Investing.com - Over the last month or so, markets have been gripped by a grim narrative: the “SaaSpocalypse.” As agentic AI surged, investors fled traditional software names like Salesforce, ServiceNow, SAP, Cadence, Synopsys, and even legacy giants like IBM, on fears AI agents would replace the need for software altogether.

In a wide-ranging interview with CNBC’s Rebecca Quick, NVIDIA (NASDAQ: NVDA) CEO Jensen Huang called that thesis not just premature, but fundamentally wrong.

His core argument is simple and deeply counterintuitive: AI agents won’t eliminate software tools; they’ll use them more than humans ever did.

A "Deep Misunderstanding" About Agentic AI

Huang frames today’s moment as an inflection point. For the first time, AI systems can perceive, reason, and do work. These “agentic” AIs don’t just answer questions; they execute tasks: writing code, planning trips, building websites.

That capability has spooked investors. If AI can write software, why would anyone need software companies?

Because, Huang says, that’s not how work actually happens.

Enterprises don’t want to rebuild the digital world from scratch. They want intelligent agents that can operate inside the world that already exists—the browsers, databases, workflows, and systems of record that companies have spent decades refining.

Why recreate Excel when you can just use Excel?

Agents Are Tool Users, Not Tool Killers

Huang’s mental model flips the SaaSpocalypse on its head. Agents are not replacements for software platforms; they are power users at machine scale.

An AI agent handling customer service won’t invent a new workflow engine; it will use ServiceNow. A design agent won’t reinvent chip tools; it will run Cadence and Synopsys. A finance or operations agent won’t bypass SAP; it will populate SAP more consistently than any human ever could.

In Huang’s words, “agents won’t replace the tools, but agents will use tools.”

And when you multiply that by scale, hundreds of thousands of digital employees working alongside tens of thousands of human ones, the implication is explosive: tool usage goes up, not down.

At NVIDIA itself, Huang already sees this happening. The number of compilers, scripts, and software instances is growing rapidly, not because there are more humans, but because there are more agents.

Systems of Record Still Matter, Maybe More Than Ever

Another fear baked into the sell-off is that AI will erode the importance of systems of record—the authoritative databases where enterprises store finalized information.

Huang argues the opposite.

Agents still need ground truth. They need places to read from and write to. Systems of record exist precisely because humans need clarity, accountability, and shared understanding. AI doesn’t eliminate that need—it reinforces it.

Agents will pull from these systems, act on our behalf, and then put information back in formats humans can trust. The platforms that own those systems don’t become obsolete; they become indispensable infrastructure for an AI-augmented enterprise.

The Robot-in-the-Kitchen Thought Experiment

To make the point tangible, Huang offers a simple analogy.

Imagine a future home robot. It’s highly intelligent. Does it invent a new microwave? Design a novel food processor? Of course not. It reads the manual and uses the tools already there.

Digital agents are no different. Software platforms are the microwaves, wrenches, and screwdrivers of the enterprise world. They exist for good reasons—and intelligent systems will rely on them, not replace them.

Not Replacing Humans - Changing the Abstraction Layer

Perhaps the most overlooked part of Huang’s argument is what this means for people.

Yes, agents will replace certain tasks humans do today—just as typing replaced handwriting. But the purpose of the work doesn’t change. Software engineers don’t exist to type code; they exist to solve problems and invent things that didn’t exist before.

In an agentic future, humans move up the abstraction stack. We describe intent. Agents handle execution. Productivity rises, creativity expands, and the surface area for innovation grows.

That doesn’t shrink the need for software or software engineers. It multiplies it.

The Market May Be Pricing Yesterday’s Fear

If Huang is right, the recent sell-off isn’t a rational response to disruption—it’s a misread of how disruption actually unfolds.

AI agents aren’t the end of SaaS. They’re the next great demand driver for it.

The real shift isn’t software versus AI. It’s software with AI, used by agents, at unprecedented scale.

And if that’s true, the so-called SaaSpocalypse may go down not as a warning—but as one of the market’s more memorable misunderstandings.