The “AI has killed software” narrative has a few loud proponents and much silent evidence against it. The companies that will survive the next five years won’t worship hyperscalers as new deities.
I always research before making claims to avoid sounding like a LinkedIn post. I wish more people in this field did the same, as there’s a widespread attitude that sees big numbers as the complete story.
When the Black Death struck, people likely thought it was the end. In times of war, people thought it was the end. Yet, we naturally overcome obstacles and turn change to our advantage.
When AI began infiltrating work and personal lives, many declared, “AI will replace people,” believing this technology would dominate our minds, hearts, and work.
Yet we still work; people still write, think, create, and build.
In recent years, more people have claimed “SaaS is dead.” This phrase originated from someone influential and soon everyone was ready for its funeral.
In August 2024, Klarna’s CEO, Sebastian Siemiatkowski, mentioned in an earnings call that Klarna “shut down Salesforce.” Workday was next.
Klarna planned to build its own AI-driven alternatives, news that moved markets. Articles proclaimed the death of SaaS. Salesforce’s Marc Benioff, at Dreamforce, faced a customer believing AI was the future, making him visibly embarrassed.
Six months later, Siemiatkowski clarified that Klarna hadn’t replaced Salesforce with AI, but with other SaaS solutions: Deel for HR, third-party CRM tools, and Neo4j for data.
Klarna still uses Slack, a Salesforce product. Siemiatkowski admitted on X being “tremendously embarrassed” by the story’s spread.
“No,” he wrote, “we did not replace SaaS with an LLM.”
This revealing story highlights the gap between saying and doing in the “SaaS is dead” narrative. The headline gained attention; the correction didn’t.
An industry of analysts, VCs, and model CEOs built a year of marketing on the louder half.
Who benefits from claiming software-as-a-service is replaced by AI? The answer is narrow. Hyperscalers do, as AI workloads justify $660 to $690 billion in capital expenditure for the five largest US cloud and tech companies by 2026, according to Futurum Group analysis, nearly double the prior year.
Foundation model labs benefit because redirecting enterprise software spend to their APIs supports their valuations. OpenAI reached $20 billion in annual recurring revenue by 2025. Anthropic hit $9 billion in January 2026. Though large, these are only a small percentage of hyperscaler capex catering to them.
Venture capitalists benefit for portfolio repricing depends on the narrative of AI-native companies outpacing incumbents they once funded. Nvidia benefits until it doesn’t.
In March 2026, CEO Jensen Huang confirmed his investments in OpenAI and Anthropic would likely be the last. The circular cycle—Nvidia investing in OpenAI, OpenAI buying Nvidia chips—reached a point where the chipmaker stopped calling it a virtuous cycle.
MIT’s Michael Cusumano, quoted by Bloomberg, bluntly stated