Week 36 argued energy is a leadership job inside the factory. Week 37 zooms out to where that job gets decided first: European competitiveness. The continent doesn’t have an AI-ambition problem, it has a megawatt-and-decade problem.

The race shifted from algorithms to infrastructure: what decides location and scale is now whether power actually gets there. The IEA reports 2,500+ GW stalled in grid-connection queues worldwide; inside the EU, queues span at least 16 member states, putting ~120 GW of mature renewables at risk by 2030, with some connections taking up to ten years, a millennium in AI time. The price gap isn’t close: EU industrial power runs over twice US levels and ~50% above China’s. Four gaps define the shortfall, price, connection, planning mismatch, and an internal split between nuclear-strong and grid-constrained locations, none of them a weakness in AI itself, all in the physical layer underneath it.

The companies acting on it aren’t waiting for reform — they’re engineering around it, as with Amazon’s ~€16B Aragon campus, a secondary market chosen for available grid capacity over the multi-year queue in established hubs. Your action this week: add three lines to your next AI investment case, real power cost, real connection timeline, local capacity vs queue. Full breakdown at renegrywnow.com.

Reflection questions

* Does your next AI investment case include power cost, connection timeline and local capacity, or just model and hardware cost?

* Are you treating site selection as a competitiveness decision, or defaulting to the established hub and its queue?

* Where could efficiency or on-site generation become the fastest megawatt you have, the one you don’t have to wait to connect?

Keywords: Europe AI Competitiveness, Grid Connection Queues, EU Electricity Prices, Energy Infrastructure, Data Centre Power, Site Strategy, Physical AI, Industrial Energy, Grid Capacity, Interconnection

The blog is here

Series: Energy Dominance · Week 37 · Part INext: Part II: Strategic Dependency vs. Sovereignty: Why Supply Chains and Grids Are Becoming Political.

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit renegrywnow.substack.com

This week ran as a set: Part I named the human advantage, Part II the energy discipline, and Part III puts both inside one job description, drawing on Paul Daugherty and James Wilson’s Human + Machine (HBR Press, 2018).

For a century, “Who executes this task?” was a complete leadership question. It stopped being one when the answer became “a person and a machine, together.” Value now sits in what the authors call the missing middle, where AI amplifies human skill rather than replacing it wholesale. Leadership becomes hybrid-aware: less “who executes this” and more “how is this collaboration designed, skilled and governed?” Four practices, told through one plant’s quality problem: reimagine the process instead of automating the old flow (metric: yield and improved hours, not “model live”); build fusion skills via a 90-day reciprocal rotation; responsible normalizing with three explicit action zones and supervisors assessed on override quality; and relentless reimagining as a quarterly management system.

Across all four, the projects don’t change, what leadership pays attention to does. Leaders become system designers and judges of last resort; culture expects challenge of both human and machine decisions. Your action this week: take one process you’re about to “add AI” to and, before funding any tool, ask what it would look like redesigned around what humans and machines each do best. Full framework at renegrywnow.com.

Reflection questions

* Before funding your next AI tool, have you redesigned the process, or are you automating the old flow?

* Are your autonomy and override norms written explicitly by leadership, or inherited from vendor defaults?

* Do you review process redesign and skills before tool status in leadership meetings, or after?

Keywords: Human + Machine, Daugherty Wilson, Missing Middle, MELDS, Fusion Skills, Responsible Normalizing, Human-Machine Collaboration, System Design, Leadership, Physical AI

Link to the full blog

Series: Energy Dominance · Week 36 · Part IIIPrevious: Part I: The Human Advantage · Part II: The Energy–AI Convergence.

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit renegrywnow.substack.com

Part I argued human judgment is the scarce asset inside an AI-driven factory. Part II turns to the constraint outside it: how much of that factory can run once energy, not silicon, becomes the binding limit.

AI is two-sided, it adds load (edge compute, robots, always-on sensing) and reduces it (better control, less scrap, fewer idle losses). Leaders who see only one side over-build or over-claim. The scale is no longer abstract: the IEA found global data-centre electricity demand grew 17% in 2025 and AI-focused demand surged 50%, while Stanford reports factories in data-centre regions already facing higher costs and longer interconnection timelines. The binding constraint on AI has shifted from chips to grid access.

