LatticeLog

  • Governance
  • Infrastructure
  • Markets
  • Analysis
  • Notes
  • Signals
  • Cycles
  • Learnings
  • Commentaries
  • Home

Written by Nithinraj Kooneri

in Bifrost Systems, The Forge

I · FoundationsII · InflectionIII · Re-ratingIV · The Other Grids
Fenrir Research · US Power Markets · Part II of IV

US Power Markets: Inflection

Demand, resilience & the grid of 2035 — the flat-demand consensus is dead, the grid is hardening against an unstable climate, and the next decade will be defined by which scenario wins.
BOTTOM LINE UP FRONT

US electricity consumption is forecast at 4,250 BkWh in 2026 (+1.3% YoY) and 4,382 BkWh in 2027 (+3.1%) per the May 2026 EIA Short-Term Energy Outlook. Commercial sector demand — including data centres — will surpass residential consumption for the first time in 2026. BloombergNEF projects US data centre power demand reaching 106 GW by 2035; the IEA’s Lift-Off case sees global data centre demand exceeding 1,700 TWh by 2035.

PJM capacity prices repriced violently in response: from $28.92/MW-day in 2024/25 to $329.17/MW-day for 2026/27 and 2027/28 at the FERC-approved price cap. Hyperscalers have committed over 9 GW of nuclear PPAs at premium economics. The grid is being hardened — covered conductors, undergrounding, dynamic line ratings — at the largest sustained capex pace in modern history. COP30 in Belém failed to agree a fossil fuel transition roadmap; the narrative has shifted from net-zero to energy abundance.

We close Part II with three scenarios for the US grid of 2035 — AI Abundance, Climate-Led Decarbonisation, and the most likely Hybrid Resilience pathway. Part III translates the framework into portfolio positioning.

PART II · CONTENTS
01The demand shock — data centres, electrification, manufacturing
02PJM capacity escalator & the co-location debate
03The new generation stack
04ENSO, wildfires & storms — the climate overlay
05Grid resilience capex — undergrounding, covered conductors, DLR
06From net-zero to energy abundance — the narrative shift
07The grid of 2035 — three scenarios
GGlossary additions (cumulative from Part I)

01 · The demand shock — data centres, electrification, manufacturing

From 2005 to 2022, US electricity demand was essentially flat. Efficiency gains offset population and GDP growth almost perfectly. That equilibrium broke in 2023 and has been shattering ever since. The May 2026 EIA Short-Term Energy Outlook forecasts US electricity consumption of 4,250 BkWh in 2026 (+1.3% YoY) and 4,382 BkWh in 2027 (+3.1%) — a step-change from the prior decade.

US electricity consumption — the inflection Source: EIA STEO May 2026, historical EIA data 3,500 3,800 4,100 4,400 4,700 BkWh 2010 2017 2024 2027f ~Flat: 17 years of efficiency offsetting population + GDP Inflection 2023 4,382 +3.1% YoY

Three vectors drive the inflection. Data centres are the largest and fastest-moving. BloombergNEF projects US data centre power demand reaching 106 GW by 2035; the IEA Base Case sees global data centre electricity demand reaching 945 TWh by 2030 and 1,200 TWh by 2035, with the Lift-Off scenario exceeding 1,700 TWh. The EIA forecasts US commercial sector electricity demand growing 2.2% in 2026 and 5.3% in 2027, surpassing residential consumption for the first time. Electrification of transport and heating is a slower but cumulatively large second vector: EV adoption has put incremental load on coastal urban grids; heat pump deployment is adding measurable winter peaks across the Northeast. Re-shored manufacturing — CHIPS Act fabs, battery gigafactories, and chemical plants — is the third, with each large semiconductor fab consuming 100–300 MW of continuous load.

The geography of the shock — uneven by RTO

The demand shock is not evenly distributed. FERC data show MISO experienced 43% annual growth in transmission service requests since 2020. PJM’s data-centre-rich Northern Virginia footprint has absorbed the largest absolute load increase. ERCOT, the Southwest Power Pool, and the Southeast — particularly Georgia’s Atlanta corridor — are seeing rapid growth. CAISO, NYISO, and ISO-NE are growing more slowly, partly reflecting their lower data-centre intensity and partly tighter siting environments.

Region Demand driver concentration Anchor utilities
PJM (Northern VA)Data centres: 5,100 MW load growth in 2027/28 auction; 70% of global internet traffic flows through regionDominion (D), AEP (AEP), Exelon (EXC), Constellation (CEG)
ERCOT (Texas)Data centres + crypto + LNG + manufacturing; 40+ GW pipelineVistra (VST), NRG (NRG), Oncor (Sempra), CenterPoint (CNP)
SERC (Atlanta)Data centres + manufacturing reshoring; 50 GW potential pipelineSouthern (SO), Duke (DUK), TVA (federal)
MISO43% YoY transmission request growth; manufacturing + dataXcel (XEL), Entergy (ETR), DTE (DTE), Ameren (AEE)
WECCData centres (Phoenix, Reno); EV electrificationNextEra (NEE) Arizona, PCG, EIX, PNM, IDA

02 · The PJM capacity price escalator & the co-location debate

PJM Interconnection — the largest wholesale electricity market in North America, serving 67 million people across 13 states — has become the empirical proving ground for the data centre demand shock. The price signal has been violent.

