AI: THE FUTURE OPERATING SYSTEM "Software" was the Bubble
- Feb 25
- 12 min read
Observing the Herd
THE REPRICING OF PERFECTION
It has been a few months since I last wrote an Observing the Herd. With the sharp repricing across software and SaaS, it feels like the right time to step back and reassess what is actually happening, at least in my view. We briefly discussed this in our last update, but I feel it is worth diving deeper given the pivotal technological change in which we now sit.
For more than a decade, software was the market’s favorite child. Asset light. Capital light. Minimal incremental cost. Infinite scalability. Recurring revenue. Expanding margins. It was hailed as the purest form of modern capitalism. The cleaner and better use of capital. Low capex, high return on invested capital, global distribution at the click of a button. The market happily began to lend massive premiums for this outlook. That, by most terms, is typically considered a bubble in itself.
Markets priced it for perfection. As of the publishing of the chart above, even after severe declines, the industry still trades over 25x NTM EV/FCF levels, which most value investors would find extremely high without producing immense growth. And it is that growth that is under attack and facing immense uncertainty.
Multiples expanded on the belief that each incremental dollar of revenue would translate into disproportionate earnings power. Even after significant declines, many software businesses still trade at valuations that assume stability and durability. But, I will admit, some are now becoming shockingly cheap by historical standards, but the certainty that once surrounded the space has fractured.
Artificial intelligence has more than officially arrived, not as another feature, but as a structural force. It’s becoming what I believe we have traditionally called an operating system.
It would be unreasonable to say software is dead. It will simply look different. But what once appeared to be defensible moats around code, integrations, and scale now look more permeable, with floods of competitors coming at legacy SaaS companies daily.
The value of owning a static codebase diminishes when code can be generated instantly. Users will soon become spoiled with limitless customization and error debugging that never makes the user diagnose, but rather, AI can diagnose and patch itself, while eventually transforming into any hardware desire, which previously took an agreed upon operating system.
If you have experimented with something like OpenClaw, you can see this clearly. For many, it has been a window into the autonomous future, one that has existed for well over a year now. You give high-level instruction, and agents coordinate tasks, generate applications, connect systems, debug errors, and iterate. It feels less like using software and more like directing a team.
Watching people rush to buy Mac minis recently reminds me of the early iPhone era. Most people forget that the original iPhone could not even copy and paste nor did it even ship with an app store of its own. Yet developers were already jailbreaking devices, building underground app stores, and experiencing the future before it was commercialized for the masses.
OpenClaw feels like that moment. Early. Imperfect. A little raw. But clearly pointing toward something much larger.
FROM DASHBOARDS TO DIRECTION
Artificial intelligence is not just another application layered on top of existing systems. It is becoming the operating system itself and eventually removing nearly all of our interaction with it. Not the kind that organizes files and icons, but an adaptive intelligence layer that sits between human intention and digital execution, understanding what we want done in ways that will be natural to us, granted to the AI by the flow of our data. Systems will simply just know what to do.
Microsoft democratized and unified personal computing for the masses with Windows and the PC ecosystem. Apple refined it into an elegant, integrated experience with the Macintosh and later macOS/iOS. The internet connected it all, turning isolated machines into a global, always-on network of data, services, and cloud intelligence. Google (via Android) and Apple (via iOS) then unified mobile computing, putting supercomputers in every pocket with instant access to that network. Now, artificial intelligence unifies intention, shifting the paradigm from explicit apps and commands to natural, proactive understanding and execution of what the user truly wants, powered by the connectivity foundation that came before, ripping through any digital limitations or bottlenecks of the past by now being capable of coding itself.
The paradigm is transforming quickly, especially when realizing true “super” intelligence is no more than five years away. We humans need to recognize that and understand what it means for work, which I believe will accelerate, not disappear. Instead, work becomes execution, not planning or being inundated with maintenance or simple data entry. Those days will seem like a distant memory ten years from now, with most people not comprehending how we once “worked,” at least what we considered work, which will look trivial in hindsight and leave future generations dumbfounded at how serious we once thought we were filling out endless forms where future generations will view forms like we now view fax machines, rarely ever having to interact with them.
Dashboards assume the human is the analyst. That we must ingest information, synthesize it, and execute action. Artificial intelligence inverts that relationship. It consumes data continuously, far better than any human could imagine. And it can already do that today. It learns from communications, transactions, habits, biometric signals, and behavioral patterns. It anticipates.
