AI & Automation
From hype to operating model.
Latest: Beneath near-universal AI adoption, economic value is concentrating in a minority of firms that have scaled AI into reinvented workflows; the binding 2026-2028 divide is value capture and...
Past capability, into liability.
The May briefing's org-chart thesis has hardened into a counterparty thesis, and the binding constraint has dropped a layer: from model capability to insurance wording, agent identity, contract law, the grid queue and the inference invoice.
After the Copilot: When AI Starts Signing, Spending and Failing
The frame that organised AI conversation last cycle, that AI had joined the org chart, has hardened into something blunter: AI has started signing, spending and failing.
The Value-Capture Divide: Why AI's Gains Are Concentrating Beneath the 88% Adoption Headline
Beneath near-universal AI adoption, economic value is concentrating in a minority of firms that have scaled AI into reinvented workflows; the binding 2026-2028 divide is value capture and...
Posture: PrepareThe Inference-Cost Inversion: Why Cheaper Tokens Are Making Enterprise AI More Expensive
Agentic and reasoning workloads are consuming tokens faster than unit prices fall, inverting AI's cheap-tokens story.
Posture: PrepareBeneath the Leaderboard: How Chinese Open-Weight Models Quietly Captured the Deployed AI Base
The deployed AI inference base is shifting to Chinese open-weight models even as US frontier performance still narrowly leads.
Posture: PrepareThe execution layer goes live as the rulebook stands down
Inaugural Change Tracker on Artificial Intelligence and Automation. Eight signals: 5 Accelerating (agentic AI to production, AI capex supercycle, open-weight frontier parity, EU AI Act delay via the...
Off the Books: How AI's Build-Out Quietly Became a Financing Risk
A growing share of hyperscaler AI build-out financing has migrated off balance sheet into special-purpose vehicles, private credit and residual-value-guaranteed leases.
Posture: PrepareThe Underwriting Brake: How AI Exclusions Are Becoming the Real Constraint on Enterprise Automation
Insurability, not capability, is becoming the binding constraint on agentic-AI scale-up. Commercial insurers are rewriting standard liability policies to exclude AI-related losses while only a thin...
Posture: PrepareIdentity Before Capability
Beneath the agentic-AI capability narrative, NIST's February 2026 AI Agent Standards Initiative, the NCCoE identity-and-authorization concept paper, OpenID Foundation's NIST response, and Okta and...
Compute as Statecraft
Beneath the US-China rivalry narrative, sovereign AI compute is emerging as a new geoeconomic asset class: Gulf capital (G42, HUMAIN) plus US silicon plus India and middle-power jurisdiction are...
Worlds for Machines
The economic centre of gravity in immersive computing has shifted from VR headsets and consumer metaverses to AI-trainable world models for robots and autonomous vehicles, with capital, platform...
The Superconducting Hedge
The headline quantum milestones obscure an architecture shift: neutral-atom and trapped-ion systems now lead on verified logical qubits, and superconducting's own pioneer is hedging, with a 2026 to...
When AI Joins the Org Chart
Three years after generative AI moved from demo to workflow, the strategic question is no longer whether to deploy AI.
Why "We Use AI" Is About to Stop Winning Contractor Tenders
The contractor tender-narrative pivot from generic "we use AI" claims to cost-driver-first AI mapping is moving from emerging differentiator to binding tender requirement on a 2026-2028 horizon...