What if the Next Moonshot is a Power Line?
Digging into ENACT’s gigawatt-first roadmap... and why wiring the planet could make or break AI’s future (and ours).
Earlier this week, ENACT convened for its second major gathering, its first on U.S. soil, bringing senior executives from Masdar, ADNOC, Exxon Mobil, BP, XRG, OpenAI, Crusoe, Microsoft and a handful of large-cap utilities into one room with a single mandate: rewrite the wiring diagram of America’s digital future.
Founded in late 2024, ENACT (Energy-AI Nexus Action Catalyst) began as an Abu Dhabi majlis where eighty cross-sector leaders hammered out a five-point plan for the AI-energy nexus.
Since then the platform has evolved from a one-off dialogue into a standing “do-tank” that weds Emirati inclusivity with hard-hat pragmatism, translating talk into integrated roadmaps for power, policy and capital.
A few quick takes from their report:
⚫️ U.S. data-center compute could add 45 GW by 2030, a load equal to 30 million households, demand that would need either 50 GW of new gas or 150 GW of renewables if nothing else changes.
⚫️ Keeping AI’s lights on will require ≈ $700 billion a year in the U.S. and $1.5 trillion globally through mid-century, forcing a rethink of 15-year offtake models and sluggish permitting regimes.
⚫️ Nearly 2,600 GW of projects languish in U.S. interconnection queues, while half the nation’s transformers are already past mid-life.
ENACT’s Working Roadmap:
🗓️ Near-term (2025-2030) priorities:
• Sweat existing generation harder; delay retirements and re-rate assets.
• Unlock stranded capacity with dynamic line rating, advanced conductors and storage.
• Make demand flexible: task-shift compute, mainstream demand-response.
• Protect households via transparent beneficiary-pays PPAs.
🗓️ Long-term (2030-2050) pillars:
• Site smarter with a Data-Center Readiness Index.
• Build a backbone grid that moves electrons like fiber moves photons.
• Stand up a smart energy economy where buildings, mobility and industry absorb 60-70 % of new load.
• Accelerate next-gen tech: AI-driven materials discovery, SMRs, even fusion.
A few of my thoughts...
🤔 Nested feedback loops, not linear upgrades
Every new block of compute doesn’t just draw a bit more electricity, it tugs on an entire chain of resources: more nickel and lithium to build the batteries that back up the servers, thicker copper to reinforce the transmission spine, extra cooling water (or sorbents) to mop up the waste heat. Ignore any one link and the strain ricochets somewhere else: higher mining tailings, transmission bottlenecks, neighborhood water shortages.
The practical takeaway: we need to frame expansions in “loops,” not silos. When a cloud provider adds a 100 MW data-centre, planners should already be lining up upstream critical-metal contracts, mid-stream grid upgrades, and downstream heat-re-use or water-recycling systems. Close the loop proactively, and you prevent the system from snapping back in the form of price spikes, supply crunches or political push-back.
🤔 Rebound vs. flywheel
Efficiency gains can cut two ways. History’s littered with Jevons rebounds: cheaper steam engines drove more coal use; LED bulbs cut energy per lumen yet boosted total lighting demand. AI is no different. If inference costs fall but price signals lag, we just spin up bigger models and burn the savings.
Flip the script by tightening the feedback between electrons consumed and the carbon they embody. Hour-ahead carbon-intensity pricing, granular RECs, or on-site renewables that settle in real time mean the cheapest watt is also the cleanest. That price clarity turns efficiency into a flywheel: money saved flows into more renewables, which drop the marginal cost further, which attracts even more clean load. Lose the latency and abundance wins.
🤔 Transmission as social technology
Stacking steel into the sky is a solved engineering problem; convincing forty counties and a dozen regulators is not. Right-of-way battles, permit delays, and “not-in-my-backyard” (NIMBY) lawsuits add more time to a line than welding and tensioning ever will. In that sense, a high-voltage corridor is as much a social artifact as a physical one.
Solutions live in the people stack: AI-assisted route mapping that avoids sacred or high-value land in the first draft, community co-ownership models that share line revenue with locals, and federally pre-cleared “sandbox corridors” where planners can fast-track demonstrator lines. (‘Sandbox corridors’ can be modeled off of the Dubai-based RegLab initiative I had witnessed) Turn potential opponents into shareholders and the NIMBY friction morphs into community pride.
