that the lost profits to the SaaS companies and their employees will be gained profits at their client companies.
I think the difficult part of Citrini’s scenario description is the anticipated speed of disruption. When millions of people lose their jobs all at once, it’s a much bigger problem than when the change creeps along over a decade or two. While the AI transformation isn’t going to take decades, there are some aspects of it that might take a little longer than some think it will. Last weekend, I saw reports that a senior executive at a tech firm lost all of her emails. I’ve seen other reports of rogue AIs wiping out massive amounts of company data. These companies haven’t protected against this because they didn’t expect the data losses to be coming from inside the company. Block firing 40% of its employees overnight was a real warning. I’m just noting that these AI agents aren’t ready to be trusted without some level of human oversight.
The models matter. I’ve been using ChatGPT extensively over the past year and I don’t trust it. It gives me wrong answers with confidence. After I correct it and ask for verified information only, it continues to make the same errors and make the same suggestions based on the same erroneous assumptions. It takes me longer to edit business images using ChatGPT than it takes the DKI interns to handle it. Basic changes take 30 minutes for the LLM and 5 minutes for the interns. I’ve also found the model to express a wide range of insane takes on social and political issues. It frequently gives me responses that are racist or sexist, and some that are historically illiterate. I’ve heard that the newer Claude models are more accurate and more useful. For now, it’s clear that while AI will have a huge long-term impact on the economy, many of the models aren’t ready and don’t deserve our trust yet.
Citrini clarified that they were outlining one possible future. I agree that we could have an uncomfortable transition to newer technology. My issue with their piece is that they explored the cost to the SaaS model of declining relative utility without noting that every lost dollar of SaaS revenue is a found dollar of lower expenses to the customer. Right now, these companies are taking advantage of near-monopoly conditions and friction to extract huge monthly fees. Transitioning to a more competitive business environment with leaner margins has been something corporate America has had to do many times in its history. Do we need monopoly margins and friction to keep the economy running?
To use one final example from finance, this morning, I read about an AI-powered platform that is designed to replace many of the functions of the Bloomberg terminals found at many Wall Street firms. No one is predicting the imminent demise of Bloomberg, but is it a bad thing for society if some (or many) people switch from a $30k/year subscription to one that costs $2,400/year? Bloomberg claims they’re working on more AI-powered features as well. Sounds good. Let’s see multiple parties fight it out based on features, user interface, customer service, and cost. It might be difficult, but it won’t be all bad. I think.
Information contained in this report is believed by Deep Knowledge Investing (“DKI”) to be accurate and/or derived from sources which it believes to be reliable; however, such information is presented without warranty of any kind, whether express or implied and DKI makes no representation as to the completeness, timeliness or accuracy of the information contained therein or with regard to the results to be obtained from its use. The provision of the information contained in the Services shall not be deemed to obligate DKI to provide updated or similar information in the future except to the extent it may be required to do so.
The information we provide is publicly available; our reports are neither an offer nor a solicitation to buy or sell securities. All expressions of opinion are precisely that and are subject to change. DKI, affiliates of DKI or its principal or others associated with DKI may have, take or sell positions in securities of companies about which we write.
Our opinions are not advice that investment in a company’s securities is suitable for any particular investor. Each investor should consult with and rely on his or its own investigation, due diligence and the recommendations of investment professionals whom the investor has engaged for that purpose.
In no event shall DKI be liable for any costs, liabilities, losses, expenses (including, but not limited to, attorneys’ fees), damages of any kind, including direct, indirect, punitive, incidental, special or consequential damages, or for any trading losses arising from or attributable to the use of this report.
