Citrini "AI Doomsday" Substack Moves Markets, Anthropic Acquires Vercept, and Infrastructure Backlash Accelerates

A speculative Substack post from Citrini Research describing an AI-driven economic collapse scenario by 2028 sent shockwaves through Wall Street — S&P dropped 1%+, Uber, Mastercard, AmEx, and DoorDash fell 4-6% specifically. Separately, Anthropic quietly acquired Vercept, a Seattle computer-use AI startup, signaling a push deeper into agentic capabilities after Meta poached one of Vercept's founders. On the infrastructure front, public opposition to data centers is hardening into actual construction bans in some localities, while Pennsylvania's aging grid is being flagged as a chokepoint for AI expansion. The White House has signaled it expects AI hyperscalers to absorb electricity rate hikes — and most have already agreed.

Impact Score

8/10

Key Stories

MARKETS

'A feedback loop with no brake': Citrini AI doomsday report shocks markets

The Guardian — Speculative Substack post describing AI-driven 57% S&P wipeout by late 2027 moved real money — Uber, Mastercard, AmEx, DoorDash all dropped 4-6%. This is not about whether the scenario is accurate. It's about investor psychology shifting: AI disruption anxiety is now a market-moving force, not a thought experiment.

Anthropic acquires Vercept after Meta poached co-founder

TechCrunch — Anthropic is buying agentic computer-use capabilities directly rather than building them. Vercept specialized in complex computer-use agents — exactly what Claude Code and Operator-class products need. The acquisition timing (immediately after Meta talent raid) suggests competitive urgency.

POLICY

Public opposition to AI infrastructure hardening into construction bans

TechCrunch — Backlash against data center expansion is converting from NIMBYism to actual municipal policy. Combined with Pennsylvania grid strain reporting, this is a real constraint on AI buildout timelines — not just PR noise.

POLICY

White House signals AI companies must cover electricity rate hikes

TechCrunch — Hyperscalers have pre-agreed. This is a quiet but significant cost-pass-through: AI infrastructure costs are being socialized onto the companies building it. Signals the administration wants AI growth without visible consumer energy bill increases.

Gemini 3.1 Pro released — 1M context, 77.1% on ARC-AGI-2

LLM Stats — Google's latest Pro-tier model is benchmarking aggressively on the hardest reasoning evals. ARC-AGI-2 was designed to be resistant to LLM cheating. 77.1% is a significant threshold — puts pressure on Anthropic and OpenAI's next releases.

Analyst Take

The Citrini story is the signal to watch, but not for the reasons most are reporting. The scenario itself is speculative — but the fact that a Substack post moved Uber down 5% tells you something real: institutional investors are now primed to price in AI disruption risk at the company level, not just the macro level. When a credible-sounding narrative names your ticker, you move. That's a new dynamic. What Citrini got right, regardless of the doomsday framing: the "friction economy" companies — platforms that make money by being the middleman between desire and fulfillment — are genuinely exposed to agentic AI. Uber, DoorDash, payment processors, travel booking. This isn't 2028 speculation, it's already happening at the margins. The Anthropic/Vercept acquisition is the other story worth tracking closely. Anthropic is buying computer-use capability under competitive pressure — Meta was already circling. This accelerates the timeline to full-stack agentic products from Anthropic. Combined with Gemini 3.1 Pro's ARC-AGI-2 numbers, the capability race is compressing. Dong should watch: which "friction economy" companies are in his portfolio or on his radar? The market is now giving a discount signal. That window historically closes fast once narratives calcify into consensus.