AI Needs SaaS: Why the Extinction Narrative Misses the Point

Here’s something interesting we’ve been noticing.

Every few technology cycles, markets confuse pressure with collapse.

Market snapshot
Salesforce and Adobe have each fallen roughly 20–25% over the past year as investors price in AI risk, even as revenues continue to grow and AI features ship across their platforms. A new capability arrives, valuations wobble, headlines sharpen, and suddenly an entire category is declared obsolete. In the past year, artificial intelligence has played that role for enterprise software. The story goes like this: AI agents will replace apps, seat‑based pricing will crumble, and the SaaS model, one of the most durable business constructs of the last 20 years, will quietly expire.

It’s a compelling narrative. It’s also incomplete. If you look past stock charts and slogans and into how real organizations are actually deploying AI, a different pattern emerges. AI isn’t demolishing SaaS. It’s leaning on it. Hard. At least for the next several years, the most credible path forward isn’t AI versus SaaS. It’s AI plus SaaS.

This isn’t a defense of incumbents for nostalgia’s sake. It’s an argument grounded in behavior: how frontier AI labs operate, how the world’s largest enterprises spend money, and how work actually gets done inside complex organizations. When you connect those dots, the extinction narrative starts to look less like foresight and more like fear.

Markets Are Scared, Not Demolished

Let’s start where the panic is loudest: public markets.

Over the past year, flagship SaaS companies like Salesforce and Adobe have seen roughly 20–25% of their market value shaved off.

“Investors have marked major SaaS names like Salesforce and Adobe down ~20–25% on AI fears, but the underlying businesses and products are very much intact.” Analysts have been explicit about the reason. Investors worry that generative AI will undercut traditional software pricing models, automate away seats, and compress long‑term growth. Software stocks, as one commentator put it, “haven’t been able to shake the narrative of AI disruption.”

But narratives aren’t balance sheets.

Salesforce is still growing. Adobe is still deeply embedded in creative workflows across industries. Workday, ServiceNow, HubSpot, and others continue to expand product lines, ship AI features, and sign multi‑year enterprise contracts. What’s changed is sentiment, not utility.

This distinction matters. A 25% drawdown reflects uncertainty about future pricing power, not evidence that customers are ripping out systems en masse. Historically, markets often punish categories before operational reality catches up, sometimes rightly, often prematurely. Right now, SaaS looks less like a dying model and more like one being repriced under a new set of assumptions.

And assumptions are precisely where the AI‑kills‑SaaS story starts to fray.

When AI’s Own Builders Buy SaaS

If AI were genuinely making SaaS irrelevant, you would expect the most AI‑native organizations in the world to behave accordingly. They don’t.

Take Anthropic, one of the leading frontier AI labs, and the creator of Claude. This is a company with deep technical talent, strong incentives to build internally, and firsthand exposure to what modern AI agents can do. When Anthropic needed to scale customer support, it faced a classic build‑versus‑buy decision.

It chose to buy. Anthropic adopted Intercom’s AI agent, Fin, which resolved 50.8% of support conversations and saved about 1,700 hours in its first month.

Instead of building an in‑house support agent, Anthropic adopted Intercom’s AI‑powered agent, Fin. The stated rationale was speed to value and focus: staying concentrated on core research rather than reinventing a mature workflow. Within the first month, Fin resolved just over half of all support conversations and saved roughly 1,700 hours.

That detail is easy to gloss over, but it’s telling. An archetypal AI company didn’t replace SaaS with bespoke agents. It embedded AI inside SaaS to move faster.

Anthropic’s broader stack tells the same story.

Even a frontier AI lab still runs on enterprise SaaS like Workday, Salesforce, and NetSuite for HR, finance, and compliance. Job postings and internal materials reference systems like Workday, Salesforce, and NetSuite for HR, finance, and compliance functions. These are not accidental choices. They are systems of recordboring, trusted, heavily regulated, and deeply entrenched.

Even culturally, the point shows up in unexpected ways. When an Anthropic design lead jokingly lamented the “inevitable switch to enterprise‑grade internal tools no one likes using (see: Workday),” the subtext was unmistakable: dislike doesn’t equal dispensability. Some tools persist precisely because the alternativerolling your own compliance, payroll, or financial controlsis far worse.

If the future were AI agents wiping out SaaS, Anthropic would be the first to prove it. Instead, it’s quietly demonstrating the opposite.

The Reality of Enterprise AI

Now zoom out from AI labs to the largest enterprises in the world. JPMorgan Chase spends roughly $18 billion a year on technology.

Enterprise scale
JPMorgan reports 450+ AI use cases in production and targets $1.5–$2 billion in annual value from AI by connecting it into existing enterprise systems. It has more than 450 AI use cases already in production and has publicly estimated that AI could generate $1.5–2 billion in annual business value. If any organization had both the incentive and the capacity to rip out legacy platforms in favor of AI‑native replacements, it would be JPMorgan.

That’s not what it’s doing.

Executives describe the strategy as building an “AI‑connected enterprise.” Translation: AI layered on top of existing systems, not wholesale replacement. In private briefings, JPMorgan leaders have indicated no immediate plans to use AI to eliminate traditional enterprise applications. In some cases, they expect to increase spending with incumbent vendors as those vendors roll out AI‑enabled features.

This approach isn’t conservative. It’s rational. In private briefings, JPMorgan leaders have indicated no immediate plans to replace core enterprise platforms with AI, and expect spending with some incumbents to increase as AI features roll out.

