COO Signals Radar
Weekly Intelligence Report — July 6, 2026
Last Updated: Jul 5, 2026, 10:11 AM (Manila Time)
Executive Snapshot
- •US data-centre backlash: ≥20 projects worth $42bn cancelled in Q1 2026 alone; the ~$3trn 2026-30 AI buildout now carries political risk — compute and power become ops planning variables
- •Bank of Korea (5,512 workers): AI saves 3.8% of work time but shows zero correlation with output — automation ROI hinges on workflow redesign, not tool rollout
- •Hyperscalers ($4bn/yr — Google 34+ cables, Meta's $10bn Project Waterworth) are rerouting Asia's subsea cables around Chinese-controlled waters — digital supply routes now have chokepoint geopolitics
- •Ramp/Revelio (21,000 firms): high-AI-adoption firms grew headcount ~10% in two years, +12% at entry level — workforce planning evidence against the 'automation = fewer people' reflex
Signals Overview
| Rank | Category | Headline | Score | Urgency | Action |
|---|---|---|---|---|---|
| 1 | Infrastructure & Capex | US Data-Centre Backlash: $42bn of Projects Cancelled in Q1 2026 as the $3trn AI Buildout Meets Local Politics The Economist | 87 | High | COO: treat compute, grid power, and data-centre siting as supply-chain risk categories with named owners — network strategy + energy procurement, this quarter |
| 2 | Workforce & Productivity | Bank of Korea: AI Saves 3.8% of Work Time but Output Effect Is Zero — Automation ROI Hinges on Workflow Redesign Bank of Korea Issue Note 2026-12 (Suh, Oh, Yoon) | 83 | High | COO: audit every ops AI program for throughput delta (not time saved) and assign a single workflow-redesign owner per program — ops excellence lead, 60 days |
| 3 | Geopolitics of Supply | Hyperscalers Reroute Asia's Subsea Cables Around Chinese-Controlled Waters — $4bn/Year Interconnect Capex The Economist | 80 | Medium | COO: add subsea-cable and cloud-region route risk for Asia operations to the supply-chain risk register — network engineering + risk management, 90 days |
| 4 | Workforce & Productivity | Ramp/Revelio: High-AI-Adoption Firms Grew Headcount ~10% in Two Years, +12% at Entry Level The AI Daily Brief (Ramp × Revelio Labs 21,000-firm payroll study; Box 1,600-firm survey) | 72 | Medium | COO: reframe the FY27 operations workforce plan around AI skill-mix shift rather than headcount reduction — ops HR + workforce planning, next planning cycle |
| 5 | Process/Agent Automation | Tacit Knowledge Is the Blocker for Process Automation — Monumental's Bricklaying Robots and Meta's Keystroke Tracking Show the Routes and Traps The Economist | 71 | Medium | COO: pilot video-based tacit-knowledge capture on one physical process before considering keystroke-style monitoring — manufacturing engineering + HR, 90 days |
| 6 | Process/Agent Automation | Anthropic Economic Index: Heaviest Automation Users Are the Most Optimistic — and AI Cost Tracks the Value of the Work Anthropic Economic Index (Cadences report, June 2026) | 70 | Low | COO: re-baseline AI budget allocation by value of the work performed (occupation/role) rather than per-seat licences — ops finance + IT, FY27 budgeting |
Deep Dive: All Signals
Why now: The 2026-06-27 Economist cover package quantified the cancellation ledger ($42bn in Q1 2026) and put a measurable political price on the buildout for the first time.
Summary
At least 20 US data-centre projects worth $42bn (3.5 GW) were cancelled in Q1 2026 alone after local pushback, with ~$100bn scuppered by protest overall — against a planned ~$750bn hyperscaler spend and ~$3trn global buildout for 2026-2030 that would take US AI capacity from ~12 GW to as much as ~60 GW by 2030. Pew data shows the opposition is ideological, not proximity-driven, and Ohio's binding 85% take-or-pay rule is emerging as the model policy for internalising infrastructure costs.