PJM Base Residual Auction clearing price ($/MW-day) Source: PJM Interconnection, Monitoring Analytics $0 $100 $200 $300 $400 $28.92 2024/25 $269.92 2025/26 $329.17 2026/27 $329.17* 2027/28 * Auctioned at FERC-approved price cap for two consecutive years

From $28.92/MW-day in 2024/25 to $329.17/MW-day in 2026/27: a tenfold escalation in two years. PJM’s independent market monitor estimates data centres drove 63% of the 2025/26 auction price increase and accounted for 40% of total capacity costs across the last three auctions. NRDC estimates cumulative cost to PJM ratepayers through 2033 at $100–$163 billion. Q1 2026 total wholesale power costs across PJM reached $136.53/MWh, up 76% year-on-year. Capacity costs alone rose 398% in the quarter. Average PJM household bills are projected to rise by approximately $70/month by 2028.

The co-location regulatory debate

Co-location — connecting a large load like a data centre directly to a generator at the same site, bypassing the broader transmission grid — is the most contested regulatory question in US power right now. The advantages are speed (a co-located deal can be operational in 18–24 months vs. 5–7 years for grid interconnection) and reliability. The cost is that the load no longer pays for the shared transmission system that ultimately backstops it.

In December 2025, FERC directed PJM to establish new pathways for co-location and load flexibility that protect reliability and affordability for other consumers. The Amazon–Talen Energy deal at Susquehanna Nuclear Power Station — for 1.92 GW of behind-the-meter nuclear power — was the test case. Microsoft’s revived contract at Three Mile Island (rebranded Crane Clean Energy Center) for 837 MW is the highest-profile example. Meta’s 20-year deals with Constellation (1.1 GW) and Vistra (2.1 GW across three nuclear sites) follow the same architecture.

Hyperscaler PPA deal tracker

Hyperscaler Generator counterparty Asset / type MW
MicrosoftConstellation (CEG)Three Mile Island Unit 1 restart (Crane Clean Energy)837
AmazonTalen Energy (TLN)Susquehanna Nuclear — behind-the-meter1,920
MetaConstellation (CEG)Clinton Nuclear (Illinois), 20-year PPA1,100
MetaVistra (VST)Comanche Peak, Perry, Beaver Valley nuclear; 20-yr2,100
GoogleNextEra (NEE)Duane Arnold (Iowa) restart, target 2029615
GoogleKairos PowerFirst corporate SMR PPA (advanced reactor)~500
Meta–OkloOkloAdvanced reactor campus, target early 2030s1,200
Restart: privateHoltecPalisades nuclear restart (Michigan)837

Aggregate disclosed nuclear hyperscaler commitments now exceed 9 GW, with another 3–5 GW of gas and renewables PPAs announced. These contracts are typically 15–20 years at prices well above legacy wholesale economics — converting merchant generators with commoditised wholesale exposure into long-dated infrastructure operators.

FENRIR VIEW

Co-location is the most asymmetric structural shift in US power in three decades. The hyperscalers are signing 15–20 year PPAs at prices that lock in nuclear and gas generator economics for two capacity-market cycles. For merchant generators with existing dispatchable fleets (Constellation, Vistra, Talen, PSEG), this transforms commoditised wholesale exposure into long-dated infrastructure contracts. This is the single largest re-rating catalyst in the sector. We position around it explicitly in Part III.

03 · The new generation stack

Six technologies will define the next decade of US generation buildout. Each has a distinct economic profile, deployment timeline, and policy backdrop. We treat each in turn, weighted by realistic 2030 deployment scale.

Combined-cycle gas turbines remain the marginal unit in most US dispatch hours. NextEra has approved 10 GW of new gas builds. Vistra closed a $4.7 billion Cogentrix gas deal. GE Vernova has reportedly secured over 90% of its gas turbine production capacity through 2030. The structural constraint is the OEM oligopoly — GE Vernova, Siemens Energy, and Mitsubishi Power. New combined-cycle delivery timelines have extended from 24 months to 48–60 months. This is the binding constraint on data centre buildout.

Realistic 2030 deployment: 60–80 GW new gas capacity. The OEMs are the picks-and-shovels play — GE Vernova reported $2.4 billion in Q1 2026 data centre-related orders alone, exceeding all of 2025.

The existing fleet (94 reactors, ~95 GW) has gone from stranded asset to strategic prize in 36 months. The §45U Zero-Emission Nuclear PTC, preserved under OBBBA, provides a price floor; data centre PPAs provide an explicit price ceiling well above legacy wholesale economics.

Restarts. Three Mile Island (Constellation, 837 MW, target 2028). Duane Arnold (NextEra, 615 MW, target 2029). Palisades (Holtec, 837 MW). Cumulative: ~2.3 GW of legacy nuclear coming back online with hyperscaler backing.