Large enterprises don’t optimize for novelty. They optimize for risk, continuity, and compliance. Replacing core platforms like HR, finance, CRM, or service management isn’t just a software decision; it’s an organizational upheaval. AI promises leverage, but leverage applied to unstable foundations tends to amplify problems rather than solve them.

So JPMorgan, like many enterprises, is choosing augmentation over annihilation.

Why AI Needs SaaS (For Now)

“In the real world of business, SaaS remains indispensable. AI fits into it, clings to it, feeds off it. It cannot exist without it.”
— Bertrand Duperrin, digital workplace and enterprise technology thought leader, July 2025

The common thread across these examples points to a set of structural realities that are easy to ignore in abstract debates and impossible to ignore in practice.

1. SaaS Is the System of Record

AI agents are impressive at reasoning, synthesis, and interaction. They are far less impressive at governance.

Systems like Workday, Salesforce, ServiceNow, and core financial platforms are where permissioned data lives. They encode local labor laws, audit trails, security policies, approval workflows, and compliance logic accumulated over decades. This is not glamorous software. It is essential software.

For AI to act meaningfully inside an organization, approving time off, issuing refunds, modifying customer records, and triggering payroll changes, it needs a trusted system of record beneath it. Today, that system is almost always SaaS.

2. Distribution Beats Reinvention

Another pattern is emerging: AI tools are becoming front doors, not replacements.

Chatbots and copilots increasingly sit on top of existing applications, pulling data from tools like Figma, Canva, Asana, Office, or CRM systems. Anthropic’s recent demonstrations of Claude interacting directly with workplace apps underscore this shift. The value comes from orchestration, not obliteration.

From a user’s perspective, this feels transformative. From an architectural perspective, it’s additive.

3. Incumbents Have Trust Capital

Trust is an underappreciated moat.

Enterprise buyers don’t just purchase features; they purchase assurances. Data residency, uptime guarantees, regulatory compliance, and vendor accountability matter more as AI becomes more powerful, not less. Established SaaS vendors already sit inside procurement, legal, and security frameworks. New AI‑native tools, no matter how capable, still have to earn that trust.

This is why leaders like Workday’s CEO have framed AI as a tailwind rather than a threat.

“AI is a tailwind for us, not a headwind. Our incumbency and the trust customers place in us uniquely position us to win in the enterprise.”
Carl Eschenbach, CEO, Workday Inc., in regulated workflows, is not inertiait’s leverage.

The Real Threats (And They’re Not Imaginary)

None of this is to say that SaaS is immune.

Revenue growth across many enterprise software companies has slowed relative to cloud infrastructure and database providers. Pricing models will be pressured as AI automates tasks previously tied to seat counts. Competition is intensifying as nearly every vendor markets some form of “general‑purpose agent.”

There is also a genuine risk that platforms like ChatGPT, Claude, or Copilot become the primary interface through which knowledge workers interact with software. If that happens, the locus of value could shift upward, away from applications and toward orchestration layers.

But notice the nuance: interface dominance does not automatically imply backend displacement.

Just as browsers didn’t eliminate operating systems, and APIs didn’t eliminate databases, AI interfaces are more likely to reorganize where value is captured than to erase the need for underlying systems entirely.

Build Versus Buy, Revisited

Every technology wave revives the same debate: should companies build custom tools or buy off‑the‑shelf software?

AI lowers the cost of building. That’s real. Startups will proudly announce that they’ve replaced legacy SaaS with internal tools stitched together by agents. Some of those stories will be true, especially at a small scale or in narrowly defined workflows.

But as organizations grow, the calculus shifts.

Industry voice
Saumil Mehta, global president at Ticketmaster, put it plainly: spending scarce developer time to replace enterprise apps is rarely worth the opportunity cost to the core business.

Developer time is finite. Attention is scarce. As Saumil Mehta of Ticketmaster put it, there are far more valuable uses of engineering talent than saving marginal dollars by rebuilding enterprise plumbing. Opportunity cost, not license fees, becomes the binding constraint.

This is why, even in an AI‑rich future, the build‑versus‑buy needle continues to point toward buy.

A More Plausible Thesis

So where does that leave us?

Despite a meaningful drawdown in SaaS valuations driven by AI fears, there is little empirical evidence that AI is making enterprise SaaS irrelevant in the near term. The most advanced AI labs are heavy SaaS users. The world’s largest banks are layering AI on top of existing stacks. And the most capable AI agents still depend on SaaS for data integrity, compliance, and execution.

Over the next several years, AI needs SaaS more than SaaS needs AI.

That balance may change. Business models will evolve. Interfaces will shift. Value will migrate. But extinction narratives rarely survive contact with reality, and reality, right now, looks far more hybrid than headline‑driven.

The future of work isn’t a clean break from what came before. It’s a negotiation between new capability and old constraints. And in that negotiation, SaaS remains very much at the table.

At Camino5, we spend our time mapping how emerging technologies actually intersect with consumer behavior, enterprise systems, and decision‑making. If you’re navigating these shifts and want clarity instead of hype, that’s a conversation worth having.

Ryan Edwards, CAMINO5 | Co-Founder

Ryan Edwards is the Co-Founder and Head of Strategy at CAMINO5, a consultancy focused on digital strategy and consumer journey design. With over 25 years of experience across brand, tech, and marketing innovation, he’s led initiatives for Fortune 500s including Oracle, NBCUniversal, Sony, Disney, and Kaiser Permanente.

Ryan’s work spans brand repositioning, AI-integrated workflows, and full-funnel strategy. He helps companies cut through complexity, regain clarity, and build for what’s next.

Connect on LinkedIn: ryanedwards2