Impact on Retail/CPG
Retail and CPG operations now compete with data centres for grid capacity, utility rates, and land in the same 'Silicon Heartland' states (Ohio, Michigan, Texas) where distribution networks are concentrated. Compute is also already being rationed — Anthropic throttling usage, Microsoft repricing Copilot steeply — so AI-in-operations roadmaps carry inference-cost and availability risk, not just adoption risk.
Recommended Actions
- Add grid-capacity and utility-rate exposure screening to site selection for DCs and plants in data-centre-heavy states — network design + energy procurement, 60 days
- Stress-test FY27 AI-operations budgets for inference price and rationing volatility (Anthropic throttling, Copilot repricing as leading indicators) — ops finance + IT, 30 days
- Apply Ohio-style take-or-pay logic to any co-located power or infrastructure commitments — legal + energy procurement, before next capex approval
Risks
- State-level moratoriums could concentrate compute capacity in fewer regions, worsening latency and redundancy options for operations
- The consumer-electricity-price narrative could expand into political pressure on all large industrial energy users, not just data centres
Sources
From the Second Brain
Why now: BOK Issue Note 2026-12 (June 2026) is the first representative macro dataset quantifying the adoption-to-output gap, and firm-level hiring data landing this week (Ramp/Revelio) completes the picture: adoption pulls in people and capex before per-worker output moves.
Summary
A representative Bank of Korea survey of 5,512 workers finds generative AI saves 3.8% of work time (~1.5 hours/week) but the correlation between worker-level time savings and output growth is zero — with 51.8% worker adoption against only 9.6% enterprise adoption. Output gains materialise only where structure supports them: self-employed (+1.0pp), professionals (+0.7pp), and top-50% intensity users (+0.5pp) — groups with outcome ownership, autonomy, and performance-linked incentives.
Impact on Retail/CPG
This is the strongest empirical warning yet for ops automation programs: deploying AI tools into unchanged workflows produces adoption metrics, not throughput. For retail/CPG operations the binding constraints are the paper's four mechanisms — task-level diffusion, workflow rigidity, downstream bottlenecks, and incentive misalignment — all of which are COO-owned design levers, not IT deliverables.
Recommended Actions
- Audit the top 10 operations AI pilots for time-saved vs measured throughput delta; redesign or stop the ones showing only time savings — ops excellence / CI team, 60 days
- Assign one accountable owner for workflow redesign (not tool rollout) in each automation program — COO staff, next operating review
- Create explicit reallocation mechanisms so saved hours flow to high-value work (exception handling, mentoring, new capacity) backed by performance-linked incentives — HR + ops leadership, this quarter
Risks
- The zero-output finding may reflect J-curve timing typical of general-purpose technologies — killing pilots prematurely would forfeit the later payoff
- Korean household-survey data may understate gains in operations that have already redesigned workflows around AI
From the Second Brain
Why now: Published in the 2026-07-04 Economist edition: the first cables of the Indian Ocean spine are live and Meta's Waterworth is in development — routing decisions that will define Asia's connectivity map are being locked in now.
Summary
The ~700 subsea cables carrying 99% of intercontinental internet traffic are being rewired by the AI boom: ~$4bn/year of new cable investment over the next four years, mostly by hyperscalers (Google has funded 34+ cables, 18 solo; Meta's $10bn Project Waterworth). New routes form an Indian Ocean spine (Oman → Maldives → Christmas Island → Australia → Guam) drawn specifically to avoid Chinese-controlled seabed — where repairs inside the nine-dash line need Beijing approval — and Southeast Asian chokepoints; no new US-China cables have been approved since the Obama administration.
Impact on Retail/CPG
Asia-based manufacturing, shared-services centres, and e-commerce platforms run on connectivity whose physical routes are being redrawn for geopolitical reasons by private firms. Digital supply routes now carry the same chokepoint risk discipline as shipping lanes: latency, repair-time, and redundancy assumptions for Asia operations need explicit review rather than inherited defaults.