Small Modular Reactors. NuScale’s 77 MWe NRC-certified design is the only fully approved SMR. Its partner ENTRA1 Energy is progressing planning for up to 6 GW of NuScale capacity with the Tennessee Valley Authority — potentially the largest nuclear deployment programme in US history. Kairos Power signed the first corporate SMR PPA with Google in August 2025. The DOE awarded $94 million in May 2026 across eight companies for Tier 2 site permits and supply chain. Realistic first-deployment timeline: 2029–2030 for NuScale; 2030–2032 for non-NuScale designs.

Total nuclear capacity 2030 forecast: existing fleet ~95 GW + ~2 GW restarts + ~1 GW first SMR units = ~98 GW vs current 95 GW. Modest in MW terms; transformational in earnings power.

US grid-scale battery storage capacity reached 38.1 GW by end-Q2 2025, up 63% YoY. ERCOT briefly overtook CAISO as the largest deployment market, ending 2025 at ~16 GW vs CAISO’s ~17 GW. On June 19, 2025, batteries supplied roughly 26% of CAISO’s evening peak — overtaking gas for that hour.

Two business models have emerged. CAISO storage earns through resource adequacy contracts and arbitrage against the solar duck curve. ERCOT storage operates as pure merchant — ancillary services and energy arbitrage. ERCOT’s RTC+B (Real-Time Co-Optimisation with Batteries) market design, live since December 2025, co-optimises batteries across energy and ancillary services. Rabobank projects ERCOT installed storage could exceed 70 GW by 2028.

The OBBBA preserves the §48E ITC for storage through 2034 but layers FEOC restrictions on Chinese content. Since global lithium cell supply remains China-dominated, compliance is a material near-term constraint. Realistic 2030 deployment: 150–200 GW of grid-scale storage.

Conventional hydrothermal geothermal is a 70-year-old industry concentrated in the western US (3.7 GW operating, mostly California and Nevada). The new story is enhanced geothermal systems (EGS) — applying oil-and-gas drilling techniques to engineer geothermal reservoirs in dry hot rock. The DOE’s Enhanced Geothermal Shot targets $45/MWh by 2035.

Fervo Energy’s Cape Station project in Utah (400 MW Phase 1) is the leading commercial deployment. Google signed a 24/7 carbon-free EGS PPA with Fervo for its Nevada operations in 2023, with multi-site expansion since. Eavor Technologies (closed-loop geothermal) and Sage Geosystems are scaling demonstration projects. Importantly: geothermal retained full §45Y/§48E PTC and ITC under the OBBBA, with the phase-out beginning only in 2034.

Realistic 2030 deployment: 10–15 GW. Below gas and storage in scale but the highest-quality renewable option for 24/7 firm power — ideally suited for data centre PPAs that demand round-the-clock zero-carbon supply.

Wind and solar combined produced more US electricity than coal in 2024 and 2025. Texas installed 7.4 GW of solar in the first nine months of 2025 — nearly double California’s additions. Most of the H1 2025 surge was policy-driven: developers rushing to begin construction before OBBBA deadlines bit.

The construction-start deadline (July 4, 2026) and placed-in-service deadline (December 31, 2027) for full §45Y/§48E credits create a binary: developers either move into the ground now or accept reduced or zero credit. Onshore wind faces the additional headwind that §45X domestic content credits (which support GE Vernova, Vestas’s US plants, TPI Composites) terminate after 2027.

Offshore wind is a separate, harsher story. In December 2025, Interior suspended leases for all five large-scale projects under construction: Empire Wind 1 (Equinor, >60% complete), Revolution Wind (Ørsted, >80% complete), Sunrise Wind, Vineyard Wind 1, Coastal Virginia Offshore Wind. All four developers have federal court challenges pending. US offshore wind is effectively closed for new business through 2028.

Carbon capture, utilisation, and storage (CCUS) is the most underdiscussed beneficiary of the OBBBA. The §45Q credit was raised from $60 to $85 per ton for carbon used in enhanced oil recovery (matching the rate for permanent sequestration). This is a political compromise: a clean-energy credit that flows to oil and gas operators.

Occidental Petroleum’s STRATOS direct-air capture facility (Texas) is the most prominent. Exxon’s Baytown blue hydrogen project depends on §45Q economics. For power, CCUS retrofits create a viable compliance path for gas plants in states with carbon constraints. Realistic 2030 deployment for power CCUS: 2–4 GW of retrofitted gas capacity — modest but tax-credit-rich.

04 · ENSO, wildfires & storms — the climate overlay

The US grid was built for one climate. It now operates in another. The reliability framework — NERC standards, RTO reserve margins, state IRPs — assumes a stationary distribution of weather events that no longer holds. Readers of our ENSO Primer (Part I of the Climate & Markets series) and ENSO Markets & Portfolio (Part II) will recognise the framework. We extend it here to US power infrastructure.

