Is SaaS Dead?, Here's a Senior Engineer's Perspective

"SaaS is dead" became a talking point after Microsoft CEO Satya Nadella used the phrase on the BG2 podcast in December 2024: if AI agents can take over the interface and the work, what is left of a subscription application? In his video, senior engineer Arjan from the ArjanCodes channel argues the opposite for a large part of the market — that the value of many SaaS products sits in work that has nothing to do with the code. This post summarises his argument and adds the figures and documents it leans on.

More Than Just Code: The Real Moat of SaaS¶

Arjan's position is that the debate over AI replacing SaaS usually ignores the operational and regulatory work these platforms absorb. An agent can reproduce the functionality of a simple application; it cannot reproduce years of accumulated compliance obligations, integrations and partnerships. The "moat", on his reading, is the messy invisible part.

The Unseen Complexity: Real-World Examples¶

1. Accounting and Financial SaaS¶

An accounting system looks like a series of calculations, but it is a stack of moving obligations: tax codes per country, state and municipality, security requirements for financial data, and connections to banks and payment rails. That is maintenance, not a one-off problem. Stripe's own product catalogue illustrates the point: Stripe Tax exists solely to "automate sales tax, VAT, and GST compliance on all your transactions" — a product whose reason to exist is that tax rules keep changing.

2. Web Hosting and Infrastructure SaaS¶

Spinning up a server is easy; running one reliably is not. Squarespace, the video's sponsor, documents DoS/DDoS mitigation, a fully-managed cloud platform serving billions of monthly views and 99.9 percent uptime. (An earlier version of this post repeated a 99.99 percent figure; Squarespace's own support documentation states that "no platform can guarantee 100% uptime" and that its public commitment is 99.9 percent.) Content delivery, traffic spikes and attack mitigation are operational work that does not disappear when the code can be generated.

3. Payment Providers¶

Payment providers such as Stripe and Adyen sell trust and regulation rather than transaction processing. The basis for that is PCI DSS, maintained by the PCI Security Standards Council to "encourage and enhance payment card account data security" — and mandatory in practice for anyone handling card data. PCI DSS applies to every entity that stores, processes or transmits cardholder data, including merchants, processors, acquirers, issuers and service providers, and Adyen's own compliance guide notes that although the standard is not law, "the card schemes" mandate it globally with significant penalties for non-compliance. The certification itself is graded: Stripe states it is certified to PCI Service Provider Level 1, "the most stringent level of certification available in the payments industry".

Two details of that burden are easy to underestimate. First, the standard keeps moving: PCI DSS v3.2.1 was retired on 31 March 2024, and of the 64 new requirements in v4.0, 51 were future-dated and only became mandatory on 31 March 2025. Second, fraud work is measured in money, not features: Stripe says it prevents more than 500 million dollars in payment fraud for its merchants every month.

4. Music and Media Services¶

Streaming is a licensing business with a hosting front. Spotify's own royalties guide describes how money moves through labels, distributors, publishers and collective management organisations — PROs and mechanical agencies — before reaching artists and songwriters, split by territory and by performance versus mechanical rights. The scale of what is being administered: Spotify says it has paid out nearly 70 billion dollars to artists, songwriters and rights holders to date, including 10 billion dollars in 2024, spread across a long tail — 12,500 artists generated more than 100,000 dollars on the platform that year. No AI agent reproduces those contracts.

What Does the Future Hold for SaaS?¶

Arjan's conclusion is that SaaS is not dying but evolving: AI gets integrated into existing platforms, micro-SaaS becomes cheaper to build for small teams, and the regulated, integration-heavy layer remains hard to displace with code alone.

The market data available so far does not show a collapse:

Gartner's forecast, reported by CIO Dive and TechRadar, put SaaS spending at 250.8 billion dollars in 2024 rising to 299.1 billion in 2025, within total cloud spending of 723 billion.
Synergy Research measured the 2024 cloud infrastructure market at 330 billion dollars, 60 billion more than 2023, with generative AI driving about half of that growth.
IDC attributes the "SaaS is dead" line to Nadella's podcast appearance, and its own verdict is that SaaS "is being disrupted, not by decline but by evolution"; it predicts that seat-based pricing will be obsolete by 2028, with 70 percent of vendors repricing around consumption or outcomes.