Decision cockpit
What changed across the thesis, investable targets, and source coverage. Last source update 2026-07-15 13:34 UTC. Signals below are inputs for human review—not recommendations.
Morning status
Operational counts from the current fact graph
Signal queue
Claim drift and severity first; repeated mechanics collapsed
Thesis pressure
Drift, refuting pressure, then evidence gaps
AI value accrues to incumbents with proprietary data and distribution more than to new software startups.
The market overprices SaaS extinction; moated SaaS (data, regulatory, workflow depth) survives and re-rates.
Agentic workloads grow token consumption faster than chat workloads.
Precision medicine and AI diagnosis outperform crowded drug-discovery as investment areas.
NATO-aligned defence budgets keep rising through 2027.
Chinese control of critical materials (tungsten, rare earths) tightens as a strategic chokepoint.
Investment objects
Ecosystem change with access, coverage, contributors, and the newest direct fact
New thesis evidence
Every row resolves to its original source
Cybersecurity is the exception to the no-pure-software thesis: genuinely novel threats create new spend categories.
Codex encrypting sub-agent prompts highlights a novel AI-security threat that creates new cybersecurity spend categories.
Cybersecurity is the exception to the no-pure-software thesis: genuinely novel threats create new spend categories.
Cybersecurity AI startup addresses novel threats, supporting thesis of new spend categories.
Low-cost autonomous systems out-compete legacy platforms on cost-exchange ratios.
Startup building drones and defense robotics supports trend of low-cost autonomous systems.
Low-cost autonomous systems out-compete legacy platforms on cost-exchange ratios.
Swarm robotics and unmanned vehicle startup confirms low-cost autonomous systems development.
Grid power availability, not chips, becomes the binding constraint for new datacentre capacity.
Oracle highlights chip/memory supply constraints as still binding, slightly undermining the claim that power is the sole binding constraint.
Datacentre demand exceeds buildable supply through mid-2027.
Oracle's mention of competitive supply for AI accelerators and memory indicates ongoing supply constraints, contributing to demand overhang.
Grid power availability, not chips, becomes the binding constraint for new datacentre capacity.
Oracle explicitly cites global power constraints as a major challenge for data center expansion, directly validating the power bottleneck claim.
Datacentre demand exceeds buildable supply through mid-2027.
Oracle reports increased customer demand for cloud services, supporting the thesis that datacentre demand outstrips supply.
Research
Artifacts generated from the same evidence graph
SK Hynix $26.5B raise signals datacentre demand overhang deepens
Read cited brief →Source freshness
Green sources only; freshness is not source quality
Theme rooms
Thesis state and coverage by mandate theme; order is curated, not ranked
Economic asymmetry favors autonomous attrition
Orbital infrastructure as strategic priority