We are moving from operators to directors. It’s feasible one day we do not have to interact with technology much, at least in the way we view our interaction today, as it’ll be embedded as a part of life. Imagine the wealthiest of elites, they’ve built systems. They have assistants, aids to send them morning briefs, organize their calendars, orchestrate their messages. We will all have this power, even if you don’t take the time to build it yourself like many have been over the last year. OpenClaw pushes that reality closer, even if just conceptually.
Hardware becomes a conduit. A microphone. A camera. A wearable. A vehicle. A screen. A factory floor. A robot arm. Every physical object becomes an interface layer for your personal AI. You will not log in to most things. You will communicate, or not even communicate at all. It will just be a part of life. The system will infer your objectives and act accordingly. It will eventually be frustrating to be restricted by a keyboard or mouse until they eventually become relics of the past.
Many ask, but how? Well, how do YOU do anything? It isn’t magical. And if that action is pushed through a technological device like a computer today, AI will do that better, if it doesn’t already.
I visited my mom the other day, discussing this exact concept. I said my car will simply know where I’m going and drive me there to which she said, “But how? How would it know?” So I asked, “Well, how did you know I was coming here?” We had texted about where we were meeting and when. That’s all the data it needed if accessible to it. Everyone will eventually demand no data is unprocessed by their AI.
Software, once scarce and expensive to build, becomes effectively abundant. Integration barriers collapse when AI can read documentation, generate APIs, debug itself, and optimize continuously. The supply of functional software logic trends toward infinity.
When supply becomes infinite, pricing power erodes. The supply of knowledge itself trends toward abundance, which is massively beneficial for the majority of society. Individuals denouncing AI specifically in pro-knowledge areas are, in my opinion, naively misinformed. I assure you, we will not be capable of clicking or typing faster than any AI machine will be able to operate, ingest infinite data, and act near instantly more informed than we could ever dream of being. Pro-AI developlers have alrady witnessed this "magic."
We must look to how these machines will work for us in the future and disregard how that looked in the past. Imagine describing business today prior to the internet. Insane. So, why not look to the future? Why not practice humility, understand our limits, and leverage that? I’d recommend doing so quickly.
It is not hard for me to picture a day when traditional software engineers are viewed the way we now view telephone line operators. Essential in their time. Foundational. But abstracted away by automation. They were once a bottleneck of infinite demand waiting to use these wonderful systems but were limited by the gatekeepers. AI blows that door open for technology as a whole, limited to no area whatsoever. I do believe engineers will be in high demand, but as orchestrators, not as the independent coders.
Consider Salesforce or HubSpot. Their value was built on centralizing data and workflows inside structured platforms. But when a user can instruct their AI to build a contextual system in hours, ingesting emails, calendars, transaction histories, and documents, the competitive equation shifts. Incumbents must pivot aggressively. Simply operating a large legacy system is no longer a moat. It can become (already is) friction.
The difference with AI remains the speed of development, which is accelerating. It changes the business narrative to act fast, rather than waiting to see what happens. But, business cycles and the shift in where the market holds premiums is nothing new.
THE RETURN OF THE PHYSICAL PREMIUM
For years, markets discounted the physical world. Manufacturing. Heavy industry. Infrastructure. Energy. Materials. These were considered capital intensive, cyclical, and expensive with little growth. Software was the cleaner growth story. Less capex. More scalability.
Artificial intelligence flips part of that narrative. In the markets, it already has been underway for a over a year now.
If software logic becomes abundant and commoditized, scarcity shifts to what cannot be instantly replicated. Energy. Compute. Rare earths. Semiconductor fabrication. Advanced robotics. Physical infrastructure. Real world logistics. Human-authenticated environments. While the digital world can scale unlike ever before, it doesn’t change physical limits.
As digital abundance grows, tangible constraints become more valuable.
We are witnessing a potential temporary inversion. Physical businesses that were once discounted for capital intensity could command premiums reminiscent of what software once enjoyed. Until markets fully understand with more clarity around robotics, autonomous systems, and physical AI risks, tangible assets may appear more defensible than infinitely replicable code.
It took markets years to grasp the implications of artificial intelligence in the digital realm, surprised by its sustained and rapid development . The repricing is still underway. The same may happen in the physical realm once robotics scales. But in the interim, premiums could migrate.
The era that worshiped asset light models may rotate toward asset-backed, truly physical realities.
Meanwhile, the deeper transformation continues.
Artificial intelligence as an operating system means we no longer need to be the constant input mechanism. No more endless toggling between apps and dashboards. Simply see what information pertains to you and be bothered by nothing else. No more manually reviewing terms of service. No more filtering spam. Agents will negotiate contracts. File taxes. Reconcile books. Schedule maintenance. Optimize supply chains. Rebalance portfolios. Authenticate identities.
