Data-Center Load Growth, Grid Bottlenecks, And Self-Generation/Off-Grid Pathways
Sources: 1 • Confidence: Medium • Updated: 2026-03-02 12:58
Key takeaways
- A key watch item is how much new large load (especially data centers) will choose to go fully off-grid with no intent to interconnect.
- Self-driving cars are already operating in several cities.
- Over the next decade, it is uncertain whether utility reform, permitting reform, and transmission buildout will reduce electricity prices in an enduring way.
- Within the next five to ten years, the trajectory of industrial electrification as a viable decarbonization pathway will become clearer and could appear to stall or fail.
- In fast-growing battery demand regimes, recycled material supply structurally reflects demand from roughly a decade earlier and cannot fully solve near-term mineral supply constraints until growth plateaus.
Sections
Data-Center Load Growth, Grid Bottlenecks, And Self-Generation/Off-Grid Pathways
- A key watch item is how much new large load (especially data centers) will choose to go fully off-grid with no intent to interconnect.
- Several uncertainties could materially change data center electricity demand, including whether architectures shift from very large centralized facilities toward more distributed edge data centers and whether chip efficiency improves enough to reduce projected buildout.
- Even if near-term data center demand forecasts are overstated, economy-wide electrification of transport, buildings, and industry still implies a need for much more electricity and a stronger grid regardless.
- "Bring your own generation" for data centers is already occurring at enormous scale, distinct from fully off-grid operation.
- If transmission and grid capacity cannot be built fast enough, around-the-edges solutions may emerge that spur unanticipated innovation and new infrastructure patterns.
- David Roberts says he is observing many large gas plants being built right now to serve rising power demand.
Autonomous-Vehicle Deployment And Electrified Fleet Implications
- Self-driving cars are already operating in several cities.
- Autonomous ride-hailing vehicles have become normalized in San Francisco, and David Roberts claims roughly one out of every three cars there is a Waymo.
- David Roberts asserts that self-driving cars currently operating on roads (including Waymo) are pure battery-electric vehicles.
- Within five to ten years, autonomous vehicles will be widespread enough in multiple cities to provide directional evidence on effects like total vehicle miles traveled, congestion, and sprawl.
- David Roberts predicts his four-year-old son will never need a driver's license because autonomous ride services will be sufficient by the time he reaches driving age.
Electricity Affordability As The Binding Constraint On Electrification Politics
- Over the next decade, it is uncertain whether utility reform, permitting reform, and transmission buildout will reduce electricity prices in an enduring way.
- Rising electricity prices are in direct tension with rapid electrification across industry, buildings, and transportation and make affordability the central political constraint.
- Electricity prices are likely to continue rising over roughly the next three years rather than falling.
- Affordability pressures are likely to align diverse stakeholders toward deploying more clean generation and batteries because these resources are perceived as among the cheapest options even without climate motivations.
Industrial Electrification Viability Under Interconnection Scarcity And Price Pressure
- Within the next five to ten years, the trajectory of industrial electrification as a viable decarbonization pathway will become clearer and could appear to stall or fail.
- Rising competition for grid-connected large-load sites from data centers could undermine industrial electrification by making it harder or more expensive to site electricity-intensive industrial facilities.
- Grid expansion needs driven by broad electrification should be kept analytically separate from objections that are specifically about data center growth.
- Competition for scarce multi-hundred-megawatt interconnection sites and upward pressure on electricity prices can erode the operating-cost advantage that underpins industrial electrification via the spark spread.
Recycling As Strategic Mineral Supply With Structural Lag And Partial Mitigations
- In fast-growing battery demand regimes, recycled material supply structurally reflects demand from roughly a decade earlier and cannot fully solve near-term mineral supply constraints until growth plateaus.
- Battery chemistry improvements can partially offset recycling’s volume lag because recovered minerals from older batteries can support more capacity in newer batteries.
- Clean-tech recycling is expected to shift in five to ten years from a peripheral environmental practice to a strategically important source of critical minerals tied to national and energy security.
- Recycling may become the cheapest and primary source of some critical minerals because it can avoid parts of the costly separations required in virgin mining.
Watchlist
- A meaningful scaled SRM geoengineering demonstration is a plausible future development and could raise moral-hazard concerns by normalizing low-cost climate intervention.
- A key watch item is how much new large load (especially data centers) will choose to go fully off-grid with no intent to interconnect.
- Several uncertainties could materially change data center electricity demand, including whether architectures shift from very large centralized facilities toward more distributed edge data centers and whether chip efficiency improves enough to reduce projected buildout.
- As homes, vehicles, and distributed energy resources become software-mediated, platform lock-in could lead to intrusive ads, subscriptions, and exploitative pricing unless interoperability and privacy rules are established.
- Within the next five to ten years, the trajectory of industrial electrification as a viable decarbonization pathway will become clearer and could appear to stall or fail.
- Rising competition for grid-connected large-load sites from data centers could undermine industrial electrification by making it harder or more expensive to site electricity-intensive industrial facilities.
- Over the next decade, it is uncertain whether utility reform, permitting reform, and transmission buildout will reduce electricity prices in an enduring way.
- Permissionless distributed energy resources such as plug-in balcony solar are rapidly moving toward legalization in the U.S. and may have outsized social and political effects even if total megawatts are modest.
Unknowns
- What is the measured penetration (fleet size, rides, VMT share) of autonomous ride-hailing in early markets like San Francisco, and how quickly is it expanding to other cities?
- Are currently deployed AV fleets predominantly battery-electric across operators and geographies, and will that remain true as AV deployment scales?
- How many large loads (especially data centers) will choose full islanded operation (no grid interconnection) versus grid-connected operation with substantial on-site generation?
- Under realistic reliability targets, what mix and scale of dispatchable generation is required for a fully off-grid, high-utilization data center, and how sensitive is the answer to site solar resource and storage duration assumptions?
- Will AI compute architecture trends favor continued centralization (large training clusters) or shift materially toward distributed/edge deployment, and what are the net implications for grid congestion hotspots?