Is AI the Beginning of the End for large-scale SaaS?

Is AI is coming for big SaaS?

Over the past two decades, I’ve watched SaaS evolve from a disruptive innovation into the dominant model for business software. The pitch was irresistible: lower upfront costs, faster time to value, continuous updates, and the ability to scale without needing to manage infrastructure.

And for a long time, it delivered.

But I’ve started to wonder. Have we hit the limits of the one-size-fits-all SaaS model?

Platforms like Salesforce, Workday, and ServiceNow have become the digital backbones of entire departments. They promise flexibility, but in practice, they require companies to adapt to their way of doing things. Real customization often demands expensive consultants, long timelines, and ongoing effort to maintain integrations, workflows, and compatibility with frequent platform updates.

Instead of streamlining operations, these platforms increasingly dictate how we work.

And of course, they’re not cheap. Globally, SaaS now generates over $273 billion in annual revenue and is projected to exceed $317 billion in 2024 (Zylo, Vena Solutions).

Meanwhile, corporate complexity is growing. According to recent research, mid-sized companies now use up to 335 different SaaS applications, while enterprise firms average up to 473 (Spendesk). The result is a tangled web of tools, APIs, and disconnected workflows—each solving a narrow problem, but collectively creating friction, fragmentation, and rising overhead.

SaaS applications used by mid-market and enterprise companies

I’m not saying these platforms are going away anytime soon. But I do believe we’re past their growth phase and possibly at the start of something very different.

Here’s my theory. AI is going to change the math.

Thanks to large language models, intelligent agents, and AI-assisted low and no-code tools, it’s now possible for companies to build solutions tailored to their processes faster and more affordably than ever before. What once required a team of engineers and months of development can now be achieved by a small team with the right tools and vision.

Rather than renting rigid platforms designed for the masses, companies may increasingly build proprietary systems that fit like a glove, designed around their unique workflows, goals, and culture.

Not only does this offer better alignment, but it also means owning the roadmap, the IP, and the competitive advantage.

And don't just take my word for it. Investors are already following the trend. In 2024, AI startups attracted $131 billion in global VC funding (a massive 52% increase year over year) and accounted for 46.4% of all venture capital raised worldwide (FDI Intelligence, Reuters). That’s more than double AI's share from just a few years ago.

By contrast, enterprise software and SaaS companies pulled in $155 billion, representing 42% of total funding, still significant, but no longer the dominant force it once was (Sapphire Ventures, Buttondown). The momentum is clearly shifting.

And here's what’s really interesting. Employees are already doing this on their own. Tools like ChatGPT and Claude are quietly becoming part of people’s daily workflows. Without waiting for formal approval or IT buy-in, teams are stitching together more efficient ways of working with AI. It's a bottom-up movement, and it’s accelerating.

This shift won’t happen automatically. It requires product leadership. People who can identify pain points, translate business needs into user journeys, and manage the development of these AI-powered internal platforms.

If I’m right, we’re about to see growing demand for product managers who can operate outside the boundaries of vendor-defined roadmaps. PMs who can guide their teams to build smarter, leaner, and more relevant tools from the ground up. Not just optimizing features in someone else’s SaaS, but architecting the next generation of internal software.

In a strange twist, AI may bring us back to a more craft-driven era of software. Except this time, the tools are far more powerful and the timelines far shorter.

I don’t have all the answers here. This is a theory in progress, and I’m genuinely curious.

Are you seeing signs of this shift in your company or industry? Are teams starting to consider building instead of buying? Is AI really lowering the barrier to custom solutions, or is it just adding more complexity?

I’d love to hear your take.