Derek Mills

HSBC Called It the SaaSpocalypse and Now I Can't Stop Using That Word

February 23, 2026

I said it to Tory yesterday morning and he just nodded slowly and said "that's good, man, that's a really good word" in the voice he uses when he's not actually listening. His phone has been face-down on his desk all week. I don't know what's going on there. But I do know that SaaSpocalypse is now permanently embedded in my vocabulary and I refuse to apologize for that.

Here's the story, because it's a genuinely significant one and it's moving faster than most people realize.

What Actually Happened

In early February 2026, the software sector fell apart in a way that hadn't been seen in years. We've been watching this slow-motion crack for a while, but this was different - this was a cliff. The trigger was a cascade of bad signals hitting at once. ServiceNow reported Q4 earnings that beat headline estimates but buried a troubling deceleration in future contract commitments. Anthropic previewed an agentic AI product that made investors antsy about entire categories of software becoming redundant. And then the whole thing fed on itself.

The numbers are not small. Between late January and early February 2026, the damage was staggering. Salesforce lost nearly 30% of its market cap. Adobe fell 19%. Atlassian dropped 30% in a single week. ServiceNow, HubSpot, Workday, Docusign, Asana - all of them got hit. The well-followed software ETF known as IGV had already fallen 30% from its September 2025 peak by the time things stabilized. The S&P North American Software Index had fallen 15% from its December highs by early February. Hedge funds earned $24 billion by shorting software stocks during this window. Twenty-four billion dollars.

The phrase "SaaSpocalypse" - first coined by a Jefferies trader, then picked up everywhere - captures the specific fear underneath all of this: that AI agents are about to make the entire per-seat SaaS business model structurally obsolete. If an AI can do the work that previously required 100 employees, a company doesn't need 100 software licenses anymore. The per-seat pricing model, which is basically how every major SaaS company makes money, suddenly looks like it was built on a foundation that's cracking.

Where HSBC Comes In - And Why I Think They're Mostly Right

In the middle of all this panic, HSBC's head of US technology research, Stephen Bersey, dropped a research note titled "Software Will Eat AI" - a deliberate inversion of Marc Andreessen's famous 2011 "software is eating the world" line. His core argument: the market is misreading the situation. AI won't replace enterprise software. Enterprise software will absorb AI and become the vehicle through which AI reaches actual businesses.

HSBC's analysts made three specific points that I keep coming back to. First, that consumer AI platform developers - OpenAI, Anthropic, Google with its consumer-facing products - have essentially no experience building enterprise-class software and would be "architecting from scratch in unfamiliar, highly complex areas" if they tried. Second, that it's not practically or economically sound for most companies to build custom in-house AI systems to replace established software. Third, and this is the one that stuck with me: even vibe-coding - generating usable software from natural language prompts - wouldn't easily displace the vendors that run the day-to-day operations of global companies. Those vendors have decades of data, integration depth, and compliance architecture that you can't replicate with a good prompt.

HSBC put buy ratings on Oracle, ServiceNow, Salesforce, CrowdStrike, and others. Their exact framing was that "sector valuations are at historical lows, even though we believe the sector is poised to expand massively." Contrarian timing, clearly. And I think they're largely correct.

Here's the thing that keeps getting missed in the doom takes: the evidence that AI actually replaces enterprise SaaS at scale is genuinely thin. The most cited example is Klarna, whose CEO announced in 2024 that they were cutting 1,200 SaaS tools including Salesforce and Workday, replacing everything with internal AI. The AI-based customer service system reportedly replaced 700 employees and saved $40 million annually. Big numbers. Terrifying numbers, if you're a SaaS company. But by early 2025, Klarna had already started backtracking - rehiring customer service staff because the AI-powered systems were causing quality problems. The radical experiment hit a wall. That story never gets told as loudly as the original announcement.

What the Panic Got Right, Though

I want to be clear that I'm not saying the SaaSpocalypse fear is entirely wrong. Some of it is correct and important. The per-seat pricing model is under genuine pressure. The percentage of companies still relying on seat-based pricing fell from 21% to 15% in a single year. Usage-based and hybrid pricing models jumped from 27% to 41% in the same period. Salesforce is already experimenting with flat-rate "Agentic Enterprise License Agreements" for companies deploying AI agents at scale. That's not nothing - that's a fundamental renegotiation of the core business model that built the SaaS economy.

The market also wasn't wrong to punish companies that have been overselling "AI potential" without showing AI proof. The sell-off represented a real shift in what investors are willing to pay for. Software companies that kept saying "our AI features are coming" while still growing revenue primarily from headcount-based seat expansions were always going to face a reckoning when agentic AI started actually shipping. That reckoning is legitimate.

Global SaaS spending is still projected to grow from $318 billion in 2025 to $512 billion by 2028. Forrester isn't calling this the death of software, they're calling it a transformation. The core enterprise functions - finance, CRM, HR, compliance - aren't disappearing. The software running those functions is evolving. There's a meaningful difference between those two sentences, and most of the financial press blurred them completely during the panic.