ENSO impacts on US power, by phase

ENSO phase US weather signature Power system impact
El Niño (warm phase)Drier Pacific NW, Ohio Valley, SE; wetter Southwest; suppressed Atlantic hurricanes; mild winter NEReduced Pacific NW hydro output (BPA); reduced TVA hydro; lower hurricane-driven outages on Gulf/Atlantic; lower winter peak NE
La Niña (cool phase)Wetter Pacific NW, Ohio Valley; drier Southwest/Texas; active Atlantic hurricane season; colder winter NE/MidwestHigher Pacific NW hydro; Texas drought stress; ERCOT cooling demand pressure; elevated Gulf/Atlantic outage risk; winter peak risk Midwest/NE
NeutralMixed signals; regional variability dominatesBaseline reliability planning conditions; ENSO signal weak

The 2025–26 weak El Niño cycle, currently forecast to transition toward neutral or weak La Niña conditions through 2027 per NOAA CPC and IRI/Columbia, sits in the medium-risk band for both Western wildfire and Atlantic storm activity. Two specific exposures matter for US utilities.

Wildfire risk — California, Pacific NW, Rockies

After PG&E’s Camp Fire liability (2018, ~$30 billion) and the January 2025 Los Angeles fires that implicated Southern California Edison, wildfire mitigation has become the defining capex programme for Western utilities. The mechanism is well-understood: utility equipment (sagging conductors, vegetation-touched lines, faulty insulators) sparks ignitions during high-wind, low-humidity events. Climate change has extended fire seasons across the West by roughly two months since 1980. Insurance markets for wildfire-zone homes are restructuring in real time.

Hurricane & storm risk — Gulf Coast, SE, Mid-Atlantic

Atlantic hurricane activity has trended higher since the mid-1990s; warmer sea surface temperatures provide more energy for storm intensification. Hurricane Beryl (July 2024) caused multi-day outages across CenterPoint Energy’s Houston territory affecting 2.2 million customers. Hurricane Helene (September 2024) caused historic flooding across the Carolinas. Hurricane Milton (October 2024) tested NextEra’s Florida hardening investments. The pattern of grid stress from compound climate events is now well-established; reliability standards and utility capex are adapting.

05 · Grid resilience capex — undergrounding, covered conductors, DLR

Utilities have responded to elevated climate risk with the largest sustained resilience capex programmes in their history. The mitigation toolkit has four principal technologies — undergrounding, covered conductors, dynamic line ratings (DLR), and Flexible AC Transmission Systems (FACTS). Each is rate-base eligible. Each translates directly to allowed-revenue growth.

Grid resilience toolkit — cost vs ignition reduction$/mile (M)Capex intensity$0$2M$4M$6MIgnition risk reduction →DLRDynamic Line Rating~$0.05M/mi · capacity gainFACTS (STATCOM/SVC)Voltage/reactive power managementCCCovered Conductor~$1M/mi · 67% reductionUGUndergrounding~$5M/mi · ~99% reduction

Covered conductors — the workhorse

Covered conductors replace bare overhead wires with insulated equivalents. PG&E has installed over 1,640 miles of system upgrades since its Community Wildfire Safety Program launched after the 2018 Camp Fire. The utility cites a 67% ignition risk reduction per circuit. SCE’s 2026–28 Wildfire Mitigation Plan calls for 440 additional miles of covered conductor. Average cost ~$1 million per circuit-mile.

Undergrounding — the gold standard

Undergrounding eliminates “nearly all” ignition risk on the circuit but costs 4–6× covered conductor — approximately $4–6 million per mile in PG&E’s territory. PG&E plans to underground 1,077 miles between 2026 and 2028, on top of the 1,250 miles already energised since 2021. SCE plans 260 miles. Both utilities have state-level political consensus around the programme; the prudency challenge at the PUC is low. Earnings durability is high.

Dynamic Line Rating (DLR) and FACTS — the capacity unlock

Dynamic Line Ratings measure real-time conductor temperature, ambient conditions, and wind to dynamically adjust the safe current limit on a transmission line — typically increasing usable capacity by 10–40% over static ratings. FERC Order 881 (effective 2025–26) requires transmission owners to implement ambient-adjusted ratings; FERC is now pursuing DLR mandates more broadly. FACTS (Flexible AC Transmission Systems — STATCOM, SVC, series compensators) manage reactive power and voltage stability to enable higher line loadings. Both are low-capex, high-throughput investments that increase grid utilisation without new line construction.

Hurricane & storm hardening — Florida, Texas, Gulf Coast

NextEra Energy’s Florida Power & Light has spent over $5 billion on storm hardening since Hurricane Wilma (2005), including substantial undergrounding of distribution feeders and concrete pole replacement. The result has been notably faster restoration after subsequent hurricanes — a regulatory virtuous cycle that translates to higher allowed ROEs at the Florida PSC. CenterPoint’s post-Beryl resilience plan calls for over $5 billion in distribution hardening across the Houston territory through 2030.

FENRIR VIEW

Climate adaptation capex is the most undervalued earnings driver in the US utility complex. It is rate-base eligible, politically supported across both red and blue states, and effectively countercyclical to broader macro stress. For PG&E, SCE, NextEra, CenterPoint, and Dominion, climate resilience programmes alone justify ~2–3% additional annual EPS growth on top of demand-driven capex.