Five leadership lessons follow, constraint forces priority, activity isn’t progress, conflicts need a decision owner, infrastructure is strategy, silos recreate waste, and none fail loudly; they fail as a slide that says “aligned” while the P&L and the kWh meter disagree. The working pattern isn’t a new dashboard but fewer, in one room: energy per unit, unplanned downtime, and pilot-to-supervised-operation share on one page, with every agent proposal showing impact on all three. Your action this week: find the two meetings reviewing energy and AI separately and ask what decision you’re getting wrong because the numbers never share a page. Full pattern at renegrywnow.com.

Reflection questions

* Is AI adding load or reducing it in your plant, and do you track both sides in one decision?

* Are energy and AI still reviewed in separate meetings, producing local wins and global losses?

* When an agent recommends an energy-saving setpoint that raises quality risk, who owns that trade-off?

Keywords: Energy-AI Convergence, Grid Capacity, Industrial Energy, Power Quality, Energy per Unit, IEA, Data Centre Demand, Interconnection Timelines, Infrastructure Strategy, Manufacturing Leadership

Here is the link to the full blog

Series: Energy Dominance · Week 36 · Part IIPrevious: Part I: The Human Advantage in an AI-Driven Factory.

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit renegrywnow.substack.com

Week 35 closed on team quality. Week 36 opens with its operational version: as Physical AI takes over speed, scale and pattern recognition, what’s left for humans is the part that was always hardest to automate, and it gets more valuable, not less.

Machines absorb work that’s frequent, sensor-rich and rule-bounded: screening, parameter suggestions, routine scheduling, anomaly flags. They don’t absorb brownfield reality, shifting product mix, worn tooling, undocumented workarounds, and the moment several weak signals together mean “stop.” The data agrees: 81%+ of manufacturing task hours are expected to stay human-driven, because AI replicates codified but not tacit knowledge. Five human advantages compound with AI, contextual judgment, tacit knowledge, responsible override, learning transfer, and floor-level trust, none of which show up on a capability matrix, all of which show up the moment someone decides whether to trust the system. Labor research adds an edge: AI hits entry-level roles hardest precisely because experience is the hard-to-copy part.

The strategic difference: whether experienced people are designed into the loop as supervisors and improvers, or designed out as cost, the second looks efficient on a slide and fails in production. Your action this week: find your best overrider and ask what they saw that the system didn’t, then whether that knowledge is being captured. Full breakdown at renegrywnow.com.

Reflection questions

* Who’s your best overrider, and is the judgment behind their saves being captured, or walking out at retirement?

* Are your experienced people designed into the AI loop as supervisors, or treated as cost to remove?

* Can your operators challenge the system’s recommendations, or must they follow them blindly?

Keywords: Human Advantage, Physical AI, Tacit Knowledge, Responsible Override, Human-in-the-Loop, Exception Handling, Shop Floor Trust, Contextual Judgment, Manufacturing Leadership, Augmentation

Bloglink

Series: Energy Dominance · Week 36 · Part INext: Part II: Leadership Lessons from the Energy–AI Convergence.

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit renegrywnow.substack.com

This week ran as a set: Part I named the activity trap, Part II showed why fixes ripple sideways, and Part III turns to the humans inside the system — drawing on Steven Thurber and Bryan Miller’s Good Team, Bad Team (Page Two, 2024).

The core reframe: as AI agents join the team, human team quality matters more, not less — technology amplifies whatever dynamic already exists. An agent doesn’t fix a bad team; it gives one a faster way to be wrong together. Deloitte found 56% of leaders design AI for business outcomes but only 40% for both business and human outcomes. Thurber and Miller’s distinctions translate directly: good teams create psychological safety to challenge AI recommendations and hold a shared definition of winning; bad teams suppress dissent and drift into activity without alignment. The difference is rarely expertise — it’s whether disagreement surfaces before deployment or only after an incident. One plant’s challenge protocols caught edge cases on a whiteboard; another found the same ones on the production line.

Building the good version is deliberate: select for collaborative capacity, define decision rights, reward challenge of humans and AI, invest in facilitation, review team effectiveness like technical performance. Your action this week: ask when someone last challenged an AI recommendation out loud — and what happened to them. Full model at renegrywnow.com.

Reflection questions

* When did someone last challenge an AI recommendation on your team out loud — and what happened to them?

* Does your best human–AI team differ from a weaker one in expertise, or in whether dissent is allowed to surface?

* Do you review team effectiveness with the same rigor you apply to technical performance?

Keywords: Team Effectiveness, Good Team Bad Team, Thurber Miller, Psychological Safety, Human-AI Collaboration, Challenge Protocols, Decision Rights, Cross-Functional Teams, Physical AI, Leadership

Here is the full link

Series: Energy Dominance · Week 35 · Part IIIPrevious: Part I — AI Activity vs Real Progress · Part II — Systems Thinking in Autonomous Operations.

