The fears of SaaS’s Demise due to AI Are Greatly Exaggerated
Will AI Kill Off SaaS?
There’s a growing chorus in tech claiming that advances in AI Coding tools and AI development assistants will flip the “buy vs. build” equation on its head. Some have even heralded “the end of SaaS” in this new era. The logic goes like this: If generative AI and drag-and-drop platforms make it easy for any company to build software, why would businesses keep paying for SaaS subscriptions? Why not have your team (or an AI) create a custom CRM or HR system tailored exactly to your needs?
It’s an attention-grabbing idea: SaaS is dead; every company will now become a software company. But is that really what’s about to happen? Just because you can build something custom now, doesn’t mean you should, or that most companies actually will. In fact, there are plenty of reasons to be highly skeptical that AI will spell the end of SaaS for the vast majority of businesses. Let’s break down why the “SaaS is doomed” narrative is largely hype, and why off-the-shelf software isn’t going away anytime soon.
Core Competence Over DIY Ambition
Most businesses succeed by focusing on their core competencies, the products, services, and innovations that make them competitive, and offloading the rest. Running great software systems is crucial, yes, but for a bank or a retailer, that’s a means to an end, not the end itself. SaaS exists to let companies offload complexity, things like maintenance, compliance updates, security, and integration headaches, to specialized providers, so the company can focus on what it truly excels at. In other words, if you’re, say, a healthcare firm, your time is better spent developing better patient services, not coding yet another customer management tool from scratch.
AI can crank out code, and low-code platforms let non-engineers create apps. But owning your code isn’t the same as owning your customers or outperforming your competitors. A custom CRM that saps your team’s time and budget won’t automatically give you an edge with customers. In fact, it might distract you from serving them. That’s why for decades the default decision was to buy, not build software for common needs, and that rationale isn’t magically gone. Many industries using vertical SaaS (think construction, hospitality, nonprofits, etc.) don’t even have the technical teams to leverage AI development tools in the first place. These companies are often still modernizing from spreadsheets (or pen and paper) to cloud software. The notion that they’ll suddenly start churning out sophisticated internal apps is pretty far-fetched.
The Hidden Challenges of Building It Yourself
Proponents of the “build it internally” trend often gloss over a big fact: Building great software is about a lot more than writing code. Sure, today’s AI copilots can help generate code snippets or even entire app scaffolds. But what about the whole lifecycle of a production software product used by hundreds or thousands of employees or customers? Here’s where the realities set in:
Depth of Features & Best Practices: Enterprise SaaS products, such as Salesforce or Workday, have been refined over many years. They embody countless features and industry best practices learned from serving thousands of customers. Rebuilding such depth isn’t a weekend project for an AI; it’s an endless journey. SaaS vendors specialize in their domain and ship improvements faster than most in-house teams ever could.
User Experience (UX): Designing an intuitive and efficient UX is challenging. SaaS companies invest heavily in UI/UX research because a better user experience is their competitive advantage. An internal app built by a small team is unlikely to match the polish of a market-leading SaaS that’s been refined through feedback from millions of users. Poor UX can tank adoption. Employees will resist using a homegrown tool that’s clunky or buggy, gravitating back to familiar commercial solutions.
Maintenance & Upgrades: Software is never “done.” Who will patch your custom app when bugs inevitably arise? Who will update it for new regulations or when your business processes change? In the SaaS model, the vendor handles all that. You wake up to new features and fixes on a routine basis. When you build in-house, you own the ongoing maintenance burden. That means dedicating engineering time indefinitely. It also introduces risk: internal tools can become outdated or insecure if not diligently maintained.
Security & Compliance: In an age of relentless cyber threats and evolving data privacy laws, security is a massive responsibility. Reputable SaaS providers have dedicated security teams, undergo regular audits, and comply with standards such as SOC 2 and GDPR. If you take things in-house, you assume all those obligations. You’ll need to implement proper authentication, encryption, access controls, audit logging, and continuously update them as new vulnerabilities and regulations emerge. Many organizations simply lack the in-house security expertise or resources to match what a focused vendor can do.
