From London: as Australians slept last Sunday night, a 5,000-word essay on a finance newsletter was quietly detonating under global technology stocks. By the time markets opened in New York on Monday, billions of dollars had evaporated from the valuations of companies including DoorDash and Uber. The culprit was not a profit warning, a regulatory ruling, or a geopolitical shock. It was a Substack post by Citrini Research, the top finance writer on the platform, that had gone viral over the weekend.
Titled "The 2028 Global Intelligence Crisis," the essay by Citrini Research frames itself as a "macro memo from June 2028" and describes a world in which the S&P 500 has plummeted 38%, unemployment has spiked to 10.2%, and the US economy is trapped in a deflationary spiral caused by the mass displacement of white-collar workers by artificial intelligence. The paper was careful to label itself a thought exercise rather than a forecast, but the market treated it as something more.
The essay's central argument, as it relates to software, is pointed. Citrini predicts a collapse in private credit, forecasting that PE-backed software-as-a-service companies will default on billions in debt as AI coding agents allow clients to build internal software rather than pay subscription fees. In this imagined future, former Salesforce project managers end up driving for Uber. Incumbent vendors, stripped of pricing power, slash fees in a race to the bottom against a wave of AI-built alternatives. The SaaS-pocalypse, as the theory has been coined, is presented as a self-reinforcing spiral with no natural off-ramp.
The speed at which the essay moved markets is itself the most instructive element of this episode. It is fairly normal these days to see companies' stocks fall on news of AI's disruptive potential, but those cases typically involve real announcements of model capabilities. Citrini's post was a scenario for how some current business models could be disrupted. The question worth asking is whether professional investors with dedicated software sector mandates had genuinely not considered AI disruption risk until a blogger sketched out a fictional future mentioning their holdings by name. The answer, almost certainly, is that they had. What the essay did was provide a common Schelling point around which sentiment could crystallise and selling could coordinate.
The pushback from sober analysts was swift. Ken Griffin's market-making firm Citadel Securities swiftly challenged the viral narrative, with macro strategist Frank Flight systematically debunking Citrini's scenario using real-time economic data. Citadel pointed to Indeed job posting data showing that demand for software engineers was actually rising rapidly, up 11 per cent year-over-year in early 2026. Technological diffusion has historically followed an S-curve, where early adoption is slow, accelerates as costs fall, and eventually plateaus as saturation sets in and marginal returns diminish — a trajectory that leaves little room for the near-vertical disruption Citrini imagines across just two years.
Snowflake CEO Sridhar Ramaswamy, who has his own obvious stake in the outcome of this debate, nonetheless made a point that carries weight beyond self-interest. Speaking to investors, he argued that enterprise software incumbents hold an advantage that goes well beyond the cost of building code. Existing software companies counter that companies cannot simply replace their business suites with AI, pointing out that processes and policies are already built around the current IT stack, and that the burden of migrating not only data but the entire daily work of a company is enormous. Oracle, SAP, and Salesforce are not merely selling features. They are the repositories of years of transactional history, and the organisations that use them are — as any IT professional will confirm — deeply reluctant to move.
The Citrini Research essay is not entirely without merit, and intellectual honesty demands acknowledging that. One thing that emerges from the piece is the idea that it does not really matter whether it is feasible to reimplement the systems a SaaS vendor provides, as long as a buyer can credibly threaten to do so during a pricing negotiation. Even if building in-house would be a bad decision, the mere possibility functions as a negotiating chip. That is a real and underappreciated dynamic in enterprise purchasing. Margin compression is coming for pure-play SaaS vendors, even if extinction is not.
The broader structural tension is also genuine. The report argues that the US economy has spent fifty years building a massive rent-extraction layer on top of human limitations, with trillions of dollars in enterprise value depending on the fact that humans take time to make decisions, find price-matching tedious, and rely on brand familiarity to avoid the effort of due diligence. AI agents that never tire and hold comprehensive information will, over time, erode some of that friction. The honest answer is that nobody knows how much, or how fast.
For Canberra and Australian investors, the episode carries a specific relevance. Australian superannuation funds hold significant exposure to US technology equities, both directly and through index mandates. Enterprise software stocks bore the brunt of the selloff, with the iShares Expanded Tech-Software Sector ETF hitting a new 52-week low, down 5 per cent on the day and nearly 30 per cent year-to-date, erasing all gains since the ChatGPT launch in November 2022. Millions of Australians with retirement savings in balanced or growth options felt this, whether they read Citrini or not.
The Reserve Bank of Australia has previously flagged AI-driven productivity as a factor with uncertain implications for inflation and wages. The Australian Bureau of Statistics tracks technology employment trends that will, in time, offer local evidence on how quickly AI capabilities are actually displacing software development work. Right now, that evidence is thin. Using the St. Louis Fed's analysis of the Real-Time Population Survey, Citadel noted that daily use of generative AI for work is remaining "unexpectedly stable" and currently "presents little evidence of any imminent displacement risk."
The episode also reflects a deeper pathology in how financial markets process technological change. Citrini's "ghost GDP" argument assumes that displaced human wages will permanently vanish from the economy, ignoring how productivity gains have historically tended to reallocate value rather than destroy it. That assumption has been proven wrong by every previous technology wave, from mechanised agriculture to the personal computer. It may eventually be proven right by AI, but the burden of proof for that claim is high, and a single speculative blog post — however thoughtfully constructed — does not meet it.
What this week revealed is that the enterprise software market faces real questions about valuation, pricing power, and competitive differentiation in an AI-rich world. Those questions deserve serious analysis from regulators, fund managers, and technology strategists alike. The answers will emerge gradually, through quarterly earnings reports, productivity data, and the slow accumulation of evidence about where AI genuinely displaces human work and where it merely augments it. That process is inherently boring. Given what a single dramatic essay just did to retirement accounts from Sydney to Seattle, boring would be a very welcome change.