06 · From net-zero to energy abundance — the narrative shift

“They won’t fear it until they understand it. And they won’t understand it until they’ve used it.”
— J. ROBERT OPPENHEIMER · OPPENHEIMER

The political frame around US power policy has shifted decisively. The 2020–22 framing was “net-zero by 2050” — energy policy as a subset of climate policy, with decarbonisation as the organising principle. The 2024–26 framing is “energy abundance” — energy policy as a subset of industrial and national security policy, with reliability, affordability, and AI competitiveness as the organising principles. This is not subtle and it has direct consequences for capital allocation. As we discussed in Part I’s coverage of the IRA-to-OBBBA reset, the legislative architecture has already adapted.

COP30 Belém — what didn’t happen

The 30th UN climate conference concluded in Belém, Brazil on November 22, 2025. The headline outcome: the formal text failed to include a roadmap to transition away from fossil fuels, despite 80+ countries advocating for one. Petrostate opposition (notably Russia, Saudi Arabia, India among others) blocked the inclusion. Two new initiatives — the Global Implementation Accelerator and the Belém Mission to 1.5°C — were launched as voluntary, parallel-track mechanisms.

What was achieved at COP30: a tripling of adaptation finance by 2035, a Tropical Forests Forever Fund ($5.5 billion raised, 53 participating countries), a Belém Health Action Plan, and a UNEZA Alliance commitment of $66 billion annually for renewable energy plus $82 billion for transmission and storage from public utilities. What was not achieved: a binding global fossil fuel transition timeline. For the US specifically, COP30 confirmed that international climate diplomacy is no longer a binding constraint on domestic energy policy — the Trump administration disengaged early and the EU was the primary advocate for the failed fossil fuel language.

The trajectory: COP28 → COP29 → COP30

Summit Headline outcome Direction of travel
COP28 (Dubai, 2023)UAE Consensus — first explicit call to “transition away from fossil fuels”; triple renewables, double efficiency by 2030High ambition
COP29 (Baku, 2024)New Collective Quantified Goal: $300B/year by 2035 for developing countries; $1.3T overall targetFinance-focused; ambition stalls
COP30 (Belém, 2025)Mutirão decision; fossil fuel roadmap omitted; adaptation finance tripled; voluntary tracks createdAmbition retreats
COP31 (Türkiye, 2026)Hosted under Türkiye presidency; Brazilian roadmaps to be reportedForthcoming

The geopolitical context matters. As we developed in War & Markets, the post-2022 fragmentation of the global trading system has revalued domestic energy security alongside cost. The COP30 failure to agree fossil fuel transition language is a consequence of the same petrostate-versus-importer divide we mapped in the geopolitical risk framework. Energy policy has become inseparable from national security policy.

What this means for US power

The “energy abundance” framing is not anti-climate; it is a re-prioritisation. Three operational consequences for US power investment:

  1. Coal retirements are being deferred, not cancelled. PJM reports 17 power plants have postponed retirement since the 2024 auction, retaining roughly 1,100 MW of capacity — mostly coal. The economic logic is straightforward: a coal plant clearing the capacity auction at $329/MW-day generates enough revenue to defer a $300M retirement decision.
  2. Permitting reform is moving. The administration’s executive orders on critical minerals, nuclear, and gas infrastructure permitting have materially reduced approval timelines for these technologies. Solar and wind permitting has tightened in parallel.
  3. State-federal divergence is widening. California, New York, Massachusetts, and others continue to enforce aggressive emissions targets; Texas, Florida, Wyoming, and others are explicitly anti-restriction. The same generator can have radically different earnings trajectories across states. State PUC composition becomes a critical equity research variable.

07 · The grid of 2035 — three scenarios

“The reaction may proceed catastrophically.”
— EDWARD TELLER · OPPENHEIMER

We have laid out the demand shock, the new generation stack, the climate vulnerability, and the political narrative shift. The natural next question — for any investor positioning for 2030 and beyond — is what the system actually looks like in 2035. The answer is genuinely contested.

We construct three scenarios, each internally coherent, each grounded in modelled studies or stated industry forecasts. They differ on two structural axes: the primacy of climate vs growth as the organising principle, and the willingness of capital and permitting systems to enable transmission expansion. Probabilities reflect Fenrir Research’s analytical judgement.

The grid of 2035 — three scenarios, two axesFenrir Research scenario frameworkTransmission & permitting capacity →Climate priorityConstrained transitionClimate-led decarbStatus quo driftHybrid resilienceCHybrid ResilienceP ~ 45% · base caseBClimate-LedP ~ 15%AAI AbundanceP ~ 40%

Three scenarios at a glance

Dimension A · AI Abundance B · Climate-Led Decarb C · Hybrid Resilience
2035 generation~5,500 BkWh; gas + nuclear-led~5,200 BkWh; 100% clean~5,300 BkWh; ~55% clean
Gas share42% (modest decline)<5% (CCUS or retirement)28% (declining bridge fuel)
Nuclear share20% (restarts + SMRs + new build)15–20% (doubled in constrained NREL case)22% (restarts + SMRs prioritised)
Wind+solar share22% (constrained by OBBBA)60–80% (NREL least-cost)35% (deploy where economic)
Geothermal/hydro8% (EGS unlocked)8–10% (24/7 firming)10% (anchor for 24/7 demand)
Storage capacity~200 GW~400 GW + LDES~280 GW
Transmission build1.3× current1.3–2.9× current (NREL)1.6× current
Incremental cost$1.2T (capex + fuel)+$330–740B over reference$1.6T (capex + transition)
Power-sector CO₂~50% below 2005100% reduction (net-zero)~70% below 2005
Probability~40%~15%~45% (base)

Premise. The “energy abundance” narrative wins decisively. Federal policy prioritises hyperscaler power needs, manufacturing reshoring, and national-security-driven AI competitiveness. State-level emissions policies become a heterogeneous patchwork. The OBBBA framework persists or is reinforced. Coal retirements continue to be deferred. New gas combined-cycle gets fast-tracked permitting. SMRs and nuclear restarts receive sustained policy and capital support. Wind and solar continue to deploy where economic but face structural headwinds.

