.png)
Matt Sambur and the heavyweights at Apollo are staring at the software sector and seeing a ghost story. They talk about "very large unknowns." They point to the generative AI "threat" as a permanent shadow over software-as-a-service (SaaS) valuations. They’ve decided that because they can't predict exactly how a LLM will replace a seat-based license, the entire sector is a radioactive wasteland.
They are missing the point.
The current "troubles" in software aren't a result of AI's mystery. They are the result of twenty years of lazy, bloated business models finally hitting a wall. Apollo is blaming the meteor for the extinction of the dinosaurs, forgetting that the dinosaurs were already too slow, too heavy, and too stupid to survive a change in climate.
The Seat-Based Lie
For two decades, the software industry lived on a comfortable lie: the per-seat license.
Investors loved it because it was predictable. CFOs tolerated it because it was the only game in town. But per-seat pricing is fundamentally anti-efficiency. It creates a perverse incentive where the software provider wants you to hire more people to use their mediocre tool, rather than making your existing team ten times more productive.
When Apollo’s leadership frets over "unknowns," what they’re actually mourning is the death of the easy markup. If an AI agent can do the work of five junior analysts, the company doesn't need five seats of Salesforce or Zendesk. They need one.
The "trouble" isn't AI. The trouble is that these software giants never learned how to price for value.
I’ve spent fifteen years watching enterprise tech companies burn through Series C and D rounds by "scaling" their sales teams instead of their product’s utility. They didn't build software; they built digital bureaucratic layers. Now that AI can automate those layers, the software itself looks thin.
The Efficiency Paradox
The consensus view—the one Sambur is peddling—is that AI is a "headwind." This is a fundamental misunderstanding of how technology cycles work.
In every previous cycle, from the mainframe to the cloud, the "threat" was always supposed to cannibalize the market. Instead, it expanded it.
The nuance the bears miss is Jevons Paradox.
As the cost of a resource (in this case, computational tasks or data processing) decreases due to efficiency, the total consumption of that resource actually increases. When you make it cheaper and easier to write code, generate reports, or manage customers, companies don't just "do the same with less." They do a thousand times more.
The software companies that survive won't be the ones defending their "seats." They will be the ones that pivot to consumption-based models or "outcome" pricing. If your software solves a $1 million problem, why are you charging $50 a month per user? You should be taking a 5% cut of the value created.
Apollo is scared because they don't know how to value a company that doesn't have a stable "headcount" to multiply by a "subscription fee." That’s a failure of imagination, not a failure of the market.
The Great Feature-Set Purge
Let’s be brutally honest about the "Software Landscape" (a term I despise, but let's use it for the sake of the argument). Most SaaS companies are just features that managed to raise venture capital.
- Are you a platform? Or are you just a clever UI on top of a database?
- Do you actually own the data workflow? Or are you a glorified spreadsheet?
The "unknowns" Sambur mentions are only unknown to companies that provide zero unique utility. If your value proposition is "we make it slightly easier to organize a task list," you are dead. You’re not being "disrupted" by AI; you’re being rendered obsolete by basic logic.
I’ve seen private equity firms buy these "feature-SaaS" companies at 8x revenue, thinking they were getting a "robust" (another hollow word) cash cow. Now they’re realizing they bought a flip-phone in the year the iPhone launched.
The contrarian truth: The "AI troubles" are a necessary cleansing. We are flushing the toilet of the last ten years of over-funded, under-engineered junk.
Why PE is the Wrong Metric
Private equity operates on the "Return on Equity" and "EBITDA Multiples" playbooks. It is a game of financial engineering.
Software, however, has reverted to being a game of pure engineering.
When Apollo says the troubles will "persist," they are admitting their financial models can't account for a world where code writes code. They are looking for "moats" in the form of multi-year contracts and high switching costs.
But in the AI era, the only moat is velocity.
If a startup can build a competitor to your $100 million ARR product in a weekend using an LLM and a lean dev team, your "switching cost" doesn't matter. The customer will switch because the new tool is better, faster, and integrated into their workflow, not just a tab they have to remember to check.
The industry isn't facing a crisis of technology. It’s facing a crisis of arrogance.
Stop Asking if AI Will Replace Software
People keep asking: "Will AI replace [Insert Software Category]?"
It’s the wrong question. It’s like asking if the car will replace the wheel.
The correct question is: "How does the architecture of software change when the cost of intelligence drops to near zero?"
If you are an investor, you should be terrified of any company that talks about "protecting their base." Defensive postures in a tech shift are a death sentence. You should be looking for the aggressors—the companies that are actively cannibalizing their own revenue streams to build the next iteration.
The Truth About "Large Unknowns"
The "large unknowns" are actually quite simple if you stop looking at the world through a Bloomberg terminal:
- The API is the UI: Users don't want to click buttons. They want results. Software that requires "training" is a legacy asset.
- Hardware is the new Software: The real margin is moving to the chips and the infrastructure, leaving the "application layer" to fight for scraps unless that layer provides proprietary data or unique hardware integration.
- The Death of the Generalist: Generic CRM, generic ERP, and generic Project Management are commodities. Vertical-specific, deeply integrated AI agents are the only things that will hold a premium price.
Apollo is right about one thing: the pain will continue for the companies they usually like to buy. The slow, the steady, and the "stable" are about to be slaughtered.
But for the rest of us? This isn't a "troubled" period. This is the most efficient the market has been in thirty years. The fluff is being evaporated. The "unknowns" are only scary if you were planning on coasting.
If you’re still waiting for a "clearer picture" before you move, you’ve already lost. The picture is clear: the era of the lazy software subscription is over.
Now, go find the people building the tools that make the old world look like a joke. Stop listening to the billionaires who are just upset their 2019 spreadsheets don't work anymore.
The software industry isn't dying. It's just finally becoming useful.
