Anyone who has ever sat through a Q4 vendor renewal negotiation for a sprawling enterprise system knows it is an elaborate game of chicken. The vendor knows how painful a migration would be, and the enterprise knows how much the vendor relies on their recurring revenue. For anyone who has spent years in the IT trenches untangling complex systems, the dynamic rarely changes: the vendor pitches a 5% to 10% price hike, the client threatens a multi-year migration to a competitor, and they eventually settle somewhere in the middle.
The Global Intelligence Crisis
A recent piece of speculative fiction by Citrini Research titled “The 2028 Global Intelligence Crisis” models a scenario where this delicate balance is completely obliterated by AI.
The piece envisions a terrifying macroeconomic feedback loop triggered by agentic coding. In their 2028 scenario, AI gets so good that companies realize they can build in-house rather than paying for enterprise SaaS. Margins collapse across the tech sector, white-collar layoffs ensue, and we enter an “intelligence displacement spiral” resulting in 10.2% unemployment and a 38% drawdown in the S&P 500.
When reading it, the immediate architectural instinct is to roll your eyes. When your mission is solving the puzzle of legacy modernization and codifying system architecture, you know exactly what happens when you underestimate the complexity of enterprise software. You don’t just “prompt” your way to a highly available, SOC2-compliant, edge-case-handled enterprise system.
But buried in the Citrini piece - and brilliantly validated in its comments section - is a deeply cynical, highly plausible mechanism that makes their economic doomsday scenario totally viable.
The AI doesn’t actually have to work to destroy the market. The threat of it working is enough.
The AI Threat
Citrini outlines a hypothetical 2026 scenario where a Fortune 500 procurement manager is negotiating a software renewal. Instead of playing the usual game of chicken, the manager casually mentions they’ve been speaking with OpenAI about using “forward deployed engineers” to replace the vendor entirely with an AI-built alternative. The SaaS company, terrified of this new invisible competitor, panics and renews at a 30% discount.
This isn’t science fiction. It is the reality of boardroom psychology.
A commenter on the piece, who works in the Cloud and AI division of a Mag7 tech giant, confirmed this exact dynamic. They pointed out that current AI capabilities in software engineering are actually “nowhere near the current hype”. If a company actually fired their vendor and tried to replace a complex system with a “vibe-coded” AI prototype, it would likely be a “devastatingly bad decision”.
But here is the kicker: some executives and procurement managers are working off very simplistic models and genuinely believe AI is magic.
We have seen this movie before. Board members see a slick demo of an LLM spinning up a Python script in seconds, and they immediately extrapolate that it can replace a sprawling, tightly integrated SaaS platform. They take that belief into the boardroom, and they use it as a weapon. Even if the internal build would fail spectacularly in production, the SaaS salesperson sitting across the table doesn’t know that. The belief that a cheaper, AI-generated alternative is viable is a strong enough negotiating chip to fundamentally crush software business models.
The AI Hype Cycle
In the short term, the economic shockwaves Citrini describes are highly possible. Not because we have achieved a technological singularity, but because corporate leverage has fundamentally shifted. The suits are going to look at their massive annual software renewals, look at the AI hype cycle, and demand ruthless discounts or pull the trigger on reckless in-house builds.
The immediate result will be collapsing vendor margins and a wave of preemptive white-collar engineering layoffs as companies try to fund these “productivity initiatives”. The market will feel the pain.
But what happens when these same companies actually try to put their cheap, AI-generated prototypes into production? What happens when the reality of tech sprawl, security, and un-codified architecture rears its head? That is a very different story, and I’ll be diving into the inevitable “Vibe-Coded” Post Hype-Cycle Apocalypse in my next post.