Recommended Actions
- Map network dependency of Asia operations (plants, shared-services/GCC sites, commerce platforms) on Malacca and South China Sea cable routes — network engineering + BCP, 90 days
- Add subsea-cable and cloud-region concentration risk to the supply-chain risk register alongside physical chokepoints — risk management, next review cycle
- Require route-diversity and repair-SLA disclosures from cloud and telecom providers in Asia contracts — procurement + IT infrastructure, at next renewal
Risks
- Baltic and Taiwan cable-cut incidents show precautionary rerouting can become actual disruption — repair times in contested waters are the exposed variable
- Chokepoint states (Indonesia has openly mused about monetising its position) could raise transit and permitting costs on legacy routes
Sources
From the Second Brain
Why now: Episode published July 2: the first firm-level payroll dataset pairing AI adoption with hiring, landing the same week the BOK productivity-disconnect evidence firmed up — together they define the redesign-plus-hire agenda.
Summary
A Ramp/Revelio Labs study of 21,000 US firms finds high-AI-adoption companies grew headcount ~10% over two years (entry level +12%) while low adopters stayed flat, with growth starting 6-12 months into adoption. A Box survey of 1,600+ mid/large firms corroborates the direction: 58% expect headcount to rise over three years — 79% among the most mature adopters. Adopters are hiring for a new filter: 'people who know how to use AI and use it well.'
Impact on Retail/CPG
Direct counter-evidence to the 'automation means fewer people' reflex in ops workforce planning. For retail/CPG operations, the binding constraint shifts from headcount targets to hiring and retraining speed for AI-native skills — with entry-level frontline and planning roles the place adopters are investing, not cutting.
Recommended Actions
- Rebuild the FY27 ops workforce plan around skill-mix shift and AI-native hiring criteria, not headcount reduction targets — ops HR + workforce planning, next planning cycle
- Stand up an AI-native entry-level track (observe → assist → lead on open tasks) in operations and planning roles — ops academy / L&D, Q4 2026
Risks
- Endogeneity unresolved — firms that adopt AI were already fast-growing; the study's matched-control correction is asserted, not shown
- Headline numbers reach the vault second-hand via one podcast summary — trust the direction more than the magnitude
Sources
From the Second Brain
Why now: The 2026-06-27 Economist names the mechanism behind stalled enterprise-AI projects just as agent deployments move from back-office to frontline physical processes.
Summary
The Economist frames tacit knowledge (Polanyi: 'we can know more than we can tell') as the missing layer stalling enterprise AI. Dutch robotics firm Monumental only cracked automated bricklaying after video footage revealed masons vibrating their hands to bond mortar — a move no interview surfaced. Meta's Model Capability Initiative (keystroke and mouse tracking to train AI) triggered employee pushback, and an MIT survey (Danielle Li) finds workers know they hold uncodified knowledge and believe they can withhold it.
Impact on Retail/CPG
Automating frontline manufacturing, warehouse, and planning work requires capturing know-how that workers cannot articulate — and can strategically withhold. That makes process-automation programs an industrial-relations and incentive-design problem, not just a technical one: capture method choice (video observation vs surveillance-style tracking) determines whether the workforce cooperates.
Recommended Actions
- Pilot video-based capture (the Monumental pattern) on one high-variance physical process before any keystroke-style monitoring — manufacturing engineering + HR, 90 days
- Negotiate knowledge-capture terms (ownership, incentives, surveillance limits) with works councils ahead of FY27 agent deployments — labor relations, this half
Risks
- Surveillance-style capture triggers backlash and strategic withholding — exactly Meta's outcome with its Model Capability Initiative
- The Danielle Li survey findings are cited without a named paper — verify before using as canonical evidence
Sources
From the Second Brain
Why now: Report published June 26 — the first usage dataset separating agentic (responsibility-level) work from chat tasks, giving automation programs an empirical unit of measure.