2035 system features. Total generation ~5,500 BkWh — meeting all demand growth. Gas combined-cycle remains the dominant generator at 42% share. Nuclear expands to 20% via restarts (~2 GW), the existing fleet, first SMR deployments (~5 GW), and possibly large-scale new build at TVA. Wind+solar plateau at 22% (vs current 18%) — deployed but constrained. Enhanced geothermal scales to 5–8% of generation. Storage at ~200 GW absorbs variable renewable output and provides firming.

Emissions outcome. Power-sector CO₂ falls ~50% below 2005 levels — meaningful but well short of net-zero. Gas displaces residual coal; nuclear and renewables displace some gas; but absolute fossil generation rises in MWh terms as the system scales.

Equity implications. Strongest scenario for merchant gas IPPs (Vistra, NRG), gas turbine OEMs (GE Vernova, Siemens Energy), regulated utilities with data centre footprints (Dominion, Southern). Weakest for pure-play renewables developers (Sunrun, residential solar) and offshore wind. Nuclear merchants (Constellation, Talen) remain strong. Probability ~40%.

Premise. A change in administration in 2028 reinstates aggressive federal climate policy. State coalitions (California, NY, MA, IL, NJ, Colorado, others) drive coordinated regional planning. The Biden-era target of 100% clean electricity by 2035 is revived with binding interim milestones. Permitting reform unlocks transmission build at the scale modelled by NREL’s 2022 study. Capital flows back into wind, solar, and storage. Gas combined-cycle without CCUS is retired or curtailed.

2035 system features. Total generation ~5,200 BkWh (demand growth somewhat moderated by efficiency and demand-side flexibility). Wind and solar combined provide 60–80% of generation — per NREL’s least-cost modelling. Storage capacity scales to ~400 GW, with material long-duration energy storage (LDES) deployment. Nuclear expands modestly to 15–20% in the constrained NREL case (where new transmission is harder) or holds at 12–15% (where transmission expansion is easier). Conventional and enhanced geothermal hold at 8–10%. Gas without CCUS is <5% of generation; remaining gas plants run as peaking-only or with carbon capture.

Transmission build. 1.3× to 2.9× current US transmission capacity — requiring 1,400 to 10,100 miles of new high-capacity lines per year. This is the binding constraint. Without it, costs balloon and the constrained-transmission scenario dominates.

Cost. NREL modelled $330–$740 billion in additional power-system costs vs reference case over 2023–2035, with the higher end reflecting siting and transmission constraints. Health and climate benefits exceed costs in all modelled scenarios.

Equity implications. Strongest for renewables developers (NextEra Energy Resources, Avangrid), storage manufacturers (Fluence, Tesla, Sungrow alternatives), HVDC transmission EPC (Quanta, Mastec, Hitachi Energy), and clean firm power (geothermal, nuclear restarts). Weakest for gas IPPs without CCUS, coal-heavy generators, and gas turbine OEMs (though they would pivot to grid services and HVDC). Probability ~15% — the timeline is exceptionally tight even with full political alignment; the 2028 election outcome is the binary trigger.

Premise. Neither extreme prevails. Federal policy under successive administrations zigzags, but the underlying economic and physical drivers force convergence. Demand growth is real and persistent — utilities, RTOs, and state PUCs cannot afford to wait for political resolution. Climate adaptation capex (covered conductors, undergrounding, hardening) continues regardless of decarbonisation politics — it is rate-base eligible in every jurisdiction. Capital flows toward firm, dispatchable, low-carbon generation: nuclear restarts and SMRs, enhanced geothermal, pumped hydro, gas with CCUS retrofit, hybrid solar+storage with high storage ratios. Pure intermittent renewables continue to grow where economic but cease to be the policy darling.

2035 system features. Total generation ~5,300 BkWh. Generation mix: gas 28% (declining bridge fuel, increasing CCUS share), nuclear 22% (existing + restarts + early SMRs), wind+solar 35% (deployed where economic, especially Texas and the West), geothermal+hydro 10% (with EGS scaling materially), CCUS-equipped 2–4%, coal residual 1%, oil/other <1%. Storage capacity ~280 GW.

System characteristics. ~55% clean by 2035 (carbon-free generation including nuclear, hydro, geothermal, wind, solar). Power-sector CO₂ ~70% below 2005 levels — short of net-zero but materially lower than today. Reliability improves materially as the system shifts toward firm low-carbon resources. Transmission grows ~1.6× current capacity. Costs ~$1.6 trillion cumulative (capex + transition costs through 2035), distributed across rate-base recovery, merchant capacity payments, and hyperscaler PPAs.