🤔 Strategic slack beats brittle efficiency
A grid run on razor-thin reserves is like a factory that keeps zero inventory: great for Wall Street metrics, terrible when a ship blocks the canal. Climate-amplified heat waves, cyber-attacks, or a sudden generation trip can all cascade into rolling blackouts if there’s no buffer.
Inject slack through modular micro-grids that can island in seconds, seasonal storage (hydrogen, pumped hydro, even ammonia) that time-shifts surplus wind or spring snowmelt, and rapid-start peakers (batteries, turbines, or engine-generators) that shoulder the first ten minutes of a disturbance. Slack isn’t waste; it’s hydraulic head that absorbs shocks and buys slower assets the time they need to respond safely.
🤔 Choose your attractor state now
Complex systems settle into repeating feedback loops. Leave today’s incentives untouched and AI growth naturally slides toward a high-carbon ruts: more gas peakers, life-extensions for aging coal, and export bans on the renewable supply chain.
But tweak two levers, real-time carbon-based pricing and siting that favors clean-rich regions and demand migrates toward renewable over-supply. Cheap surplus wind in West Texas, stranded hydro in Quebec, or midday solar gluts in the Gulf states become magnets for compute. Those extra gigawatt-hours fund even more renewables, deepening the basin of clean abundance until it becomes the easier path for every new workload. The choice is path-dependent so the best time to tilt the board is right now.
🔋 Rise of the Neo-Utilities
Think of neo-utilities as the grid’s bilingual translators, fluent in both kilowatts and teraflops. Unlike yesterday’s rate-base incumbents, these outfits bundle three roles under one roof:
Developer–Operator - They site renewables, batteries, and small-modular peakers right next to high-density compute (or growing communities that would have otherwise might have that similar energy demand), shaving line losses and interconnection wait-times to near zero.
Market Orchestrator - Using real-time nodal pricing and AI dispatch, they arbitrage between “buy electrons cheap, sell reliability” turning flexibility into a revenue stream rather than a regulatory burden.
Service Layer - They expose energy as an API for hyperscalers, edge clusters, and even municipal fleets, complete with latency windows and uptime SLAs.
Why does this matter? Because every strategy above, loop-closing, flywheel pricing, socialized transmission, strategic slack, needs a business entity that can monetize coordination instead of treating it as an externality. Neo-utilities are that entity. They win by collapsing transaction costs: fewer permits, tighter feedback loops, faster capital recycle. We’re already seeing early signals, merchant battery farms with 5-minute trading desks, data-center-plus-solar campuses run as one balance sheet, community micro-grids that pay dividends to local subscribers.
The fight isn’t “old utilities vs. new tech.” It’s static franchises vs. dynamic integrators. Back the players who treat electrons, bits, and balance-sheets as the same design material, and the rest of the roadmap suddenly looks less like science fiction and more like standard operating procedure.
Where We Go From Here
Is this the “grid-singularity moment”? AI’s appetite has revealed every hidden weld in our energy system, now we either reinforce those seams or watch the whole structure groan under the load.
ENACT’s roadmap is a solid blueprint, but blueprints don’t pour concrete. The next 12-18 months will decide whether the U.S. grid becomes a launchpad for clean abundance or a bottleneck that throttles innovation and drags us back to the fossil default.
So, if you’re…
A policymaker: give permitting the same urgency you’d give a national security waiver because it is one.
An investor: swap the 15-year PPA mindset for agile, asset-backed structures that recycle capital as fast as the technology cycles.
A builder or operator: treat every new megawatt as a multi-loop design challenge, upstream minerals, mid-stream wires, downstream reuse, not a single-line item.
A community leader or prosumer: demand a seat at the cap table. Dividend-sharing micro-grids and rooftop peer-markets are already flipping NIMBYs into stakeholders.
And for the rest of us systems nerds looking for leverage points: keep amplifying the price signals, shorten the feedback loops, and remind ourselves that electrons are political creatures long before they’re technical ones.
The future isn’t just written in code; it’s soldered in copper, brokered in permits, and financed in milliseconds.