Continuing my thoughts after my previous article, and having too much free time lately, I have been spending more time than I should browsing LinkedIn, the place where the jobs descriptions die, but the AI influencers thrive.
The subscription problem
After reading too many AI posts, followed by too many anti-AI posts, I’ve started to forming some kind of belief that the AI psychosis we’ve seeing over the last year is- at least some extend of it- a direct consequence of something that started long before AI was there: the endless subscription models that almost every major or minor company has imposed on us.
For years, we’ve gradually moved from buying software to renting it (who remembers the good old days that when you bought a laptop, it was also included the actual CDs to install Word?). Instead of paying once for a product, we now pay a monthly subscription to keep accessing it. Yes, we are paying for infrastructure (since everything seems to be on the cloud), ongoing development, support, content or a service that has recurring costs.
But sometimes it feels like our money goes to something that it is more like a collection of features wrapped in a nice-ish interface.
Two Types of AI Enthusiasm
Upon browsing Linkedin you can see too major groups:
The first group has realised that they can build a spinoff of an application they need/want/like for a fraction of the cost. They don't necessarily need to build something sophisticated, with the best architecture or commercially viable. They just need something that solves their particular problem and AI helps them to achieve their goal.
The second group has a different thought model. They have realized that they can build a spinoff of an application they need/want/like, and then they assume that because they were able to build it more or less cheap and easy, there must also be an opportunity to monetize it.
In my humble opinion this second group is where some of the current enthusiasm around AI starts becoming disconnected from reality. Being able to build something in a low price does not mean that someone is willing to pay you for it or you solve any particular novel problem.
AI Didn't Democratise Software
So, did AI unlock our potential to democratize code or reduce (finally) the prices? Absolutely not.
What AI has done is reduce the cost and effort required to produce certain types of software. We have to consider the difference between the cost of creating that said software and the value that this software provides.
If the software is just an interface around something easy to reproduce, then yes, AI can make it much less valuable. But there are plenty of businesses where the software itself isn't the main reason people buying it. That could be the trust in brand or the security promise etc.
Software Isn't Always the Product
There are some services that you can’t really hurt and it doesn’t even matter if they use AI or not. Their value comes from something the average person can’t reproduce: their content.
Streaming services are a good example. Netflix can charge (or change whenever they want) a monthly subscription not because the application is difficult to recreate (the contrary), but because its value is the content available inside the app.
Yes, you could build your own basic streaming application version very easy. But that doesn’t give you access to the catalogue of movies or shows.
Sure, you could start buying DVDs, digitise everything, build a ridiculous amount of storage and spend the rest of my life maintaining it. But at that point, I might as well just pay for the netflix subscription.
Amazon works in a similar way. There is nothing revolutionary about the concept of an online storefront. A developer can build a basic version of an Amazon-like interface relatively quickly (with or without AI).
But Amazon isn’t valuable because it has product cards, a shopping cart and a checkout page. The real value is everything behind them: the sellers, warehouses, logistics, delivery network, payment systems, customer service and, perhaps most importantly, the convenience of having almost anything delivered to your door.
The Software We Never Built
The truth is that AI hasn't made software worthless but rather made a particular category of software easier to reproduce.
The last years, we have accepted subscription software as the default. If we needed a particular functionality, we would search for a SaaS product, create an account and add another recurring payment to a long list of subscriptions we already have. But now we can just consider: "Can I just build and maintain the thing I actually need?"
What Actually Changes?
Realistically, I don't think everyone is going to become a software entrepreneur, and I certainly don't think every AI-generated SaaS product is going to become a successful business. Actually, i believe the opposite, most of them will fail, because when everyone can build another spin-off, building anotherx2 spin-off isn't much of a competitive advantage.
In fact, we are already seeing an enormous amount of software being created that nobody actually needs or cares about. What I do think is changing is the relationship between users and software companies, now the users can make the decision between building or buying that software more or less in the same price (or even cheaper).
These tools won't make us billionaires or help us to make the next billion-dollar application. I think the future lies more in making the tiny, boring, incredibly specific tool that only you need economically reasonable.
The companies that should be worried aren't necessarily the ones selling complex technology or very good content. But the ones whose entire value proposition is a thin layer of software sitting between the user and something the user can now reproduce themselves.
So What is your take? Do you think AI will change the way we buy/rent software?
Will we start building the small tools we need ourselves or will convenience and reliability still be enough to make subscriptions withstand?
Which types of software do you think are most vulnerable to this shift?
This article reflects my own thoughts, and is written by me, but i use AI to help me with the grammar!
This article isn't about AI being good or bad, but more than how the economy changes around it. You can agree or disagree but be nice in the comments.
Top comments (11)
Great post Alex. There is a lot of thoughts I have but going to condense it the best I can.
I believe so. The main thing is that since it's easier to build software, people will monetize it. When people monetize it, people get greedy. This is heavily relying on Subscription base because we are starting to see more throughout the years, which is unfortunate in my opinion because some companies have a policy that you are basically "renting" and not owning. You seen the Sony PlayStation news about them removing physical disks? (though not related to subscription, it ties to players not owning their games because of the digital future.)
I think open source software since it's more transparent, especially in this case for the age of AI. We would like to see who built the feature and the responsibility of it. A lot of software that are "closed sourced" tend to shutdown the product, leading to users not using the product anymore because that software no longer exist.
Unfortunately, that occasionally works lol. I feel like this is more towards Y Combinator than anything, but idk.
Good post again :D
Two thoughts:
That is true, but now there is this shift that "why should i pay your subscription, if i can make it myself?", most of this greed to monetize will flop eventually!
That's my point! If the content is rare, you cannot find it in a physical form, and cannot recreate it. These are the "subscriptions" that will last (probably).
This is such a good point. Software engineers now need to understand the business of software in order to add value, whether we are hustling on our own side projects or working a day job for big capitalism. Not to sound like a LLM, but, "the code was never the point!"
Yes, I also thought it was important to know the business and the users, but the world we live now makes it even more important. I guess the software that survives will be in one of the two buckets: great content or solving an important problem.
Interesting article!!
Hello,😁
Thank you for this highly realistic and refreshing perspective. Your point about "subscription fatigue" driving the current AI hype is incredibly accurate.
As someone who builds privacy-first, client-side web tools, I completely agree with your conclusion: the real future lies in "tiny, boring, incredibly specific tools" that solve actual problems, rather than thin wrappers chasing monetization.
Thank you for cutting through the noise and providing such a grounded, practical analysis. It is exactly the kind of clear thinking the developer community needs right now.
Wishing you continued success, progress, and creativity in your writing and future projects! 😊
With sincere appreciation,
A fellow developer & enthusiast
🧊🗻🌊
Thank you so much! 😊 I really appreciate you taking the time to read it and share such a thoughtful perspective.
the amazon/netflix point is the sharpest part here. we build small automation tools at viaSocket and see this exact split daily - the thing people actually pay for isn't the trigger-action logic (AI can write that in a weekend now), it's someone else already having solved the auth handshakes, rate limits and edge cases for 500 different APIs. that's boring, unglamorous, and exactly the kind of moat AI doesn't touch.
Interesting perspective. As AI becomes more integrated into our daily lives, the challenge is not only about improving technology but also about creating healthier ways for people to interact with it.
Balancing expectations, accessibility, and responsible usage will be important as the AI ecosystem continues to grow. Great discussion on a topic that deserves more attention.
Some comments may only be visible to logged-in visitors. Sign in to view all comments.