Summary
Anthropic's June 2026 Economic Index reports Claude sessions increasingly consist of long-running agentic tasks, with 93% of conversations producing a classifiable artifact. Token cost scales with the wage of the occupation performing the underlying task (marketing managers at ~2× editors' wages use ~2.5× the tokens), and the linked EI Survey finds users who work in the most automated way expect AI to take on more tasks yet report the most optimism about pay, job security, and meaning.
Impact on Retail/CPG
Two operational takeaways for automation programs: budget AI by the value of the work being automated (occupation/role level), not by seats or flat licences; and the automation-equals-optimism finding is empirical counter-evidence for change-management plans that assume heavier delegation drives workforce anxiety.
Recommended Actions
- Re-baseline AI spend allocation by role value rather than per-seat licences in FY27 budgeting — ops finance + IT, next budget cycle
- Use the automation-optimism survey finding in change-management communications for agent rollouts — transformation office, ongoing
Risks
- Single-lab data — Claude users skew technical/early-adopter; no cross-lab confirmation of the patterns yet
- The survey chapter reached the vault truncated; mechanism and sample sizes need a second pass before external citation
From the Second Brain
Diff vs Last Week
- US Data-Centre Backlash: $42bn of Projects Cancelled in Q1 2026 as the $3trn AI Buildout Meets Local Politics87
- Bank of Korea: AI Saves 3.8% of Work Time but Output Effect Is Zero — Automation ROI Hinges on Workflow Redesign83
- Hyperscalers Reroute Asia's Subsea Cables Around Chinese-Controlled Waters — $4bn/Year Interconnect Capex80
- Ramp/Revelio: High-AI-Adoption Firms Grew Headcount ~10% in Two Years, +12% at Entry Level72
- Tacit Knowledge Is the Blocker for Process Automation — Monumental's Bricklaying Robots and Meta's Keystroke Tracking Show the Routes and Traps71
- Anthropic Economic Index: Heaviest Automation Users Are the Most Optimistic — and AI Cost Tracks the Value of the Work70
- Walmart FY26 Annual Report: 65% of Stores and 55% of FC Volume Automated, ~20% Unit-Cost Improvement — benchmark established; no vault updates this window
- SAP Sapphire 2026 Autonomous Supply Chain (Joule Agents in EWM/IBP) — product announced; next checkpoint is GA availability
- P&G Supply Chain 3.0 Large-Scale Rollout + Berlin Automated Night Shift — rollout underway; no new public data this window
- Amazon RIVR Acquisition + Amazon Now 30-Minute Delivery — event absorbed into last-mile baseline
- Tariff Volatility Regional Supply Chain Reset — theme persists but no fresh vault evidence this window; superseded by digital supply-route geopolitics (subsea cables)
Foundations
Evergreen briefings from Sunil's Second Brain — free subscriber access.
Designing IT Roles for an AI Era — A Talent-Strategy POV Question (2026-06-02): As AI pushes humans toward higher-value work anchored in domain mastery and solution design, how should we structure IT roles inside an ente
Unlocking 10X in Domain Masters as AI Gets Better Question (2026-06-11): "As AI is becoming better each day, how can one leverage this to unlock a 10X mindset for employees that have domain mastery?" The core claim: AI i
Hourglass Organization Steven Brovich's named org shape for the agentic-AI era. Named in A Leaders Guide to Advanced Team Structures (AWS Events). The form that preserves the talent pipeline while still capturing agent-a
Headcount-to-Value Pivot The central thesis of GCC Philippines Summit 2026 (PHx): enterprises (and their GCCs) are shifting from headcount-led growth to value-led growth — revenue and impact decoupling from FTE count as
Knowledge Work Factory Redesign OpenAI's framing of Codex (June 2026, in the report The Next Era of Knowledge Work ): knowledge work is the next domain due for a factory-style redesign, and Codex is positioned as that re
Skill Change Index (SCI) McKinsey's measure (GCC Philippines Summit 2026 (PHx)) of how much AI reprices the skills demanded by a role — the degree to which a given skill's relevance rises or falls as AI automates parts o