Integration: Modern business tech stacks are a web of integrations. Your CRM likely integrates with your email, analytics tool, finance system, and other systems. Top SaaS platforms come with rich APIs and pre-built integrations or an ecosystem of third-party plugins. A custom-built solution starts as an island. You’ll have to integrate it into everything else. That’s doable, but it’s a significant effort. Each integration is another thing to maintain whenever other apps update their APIs. With off-the-shelf software, much of this connectivity is handled or supported by the vendor and community.
Support & Reliability: When using a SaaS product, you typically have support contracts or SLAs. If something breaks, you call the vendor. If your internally built system goes down at quarter-end or loses data, guess who is on the hook? You are. That means on-call engineers for your “side” software project, and fire-fighting in the middle of the night for an app that isn’t even your core business. And what if the key developer who built your internal tool leaves the company? There’s a knowledge gap that can be painful. With a SaaS, you don’t care which engineers come and go at the vendor. The service continuity is their problem.
Building software is the easy part; building enterprise-grade software and running it smoothly at scale is the hard part.
Data Check: SaaS Isn’t Dying, It’s Evolving
If AI were indeed sounding SaaS’s death knell, we would expect to see companies massively cutting SaaS spending. In reality, SaaS usage and spending are still rising. Gartner projected SaaS end-user spending to grow about 20% in 2024 to reach $247 billion. And despite tighter budgets in 2023-24, global SaaS spend still grew by ~17.9% over a 12-month span.
At the same time, the low-code/no-code market is also growing, expected to hit around $32 billion by the end of 2024. However, note the order of magnitude difference: the SaaS market is approximately an order of magnitude larger. Companies are indeed adopting low-code tools (Gartner noted that by 2024, upwards of 75% of enterprises were using multiple low-code platforms for app development). However, they’re mostly using them to augment their operations with niche apps and automations, not to replace mission-critical enterprise systems wholesale. The rise of low-code is producing lots of small, specific applications inside organizations (think custom forms, simple workflows, department-level tools). It’s not producing thousands of full-blown alternatives to Salesforce or SAP overnight. In fact, many of those citizen-developed apps still rely on the data and foundation provided by core SaaS platforms.
The 10% (or Less) Who Should Build
To be fair, none of this is to say no company should ever build internal software with AI. There are absolutely cases where it can make sense, and those will likely increase somewhat with improved tools. If a company has truly unique processes or niche needs that off-the-shelf software can’t meet, an internal build can be justified. Additionally, organizations with already established strong engineering teams and a tech-centric culture (such as large banks or Fortune 500 companies with sizable IT departments) may be able to sustainably develop certain internal systems.
Startups, too, might choose to build a lightweight tool instead of paying for an expensive enterprise product, especially if the available SaaS is overkill.
There will be selective victories for the build approach in the coming years. Particularly, internal tools that wrap around existing systems or address very company-specific workflows. These are prime candidates for AI-assisted development, since they don’t need to be commercial-grade products with broad scope.
Conclusion: SaaS Isn’t Dead, Just Because You Can Build It Doesn’t Mean You Will
The bottom line: AI development tools are exciting, they lower barriers and will undoubtedly enable more software to be created by a wider range of people. Some businesses will take advantage of this to craft bespoke apps and maybe replace a few SaaS subscriptions. But the idea that this spells doom for SaaS as a whole is overblown. Companies have finite time, talent, and attention. For the lion’s share of needs, it’s still going to make sense to buy the solution and get back to doing what your company is actually in business to do.
Rather than the end of SaaS, we’re more likely to see an evolution of SaaS. The future is one where SaaS platforms incorporate AI and offer more flexibility, and businesses integrate a few custom-built pieces where it truly counts. It’s a hybrid model: use off-the-shelf for what’s standard, build in-house for what’s unique.
