Electrification
How fast electricity demand is growing, how much is being spent on the grid, how strained and how volatile wholesale power is, and how much of the car market has gone electric
Thesis
Is electrification outrunning the grid?
- Electricity demand grows faster than at any time since the 2000s.
- Grid investment is at a record but no longer accelerating.
- Price dispersion and volatility say the grid is the bottleneck.
Index
Five indicators, standardized and equal-weighted, updated monthly — 100 at the start of the common sample, higher means faster electrification and a tighter grid
Electrification Index
130.6
1 mo
↑ +5.0
1 yr
↑ +8.5
Source: Liquiqui, from EIA / Census / Argonne
Where each constituent stands today, in standard deviations from its own average over the sample
- Electricity demand growthMonthly · Jul 1, 2026+0.2
- Power construction growthMonthly · Jul 1, 2026+0.0
- Hub price dispersionMonthly · Jul 1, 2026+1.4
- PJM price volatilityMonthly · Jul 1, 2026+2.8
- Plug-in share of salesMonthly · Jul 1, 2026+0.8
Notes
- What is happening
- The index stands at 130.6 in July 2026, up from 125.6 in June, against a base of 100 in January 2017 and a record of 137.9 in January 2026, the month of the cold snap. It hit its all-time low of 88 in November 2020, then climbed in steps: 112 at the end of 2019, 115 on average in 2024, 121 in 2025. July is the cold snap's mirror image: the heat wave took PJM West's within-month volatility to 96% and its average price to $157. Today the two grid-strain measures dominate — volatility at +2.8 standard deviations and price dispersion at +1.4 — followed by the EV share (+0.8), while demand growth (+0.2) is barely above its average and investment (0.0) sits exactly on it.
- Why it matters
- Electrification shows in two places: in demand — how much electricity the economy consumes, how many cars plug in — and in the grid that has to deliver it — how much is invested in it, and how badly it strains. This index combines five measures at equal weight — demand growth, growth of grid investment, the dispersion of wholesale prices across regions, the volatility of the PJM price and the EV share of car sales — each standardised against its own history since 2017. It rises when the economy electrifies faster and the grid is tighter than usual.
- What's driving it
- Grid strain — record price dispersion in early 2026, record volatility in January and again in July — carries the index; demand and investment sit near their averages.
- What to watch
- The composition matters more than the level: an index carried by dispersion and volatility says the grid is the bottleneck, not demand. An index above 130 with investment back above its average would say supply is catching up; the same level with investment below average would say prices are doing the work instead of wires — July crossed 130 with investment exactly at its average, and prices did the work. The retreat of the EV share since the federal credit ended is the constituent pulling down in 2026.
Signals
The demand, the investment, the strain and the adoption — each scored by how far it has moved, and the arrow points the way it is heading
| Criterion | Current Reading | Trend | Comment |
|---|---|---|---|
Electricity demand growthin the index Is electricity consumption growing faster than it used to? | Moderate | ↑ | +1.3% over the twelve months to July 2026 for retail sales, +3.1% for grid demand over the 365 days to 10 October. From 2005 to 2019 US demand grew 0.6% a year on average: the pace has doubled, or more than quadrupled depending on the measure, on data centres and reindustrialisation. |
Grid investmentin the index How much is being spent building power infrastructure? | High | ↑ | $186 billion at an annual rate in August 2026, a record, against 138 in 2019 and 126 at the 2022 trough. The private sector carries 166 billion of it, the public sector 20 — more than double 2019. Growth is back up to +8.5% on the year, from +5% in the spring: strong in level, and the pace is picking up again without matching the 2023 surge (+30%). |
Grid congestionin the index Are wholesale prices diverging from one region to another? | High | → | Price dispersion across seven hubs hit a record 83% in April 2026, after 78% in January, before falling back to 15% in August, a seasonal floor: prices converge in summer when the whole country is hot at once. On average 2025 ran at 34% against 28% in 2024 — the grid is moving power between regions less and less well. |
Power price volatilityin the index Has the wholesale price become unpredictable from one day to the next? | High | ↑ | 36% in August 2026 for the within-month dispersion of PJM West's daily prices, after a record 134% in January 2026 during the cold snap. The annual average has gone from 18% in 2019 to 40% in 2024 and 47% in 2025: a grid running closer to its limits amplifies every swing in temperature. |
EV share of salesin the index How much of the new-car market plugs in? | Moderate | ↓ | 7.7% of light-duty vehicle sales in August 2026 for BEVs and PHEVs, 7.5% over twelve months, after 6.7% in July. The share had peaked at 12.9% in September 2025, the last month of the federal tax credit, then fell to 6.1% in December. On twelve months the 10.3% peak of September 2025 is eroding: adoption has gone back two years. |
Raw data
Electricity Demand Growth
Lower 48 demand, trailing year, year on year (daily)
3.21%
1 wk
↑ +0.10%
1 mo
↑ +0.02%
1 yr
↑ +0.67%
Retail sales, trailing year, year on year (monthly)
1.25%
1 mo
↓ -0.06%
1 yr
↓ -0.95%
Source: EIA
Notes
- What is happening
- +3.1% over the 365 days to 10 October 2026 for grid demand, +1.3% over the twelve months to July for retail sales. From 2005 to 2019 US demand grew only 0.6% a year on average, with negative years — −1.2% in 2019, −1.4% in 2023, two mild winters. Since 2024 the regime has changed: +2.6% in December 2024, +2.1% in December 2025. The daily series, being faster, runs ahead of the monthly one. The EIA's September outlook expects sales to rise almost 2% in both 2026 and 2027, to a record.