Why this is the base case. Three structural forces converge: (1) capital is already flowing into firm low-carbon assets — hyperscaler PPAs are doing the policy work; (2) climate adaptation capex is non-political and well-supported; (3) the technology cost curves for storage, geothermal, and SMRs are improving fast enough that the economics dominate without policy push. State-federal divergence persists but the equilibrium clusters around mid-path system design.

Equity implications. Strongest for diversified utility names with all-of-above capex (NextEra, Dominion, Duke, Southern), nuclear-anchored merchants (Constellation, Vistra, Talen), grid equipment manufacturers (GE Vernova, Eaton, Quanta), and climate resilience plays (PG&E, SCE, CenterPoint). Cleaner-than-A but more dispatchable-friendly than B. Probability ~45% — this is our base case for Part III positioning.

2035 generation mix — three scenarios visualised

2035 US generation mix — three scenarios (% of total)Fenrir Research framework · NREL, BNEF, EIA inputs0%25%50%75%100%Gas 42%Nuclear 20%Wind+Solar 22%Geothermal+Hydro 8%Other 8%A · AbundanceP ~ 40%Wind+Solar 70%Nuclear 18%Geothermal+Hydro 9%Other 3%B · Climate-LedP ~ 15%Gas 28%Nuclear 22%Wind+Solar 35%Geothermal+Hydro 10%CCUS 5%C · Hybrid (base)P ~ 45%Gas (incl. CCUS)NuclearWind+SolarGeothermal+HydroOtherCCUS

The five enabling systems any 2035 grid must build

All three scenarios require — at different scales — the same five enabling systems. These are the actual deliverables behind the headline generation mix:

  1. Firm 24/7 zero-carbon capacity. Existing nuclear extensions, restarts, SMR rollout, enhanced geothermal, expanded pumped hydro. Required across all scenarios. Strongest scaling in B and C.
  2. Storage at three time horizons. Sub-hourly (frequency regulation, voltage support) via batteries; 4–8 hour (daily arbitrage and solar firming) via lithium-ion BESS; multi-day to seasonal (long-duration energy storage) via flow batteries, compressed air, iron-air chemistry, hydrogen, pumped hydro. B requires materially more LDES; C requires moderate LDES; A requires minimal LDES.
  3. Transmission expansion at 1.3×–2.9× current capacity. Long-distance HVDC links for moving wind from Plains to coastal load (especially in B); intra-regional reinforcement (all scenarios); offshore wind interconnection (limited in A and C). Permitting is the binding constraint.
  4. Resilience and adaptation. Undergrounding, covered conductors, dynamic line ratings, FACTS deployment, microgrid integration, advanced inverter capabilities. Rate-base eligible in every scenario; particularly intensive in California, Texas, Florida, the Carolinas.
  5. Demand-side flexibility. Time-of-use rates, demand response, virtual power plants (VPPs), data centre load shifting, EV smart charging. Increasingly cost-competitive with new generation; underweighted in current planning.
FENRIR VIEW

The scenarios differ on the surface mix but converge on the underlying enabling systems. Firm zero-carbon capacity, multi-horizon storage, transmission expansion, climate resilience, and demand-side flexibility are required in every plausible 2035 grid. The investable companies are those exposed to multiple of these enabling systems regardless of which generation-mix scenario plays out. This insight is the central thesis of Part III’s portfolio positioning.

Bottom line · what Part II established

The structural drivers from Part I have crossed into operational reality. Demand is rising at rates not seen since the 1970s. The PJM capacity market has repriced tenfold in two years. Hyperscalers have committed over 9 GW of nuclear PPAs at premium economics. Wildfire and storm risks have driven the largest sustained grid hardening capex programmes in modern history. COP30 confirmed that international climate diplomacy is no longer a binding constraint on US energy decisions. The narrative has shifted from net-zero to energy abundance.

Our three scenarios for 2035 give analytical structure to the next decade. AI Abundance (~40%) sees gas, nuclear restarts, and SMRs lead the buildout with modest decarbonisation. Climate-Led Decarbonisation (~15%) requires a political pivot in 2028 and 100% clean by 2035 per NREL’s modelling. Hybrid Resilience (~45%) — our base case — sees firm low-carbon capacity, geothermal scaling, and gas-with-CCUS converging at ~55% clean, ~70% emissions reduction. All three require the same five enabling systems: firm 24/7 zero-carbon capacity, multi-horizon storage, transmission expansion, climate resilience capex, and demand-side flexibility.

The market has begun to price this. The utility sector has been one of the strongest performers in the S&P 500 over the past 18 months. The merchant generators have re-rated more violently than any other sector. Part III takes this analysis and translates it into specific portfolio positioning across all three scenarios — the S5UTIL trajectory, the P/E expansion, the move from defensive bond proxies to growth narratives, and named anchor stocks across five distinct tracks. We also map the risks that could compress the trade.

G · Glossary additions (cumulative from Part I)

New terms introduced in Part II. Full glossary including Part I terms is at the end of Part I — Foundations.