This reminds me of the conversation I have every time I bring up The Last Jedi. People see the surface - Luke Skywalker failing, the Resistance on the run, things looking terrible - and they call it a disaster. They miss that it's actually a deliberate structural break, a burning of the old model so something new can emerge. Rian Johnson got blamed for killing Star Wars the same week that Star Wars made a billion dollars. The SaaSpocalypse coverage has that same energy. The incumbents are "dying" while quietly retooling to eat the very thing threatening them.

I told Chris this and he nodded along in that way he does, that extremely kind, slightly vacant nod. He asked if I wanted the last of the conference room coffee. Honestly the man is too good for this office.

The Part That Actually Matters for People Running Businesses

If you're not an investor in software stocks, here's why this still matters to you: the SaaSpocalypse is going to change your software costs, your contracts, and your vendor relationships over the next two to three years. Not maybe. Definitely.

Vendors are going to restructure pricing. The companies that survive this transition will be the ones that shift to outcome-based or usage-based billing - charging for results rather than for seats. For businesses on the buying side, that means the negotiating leverage is going to shift. For the first time in a decade, you're going to have real power in contract renewals, because software companies are scared. They need retention. They'll deal.

We ran a software audit here a while back - it got weird fast - and the thing that stood out was how many tools we were paying for on seat counts that no longer matched actual usage. That mismatch is now a negotiating argument. "Your competitor charges per outcome now. What are you going to do about this contract?" That conversation is available to you right now and most businesses aren't having it.

Stephanie, when she saw the Salesforce valuation drop, asked whether we should just buy the company. She was not joking. She doesn't really have a concept of what's a reasonable thing to consider purchasing. But her instinct - that right now is a strange buying opportunity rather than a moment to panic - is actually the same thesis HSBC is running with. Except HSBC is talking about buying stock, not the company outright.

The companies most at risk aren't the Salesforces and the ServiceNows - they have the data moats, the compliance infrastructure, the enterprise relationships. The companies most at risk are the mid-tier specialized SaaS tools that do one thing, charge per seat, and don't have enough proprietary data to make AI work better than a general-purpose agent. If your business relies heavily on tools like that, pay attention to what's happening in this space right now. Some of those vendors are going to get consolidated, acquired, or quietly defunded in the next 18 months.

If you're using a CRM or any major sales tool, the companies behind those products are actively trying to show that their AI is additive, not just defensive. The smart ones are winning that argument. But your job as a buyer is to make them prove it.

The Word Itself Is Doing Real Work

I keep coming back to the fact that HSBC named this thing. They didn't just analyze it - they gave it a word that stuck. "SaaSpocalypse" is now everywhere, in analyst notes, in Inc. Magazine, in conversations on trading floors and in Slack channels. Language matters in markets. When something has a name, it has gravity.

And I think HSBC named it specifically because they wanted to own the narrative around it. You don't call something a "SaaSpocalypse" in a note where you're recommending buying SaaS stocks unless you want to control the conversation. It's a bold move. Acknowledge the fear, name it dramatically, then explain why you think the fear is overblown. "Software will eat AI" as a counter-thesis only works if people already believe the apocalypse is possible. The drama is the point.

Does that make me trust the analysis less? A little, honestly. HSBC has buy ratings on Oracle, Salesforce, ServiceNow - they have incentives here. The distribution-beats-innovation argument they're making is compelling but not airtight. If AI agents become sophisticated enough to replace entire application categories - not just augment them - then incumbent advantages might not matter as much as HSBC thinks. That's a real risk they acknowledge in the fine print while leading with the bullish headline.

But the base case - that enterprise customers don't want to rip out their existing software infrastructure and start over with AI-native tools - feels correct to me. Most businesses I know are not scrambling to replace their core systems with custom AI builds. They're asking their existing vendors to make things smarter. That's exactly the structural advantage HSBC is pointing to.

The cloud computing era is the useful precedent. When cloud arrived, the initial story was that it would destroy Microsoft and Oracle. Instead, Microsoft became one of the most valuable companies in history by adapting to cloud faster than anyone expected. The narrative of replacement gave way to the reality of absorption. HSBC is betting the same thing happens with AI. I think they're right, with one important caveat: the vendors that don't move fast enough won't get a second chance this time. The window is shorter.

Jamie walked past my desk today and asked if I had "any thoughts on the SaaS situation." I said yes. A lot of thoughts. He nodded like he'd heard of it and kept walking. Classic.

Linda mentioned that Gerald has been asking about whether their retirement accounts have software stocks in them. Which is how you know a market event has become a real conversation and not just a finance thing. When Gerald is worried, it's news.

The SaaSpocalypse is real as a name, partially real as a threat, and significantly overstated as a death sentence. The businesses that understand the difference between those three things are the ones that come out of this period in better shape. The ones that either panic and over-rotate, or ignore it completely and assume nothing will change, are going to have a harder time. That's not hedging. That's just what the data says.

Now if you'll excuse me, I need to go use the word in another three conversations before end of business.