One day (soon), even accepting a calendar invite will seem absurd. You already discussed the meeting. Why the extra step? Why tell your system, yes, I really am going even though I just clearly stated I was, but I will additionally click this checkmark to tell you so? These small frictions will disappear, and we’ll all be more productive for it. Velocity will accelerate.
Future generations will find our debates over job displacement almost quaint, similar to every single one in the past. Agriculture once employed roughly half of Americans just one hundred years ago, not even breaching the full age of the oldest people on earth. Today it employs less than two percent, yet food is more abundant than ever. Switchboard operators vanished, but communication exploded. Floors of typists were erased, but corporate communications have grown exponentially.
Artificial intelligence simply raises the baseline again.
FUTURE GENERATIONS WILL KNOW NO DIFFERENT
My children will not experience AI as a novelty. It will be ambient. They are already building custom games, even multiplayer builds with friends, beyond my imagination without touching any lines of code. That is happening already today. By the time they reach midlife, they may struggle to comprehend that we once manually searched emails or clicked through dashboards to extract information. It will feel as primitive as punch card computing feels to us.
Consumers will demand this convenience. How many times have you stared at screen knowing what you want done but either overwhelmed with the steps to do it or frustrated without the know how? That feeling will not be understood by future generations. As these tech hurdles are obliterated, your spam and unwanted communications will be overwhelming. You’ll want an intelligent assistant reviewing these and notifying you only of what’s important. AI will become the gatekeepers. As that relationship grows, so will the tasks it handles.
I cannot even fathom what the world will look like in just a few decades. Humanoids cleaning homes and caring for most basic needs may be viewed no differently than we view nearly every American owning a cell phone today. We may feel we’re special and this AI technology is “new,” but every technological innovation of the past was new to that generation too.
Markets are wrestling with uncertainty in software because it was priced for perfection. AI dismantled that illusion and replaced it with ambiguity.
But zoom out.
This is not the death of technology. It is the next chapter. A reallocation of premiums. A migration of scarcity. A shift from dashboards and manual inputs to autonomous orchestration. We viewed our input as superior, but we must eventually accept when it is no longer the case. And this does not only apply to data and knowledge , it will impact every interaction with technology.
Optimists in general have historically been rewarded. I remain in that camp. Engineers and programmers will still be vital to this new era. Imagine all the jobs created once telephone line operators were no longer a bottleneck? AI will connect our technology, and demand for that technology will and already is exploding.
And while this feels faster than any technological innovation before (because it is from a large base), but every generation likely felt the same from the inside. In decades’ hindsight, this period may simply be viewed as another technological inflection point that pushed humanity higher once again. And I could not be more excited to see what comes next.
All the innovations and escalation in productivity that will occur in the period ahead of us will let history laugh at those that feared economic demise from one of the greatest technologies yet to be invented. But, many of these shops blaring these theories are typically entrenched in research alone. If I were in that position, I would be concerned too, but I'd already be navigating a pivot.
For others, just imagine, every healthcare facility operating at the level as if it were a Mayo Clinic. Each financial advisor giving the absolute best possible advice. Some say this will erode knowledge much in the way Socrates decried writing as an erosion of human memory. I clearly disagree. It will empower those with ambition with the ability to be one of the best, elevating all peers alongside them. Every one wins in the long run.
Yes, displacement, especially in the next few years will occur, much in the way entire industries were eliminated from past innovations, always leading to a better future. Yes, AI's application is more broad, but it is leading to a future that will quickly make our corporate processes seem ancient and interaction with technology elementary for what will soon be possible. It will alter the entire landscape and be a massive reshuffling for many industries. But, the displacement will pail in comparison to the growth that ensues.
I see AI as the next augmentation, a tool, to make us all better, much the way tech in the past has before for those leveraging it, where the tide is rising for all boats long term. We’re observing the markets trying to determine individual impacts in real time, which will remain volatile as that future around technology is more unlcear today than ever before.
But, AI is simply making technology be the tool we’ve always wanted, intelligent and convenient.
It will simply... just... work.
Companies, especially technology firms, must build for that future where dashboards are rare, data is instantly analyzed and intelligently acted on, only bothering human users when absolutely necessary. If not, I believe they’ll face extinction; not from a "fear mongering" perspective, but rather a simple realization that AI is nothing more than the next evolution of technology that any business would struggle without.
Comments