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit renegrywnow.substack.com

Part I argued most AI programs confuse activity with progress. Part II asks why, even when chasing genuine progress, fixing one problem in an autonomous operation so often creates another. The answer: linear troubleshooting — find the cause, fix it, move on — breaks down when the cause isn’t in any single component but in the interactions between them.

The concept is emergence: system behavior that arises from component interactions but can’t be predicted from any component alone — ant colonies routing around obstacles, flash crashes from rational algorithms, flocks synchronizing through local rules. A 2025 study showed hundreds of AI agents with no central coordinator converging on conventions nobody programmed. Systems thinking is the countermeasure — a set of habits: mapping feedback loops, understanding delays, finding leverage points, anticipating the unintended consequences of local optimization. Organizations are now building teams specifically to read emergent behavior rather than debug parts.

Systems thinkers ask different first questions — how will this agent interact with existing controls, what happens when agents optimize for different goals, where are the stability loops — and answer them in a digital twin before deployment, not after an incident. Your action this week: take one fix that created a new problem elsewhere and map the interfaces your root-cause meeting missed. Full toolkit at renegrywnow.com.

Reflection questions

* When did fixing one thing on your floor quietly break another — and did anyone map why?

* Do your reviews examine system-level performance, or only local KPIs that can each look green while the whole degrades?

* Are you rehearsing multi-agent interactions in simulation, or discovering them live?

Keywords: Systems Thinking, Autonomous Operations, Emergence, Multi-Agent AI, Feedback Loops, Leverage Points, Cyber-Physical Systems, Digital Twin, Interface Risk, Industrial AI

Here is the blog

Series: Energy Dominance · Week 35 · Part IIPrevious: Part I — Why Most Organizations Confuse AI Activity with Real Progress.

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit renegrywnow.substack.com

Week 34 closed with a framework for turning vague AI ambition into strategic choices. Week 35 opens with the trap that framework exists to prevent: mistaking activity for progress. A dashboard full of green metrics and a CFO’s spreadsheet full of red questions can both be true at once — and that gap is where AI programs die.

The numbers are stark: manufacturing AI projects fail at ~76%, and 84% of those failures trace to leadership decisions, not technical limits — most often launching without defined success metrics. Projects with quantified metrics succeed 54% of the time; without them, 12% — yet ~73% still launch with no clear metrics. One analyst calls it “AI theatre.” The confusion persists because activity is easy to report and impact is hard, reinforced by misaligned incentives, technical debt, and AI teams isolated from decision-makers.

Real progress needs no showcase — it shows up in downtime, energy per unit, yield, and in capability: decision rights, data foundations, teams that supervise and improve. Grant Thornton found fully-integrated-AI organizations nearly 4x more likely to report revenue growth than those still piloting (58% vs 15%). Your action this week: hold every initiative against one question — which operational number does it move? Full breakdown at renegrywnow.com.

Reflection questions

* How many of this quarter’s AI initiatives could you translate into a moved operational number — downtime, energy, yield?

* Does your reporting count pilots, or impact achieved and scaled?

* Have you defined what “progress” specifically looks like in your context — before approving the next pilot?

Keywords: Industrial AI, AI Activity Trap, AI Theatre, Progress Metrics, Pilot to Production, Manufacturing AI ROI, Success Metrics, Governance, Physical AI, Operational Outcomes

Here is the blog

Series: Energy Dominance · Week 35 · Part INext: Part II — Systems Thinking in the Age of Autonomous Operations.

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit renegrywnow.substack.com

This week ran as a trilogy: Part I diagnosed how AI exposes weak leadership, Part II priced the paralysis that follows, and Part III hands over the tool — A.G. Lafley and Roger Martin’s Playing to Win strategy cascade, applied to Physical AI.

Most AI investment is activity without strategy: ask ten manufacturing leaders why they’re investing and eight say “because everyone else is.” Capgemini found 79% of leaders engaging with Physical AI but only 4% at large-scale deployment — a field of scattered, disconnected pilots. The cascade replaces that scatter with five integrated choices, made in order and checked against each other: Winning Aspiration (concrete, e.g. “zero surprise downtime”), Where to Play (which plants and lines first, concentration over dilution), How to Win (the distinctive advantage every investment must serve), Capabilities (build AI-supervision skills in operators, not just hire data scientists), and Management Systems (stage-gates, governance, energy-linked KPIs).

