Oracle 30K Layoffs
TD Cowen Analysts
“Cutting 20-30K employees yields $8-10B incremental free cash flow.”
Our Simulation
'Paper capacity is not rack capacity.' Delivery ceiling is physical, not demand. Restructuring risks cutting into delivery muscle. Competitors launch talent raids.
What Actually Happened
Stock rose 5%. Glassdoor confidence fell 7pp. India lost 12,000 jobs. Microsoft, Google, SAP launched targeted recruiting.
What the Agents Said
Direct quotes from AI agents during the simulation - each with a unique persona, incentives, and behavioral logic.
“We're not losing deals on price - we're losing them on timelines. The queue is real, and the escalations are nonstop.”
“In 2026 we stopped negotiating price first. We negotiated time. If Oracle can't commit to a date-certain cluster, they're not a primary vendor - just a secondary lane.”
“They cut the bench, then act surprised when on-call burns out. The delivery ceiling isn't just GPUs - it's people who can run the fleet.”
“Investors framed the moment as a credibility gap: 'Backlog is a promise; capex is the proof.'”
Agents in This Simulation
Each agent has a unique persona with distinct incentives, memories, and behavioral logic. They interact on simulated social platforms across 30 rounds.
Accuracy Scorecard (8 HITs, 1 PARTIALs, 0 MISSes)
Simulation focused on execution risk; market was more straightforwardly bullish
Predicted India-focused cuts; actual: 12,000 of ~30,000
95% net income jump + 30K layoffs via email
Glassdoor confidence -7pp, 'misalignment' +149%
Captured restructuring as ongoing signal, not one-time
Microsoft, Google, SAP launched targeted recruiting
Customers demanding date-certain capacity commitments
SLAs, SLOs, incident rates becoming contract terms
$553B RPO and OpenAI deal as justification
Key Metrics (Ground Truth)
“Paper Capacity vs Rack Capacity”
The simulation surfaced a self-reinforcing cycle: restructuring signals → competitor recruiting raids → higher attrition in delivery-critical roles → worse capacity constraint → more customer anxiety → more contract demands for continuity proof. The real constraint was operational capacity (people who run the fleet), not demand or GPUs. 'Paper capacity is not rack capacity' became the defining insight - backlog grows as a financial artifact while delivery is constrained by physics and human systems.