“` “`
MARKETS & DEMAND
AEOAnnual Energy Outlook. EIA’s long-term US energy forecast, published annually.
BkWhBillion kilowatt-hours. EIA’s preferred unit for national consumption (= TWh).
BNEFBloombergNEF. Energy research and forecasting unit of Bloomberg.
Co-locationDirect connection of large load to generator at same site, bypassing grid interconnection.
STEOShort-Term Energy Outlook. EIA’s monthly two-year forecast.
VPPVirtual Power Plant. Aggregation of distributed resources providing grid services.
TECHNOLOGY & INFRASTRUCTURE
BPABonneville Power Administration. Federal Pacific Northwest hydro operator.
DLRDynamic Line Rating. Real-time conductor capacity assessment using temperature, weather.
EGSEnhanced Geothermal System. Engineered reservoirs in hot dry rock.
FACTSFlexible AC Transmission Systems. Power electronics for reactive power/voltage management.
FEOC / PFEForeign Entity of Concern / Prohibited Foreign Entity. OBBBA restrictions on China, Russia, Iran, DPRK.
LDESLong-Duration Energy Storage. Multi-hour to seasonal storage (flow batteries, compressed air, hydrogen, iron-air).
NRELNational Renewable Energy Laboratory. US DOE research institution; produces ReEDS capacity expansion modelling.
ReEDSRegional Energy Deployment System. NREL’s flagship capacity expansion model.
RTC+BReal-Time Co-optimisation with Batteries. ERCOT’s storage-aware market design.
STATCOM / SVCStatic Synchronous Compensator / Static VAR Compensator. FACTS device categories.
TVATennessee Valley Authority. Federal utility serving 10 million people in 7 SE states.
POLICY & INTERNATIONAL
COPConference of the Parties. Annual UN climate summit under UNFCCC.
NCQGNew Collective Quantified Goal. COP29 finance target: $300B/year by 2035 from developed countries.
NDCNationally Determined Contribution. Country-level climate plan under Paris Agreement.
UAE ConsensusCOP28 Dubai final text. First explicit call to “transition away from fossil fuels”.
UNFCCCUnited Nations Framework Convention on Climate Change. Treaty body convening COPs.
DATA SOURCES & REFERENCES

US Energy Information Administration (EIA) — Short-Term Energy Outlook (May 12, 2026); Annual Energy Outlook 2026 (April 8, 2026); Electricity Data Browser. BloombergNEF — US data centre demand outlook (Dec 2025). International Energy Agency (IEA) — Energy and AI report 2025 (Base, Lift-Off, and Headwinds cases). Brookings Institution — Global energy demands within the AI regulatory landscape (April 2026). FERC — quarterly transmission monitoring data; Orders 881, 1920, 2023. PJM Interconnection — BRA results 2024/25 through 2027/28. Monitoring Analytics LLC — independent market monitor reports. Inside Lines (PJM), Utility Dive, Canary Media, IEEFA, NRDC — capacity market and data centre demand analysis. National Renewable Energy Laboratory (NREL) — “Examining Supply-Side Options to Achieve 100% Clean Electricity by 2035” (2022); National Transmission Planning Study (2024). US Department of Energy — Enhanced Geothermal Shot, SMR Tier 1 and Tier 2 funding announcements. NuScale Power Corporation — SEC Form 8-K filings (FY2026 Q1). UN Climate Change (UNFCCC) — COP30 outcomes and Mutirão decision text. World Resources Institute, IISD, European Commission Climate Action — COP30 outcome analyses (November–December 2025). Pacific Gas & Electric Company — 2026–28 Wildfire Mitigation Plan. Southern California Edison — 2026–28 Wildfire Mitigation Plan. S&P Global Commodity Insights — US battery storage capacity reports (Q2/Q3 2025). NBC News, PBS NewsHour, Offshore Wind, Utility Dive — offshore wind regulatory actions (Dec 2025–Jan 2026). NOAA CPC, IRI/Columbia, ECMWF — ENSO outlooks. Fenrir Research prior publications: Power & Markets Part I — Foundations, ENSO Primer, ENSO Markets & Portfolio, War & Markets.

DISCLAIMER

This analysis is for informational purposes only. Not investment advice. All scenario probabilities and forward-looking judgements are analytical synthesis based on cited sources. Stock-specific references are illustrative and not buy or sell recommendations. The author and Fenrir Research may hold positions in securities mentioned.

FENRIR RESEARCH · YGGDRASIL LEDGER
POWER & MARKETS · PART II · LATTICELOG.IN · MAY 2026

←Dead Reckoning – W.E. 05/15
Dead Reckoning – W.E. 05/22→

Comments

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

More posts

  • A Billion Consumers: The Incentive

    June 5, 2026
  • A Billion Consumers: Consumption vs Productivity

    June 5, 2026
  • A Billion Consumers: From Scarcity to Aspiration

    June 5, 2026
  • <Note 1> Climate x Insurance

    June 3, 2026

LatticeLog

Structural research across markets, infrastructure, climate, and the systems that connect them. Published under Fenrir Research, a division of Yggdrasil Ledger.

  • Blog
  • About
  • FAQs
  • Authors

Twenty Twenty-Five

Designed with WordPress