It fits Physical AI unusually well because these investments are capital-heavy, safety-touching and lasting — so vague strategy is especially dangerous. The cascade forces hard conversations early and cheaply, and gives technology, operations and leadership one shared language. Siemens/NVIDIA’s single governed Erlangen blueprint is the cascade at work. Your action this week: hold every current pilot against one question — which winning aspiration does this serve? Full cascade at renegrywnow.com.

Reflection questions

* Can you name the single winning aspiration each of your current AI pilots serves — or do some have no answer?

* Are you playing everywhere at once, or have you concentrated Physical AI where it wins first?

* Does every AI investment demonstrably support one distinctive “How to Win” — or five different ones?

Keywords: Playing to Win, Strategy Cascade, Physical AI, Lafley Martin, Winning Aspiration, Where to Play, How to Win, Manufacturing Strategy, Competitive Advantage, AI Roadmap

The blog is here

Series: Energy Dominance · Week 34 · Part IIIPrevious: Part I — AI Will Not Replace Leaders, It Will Expose Bad Ones · Part II — The Cost of Delayed Decisions.

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit renegrywnow.substack.com

Part I argued AI exposes weak leadership. Part II puts a number on one of the clearest symptoms: the decision that never gets made. Most operational decisions in a complex plant are reversible — a parameter adjusted back, a maintenance call revised — yet get deliberated like bet-the-company moves. The cost of a fixable mistake is almost always smaller than the cost of delay.

West Monroe’s 2026 study of 1,200+ leaders at $250M+ companies found nearly three in four estimate their organization loses up to 5% of annual revenue to slow decisions — the “Slowness Tax” — driven by leadership behavior, not technology, with AI gains largely absorbed by process friction first. The velocity matrix shows why: complex industrial environments gravitate toward the low-speed/high-quality quadrant — analysis paralysis — because a wrong decision is attributable while delay is diffuse. Underneath sits an unspoken incentive: the decider bears the risk of being wrong; the delayer bears almost none.

The fix isn’t faster models but clearer decision rights: match deliberation to reversibility, make delay costs as visible as mistake costs, and make inaction a reviewable outcome. IBM’s 2026 CEO Study: AI rewards decisiveness and penalizes hesitation. Your action this week: sort recurring decisions into reversible vs. irreversible and check how much committee time each consumes. Full model at renegrywnow.com.

Reflection questions

* Are your reversible decisions eating the same deliberation as your genuinely irreversible ones?

* Do you track the cost of delay with the same rigor as the cost of mistakes — or does delay just get a shrug?

* In your organization, who bears the risk of a slow decision — anyone at all?

Keywords: Decision Velocity, Slowness Tax, Analysis Paralysis, Decision Debt, Reversibility, Decision Rights, Incentive Asymmetry, Industrial Leadership, Decision Architecture, Operational Excellence

Blog is here:

Series: Energy Dominance · Week 34 · Part IIPrevious: Part I — AI Will Not Replace Leaders. It Will Expose Bad Ones.

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit renegrywnow.substack.com

Week 33 argued that judgment, not information, is the scarce resource. Week 34 asks what happens to leaders who never had much judgment — now that AI has removed the excuse that covered for them.

AI doesn’t create leadership weakness; it removes the fog that hid it. For years, slow information gave average leaders cover — a slow decision looked reasonable while everyone “waited on the numbers.” When analysis that took days now takes minutes, that excuse expires, and AI works like an X-ray for competence. Grant Thornton’s 2026 survey found many executives doubt their organization could pass an independent AI governance audit within 90 days — an accountability gap, not a technology one. The old playbook (confidence, visibility, decisiveness) now partly backfires: confidence without humility becomes a failure to question outputs or tell fluency from accuracy.

The exposure shows up in three deficits — trade-off avoidance disguised as “more information,” a strategic judgment gap, and a missing decision framework — and, most tellingly, in a quiet reallocation of trust as teams route around slow leaders. Your action this week: ask honestly whether your best people still bring you the hard calls, or have started solving around you. Full diagnostic at renegrywnow.com.

Reflection questions

* Do your best people still bring you the hard calls — or have they quietly learned to solve around you?

* Could your organization explain, within 90 days, how a given AI-supported decision was made and who owns it?

* Where is “gathering more information” functioning as genuine need — and where has it become cover for avoiding a trade-off?

Blog Link:

Series: Energy Dominance · Week 34 · Part INext: Part II — The Cost of Delayed Decisions in High-Complexity Industrial Environments.

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit renegrywnow.substack.com