The Moat Is Dead. Long Live the Moat.

The Moat Is Dead. Long Live the Moat.

On February 3rd, 2026, the cloud software sector had its worst single-day performance in years, with the S&P 500 Software Index dropping 5.7%. More than $800 billion in market cap evaporated from companies that spent the last decade building what seemed like unassailable positions in enterprise software. The latest victim? Figma dropping 7%after Anthropic released Claude Design.

The markets have been terrified that SaaS is dead, with the latest shock being Figma. I’m here to tell you it’s far from the grave. What’s getting a complete overhaul is how SaaS companies win.

I’m Ethan, a first-year in Kellogg’s MBAi program, a joint degree with McCormick’s engineering school focused on marrying traditional business with AI. Before this, I worked in R&D and design across product and hardware and software engineering in both the aerospace and consumer tech categories. The intersection of technology and strategy is something I think about constantly. This is my take on what’s happening.

The SaaS playbook over the past decade was plug and play: code software that solves a business problem, wrap it in a subscription, and start printing money. Popular wisdom was that the moat was the code, made as complex and customized as possible. Today, that moat is shrinking quickly. Retool’s 2026 Build vs. Buy report found that 35% of enterprise builders had already replaced at least one SaaS tool with a custom build, and 78% expected to build more this year. The transition from “Software as a Service” to “Service as Software”, where AI agents execute the work and you pay for outcomes instead of subscriptions, is driving the “seat compression” that spooked the market.

The market is looking at these signals and concluding that SaaS is dead, but these companies aren’t going to vanish. Software subscriptions will still exist. The difference is that it’s time to come to terms with the fact that code was never really the moat. It just felt like one because writing it was so expensive.

If code isn’t the moat, then what is? As we look forward into the new AI era, I think these three differentiators will hold up and can’t be vibe-coded in a weekend:

1. Trust and Accountability

Trust is a combination of three ingredients: expertise (knowing what you’re doing), risk absorption (being willing to own the outcome), and differentiated data (proprietary information). All of these take years to collect and create a flywheel that keep customers coming back.

Expertise is the most obvious one, and it’s the one that AI is chipping away at fastest. Consider SOC 2 compliance. AI platforms are compressing the required prep work from months to days. AI, however, can’t replicate the judgment calls that come from experience: which risks actually matter for your business, how regulations get interpreted, and the gut feeling that only comes from seeing hundreds of audits go sideways. A startup using an AI compliance tool will get 80% of the way there. The last 20% is where the real risk lives, and that’s where expertise earns its premium.

Risk absorption is what turns expertise into a moat. Knowing the space is one thing, but being willing to publicly own the outcome is another. That’s why companies like Stripe and SAP will continue to exist even as cheaper alternatives emerge. When money is involved or a missed shipment can trigger legal liability, companies don’t want to be the ones holding the bag. They go with someone who will. I think any SaaS company that handles money, compliance or legal consequences has a stronger position than they realize right now, but only if they lean into accountability rather than just feature development.

Data is the ingredient that compounds over time and helps foster the previous two. A compliance firm that’s seen thousands of audits has information no foundation model can match. A fintech company processing millions of transactions has fraud signals that only exist because of the volume they’ve already handled. A competitor can spoof the code, but they can’t fake the historic data baked into the model.

The companies that win here are selling their willingness to deliver on the outcome every time, not software.

2. Operations and Systems

The second moat is the operational infrastructure SaaS companies have built around their software: the physical and organizational systems that are much harder to replicate than the code itself.

Think about what makes a company like SAP hard to leave. What locks you in is the years of customization, the business processes built on top of it, and the organizational knowledge embedded in the system that goes undocumented. Moving off SAP requires an army of consultants, years of work, and hundreds of thousands of dollars. That’s an operational moat, not a software one.

Beyond the software itself, there’s on-prem systems, hybrid deployments, compliance environments, implementation services, and customer success teams that spend months configuring the product to a client’s specific workflows. Salesforce’s AppExchange has 6,000+ apps built on top of it. That ecosystem represents operational depth that took years to build and would take years to replace. This compounds when considering running these operations at scale. A vibe-coder can ship a CRM in a weekend, but they can’t provide the comprehensive ecosystem surrounding it.

The broader pattern confirms this. An MIT study from 2025 found that 95% of generative AI pilots deliver no measurable return because the companies weren’t operationally ready. The companies that invested in operational excellence before the AI wave (clean data, standardized processes, disciplined governance, etc.) are the ones capturing value from it. Everyone else is running pilots that go nowhere because they feel like they have to add “AI” to their offering.

In a world where anyone can build software, the companies that have built real operational depth around their product will be the ones that last.

3. People and Networks

The third moat is the oldest one in business: network.

Google may be one of the best companies positioned to win the AI foundation model race. Not because they published “Attention Is All You Need”, nor because they’ve been building custom AI chips for over a decade (while those are still going to be vital to their success), but because of their billion-person install base. When you can deploy Gemini inside Search, Chrome, Workspace, and Android simultaneously, distribution becomes the moat. The model isn’t the differentiator. The install base is.

A marketplace’s real value lives in the network of buyers and sellers, not the code connecting them. You can’t build a buyer-seller network with a prompt. Take NVIDIA, a company with a moat that goes far beyond the chips themselves. It’s CUDA and the 4+ million developers building on it. That’s why AMD, which already fabs competitive hardware, still struggles at enterprise scale. You can’t vibe-code a developer ecosystem any more than you can vibe-code a semiconductor fab.

People go to these companies because they trust them to deliver, and that trust compounds through networks the same way data does. You can’t replicate that by throwing more compute at the problem.

The Moat Was Never the Code

The bygone SaaS era trained us to think of software as a moat. It was a temporary advantage maintained by the difficulty and expense of writing and maintaining code at scale. AI removed that constraint, and now we’re watching the repricing happen in real time.

The companies that thrive won’t be the ones clinging to their codebase. They’ll be the ones who realized early that code was never the point.

The moat is dead. Long live the moat.