- Why it matters
- Electricity demand growth is the most direct measure of electrification: it adds up data centres, reshored factories, heat pumps and electric cars, and it is what the grid actually has to deliver. Two readings: daily grid demand for the Lower 48, on trailing 365 days, near real time; and the EIA's monthly retail sales, on trailing twelve months, which reach back to 2001 and feed the index.
- What's driving it
- Data centres, reshored factories and, more slowly, heat pumps and electric cars: load growth has doubled since 2023.
- What to watch
- The threshold is 2%: above it demand grows faster than the capacity the grid adds each year, and prices rise. Retail sales growth sustained above 2% would say data centres have changed the scale of the system, as air conditioning did in the 1960s. A relapse below 1% would say most of the announced projects have not been connected.
Grid Investment
Power construction spending, total
186 B$
1 mo
↑ +2 B$
1 yr
↑ +15 B$
Power construction spending, year on year
8.5%
1 mo
↑ +1.5%
1 yr
↑ +6.2%
Source: US Census Bureau, via FRED
Change on a year earlier
Source: US Census Bureau, via FRED
Notes
- What is happening
- $186 billion at an annual rate in August 2026, a record, against 138 in 2019, 126 at the 2022 trough and 171 in 2024. The private sector carries 166 billion, the public sector 20 — public spending has more than doubled since 2019 under the federal programmes of 2021 and 2022. Growth is back up to +8.5% on the year, from +5% in the spring, after +30% at the end of 2023 and +3% at the end of 2024: the level is high, and the pace is picking up again.
- Why it matters
- Grid investment is the supply-side answer: construction spending on power infrastructure — plants, lines, substations — which the Census measures every month, at an annual rate, public and private separately. It is the page's only measure that says whether the bottleneck is loosening. The year-on-year growth rate feeds the index, because the level has trended up for twenty years.
- What's driving it
- Utility rate cases and federal programmes funded the 2023-2024 surge; interconnection queues and permitting now slow the pace.
- What to watch
- The figure to watch is the gap between investment growth and demand growth: at +8.5% against +1.3%, supply is winning; at +3% against +2.5% last year, it was losing. Investment slowing below 3% while demand runs above 2% would mean interconnection queues are lengthening, and that the congestion and volatility below will keep rising.
Grid Congestion
Price dispersion across hubs (coefficient of variation)
14.6%
1 mo
↓ -40.3%
1 yr
↓ -1.3%
Dearest minus cheapest hub, $/MWh
23.0
1 mo
↓ -97.6
1 yr
↓ -3.4
Source: EIA (ICE)
Spread between hubs, monthly averages
Source: EIA (ICE)
Notes
- What is happening
- 15% in August 2026, a seasonal floor: prices converge in summer when demand is strong everywhere at once. But April 2026 set a record of 83% and January 78%, when PJM cost 180 dollars more than the cheapest hub. On an annual average, dispersion went from 28% in 2024 to 34% in 2025, and 2026 so far averages 60%. In 2019 the average gap between hubs was 65 dollars; in January 2026, 180.
- Why it matters
- A grid without congestion has one price: power flows from the cheap region to the dear one until prices equalise. When the lines are full, prices diverge. Dispersion — the standard deviation of seven hubs' monthly average prices, relative to their mean — therefore measures congestion without access to the grid operators' internal data. The prices come from the ICE files the EIA publishes; the gap between the dearest and the cheapest hub is given in dollars per megawatt-hour.
- What's driving it
- Interregional lines full while demand rises unevenly: PJM's data-centre load against a West with surplus solar.
- What to watch
- Dispersion is seasonal and a single month misleads: the figure that matters is the twelve-month average. Above 40% on an annual average, the US grid would be working as three separate markets — the East, Texas, the West — and investment in interregional lines, not in generation, would become the subject. A return below 25% would say the new lines in the Midwest and Texas have begun to deliver.
Power Price Volatility
PJM West, within-month volatility
36.0%
1 mo
↓ -59.6%
1 yr
↑ +2.7%
PJM West, monthly average price, $/MWh
73.2
1 mo
↓ -84.0
1 yr
↑ +25.9
Source: EIA (ICE)
PJM West peak price, monthly average
Source: EIA (ICE)
Notes
- What is happening
- 36% in August 2026, after a record 134% in January 2026 — the cold snap took PJM's average price to 192 dollars a megawatt-hour, against a 2019 average of 32 — and 96% in July 2026, at 157 dollars. The annual average of volatility has gone from 18% in 2019 to 25% in 2021, 40% in 2024 and 47% in 2025. PJM's average price has doubled since 2019: 64 dollars in 2025, 73 in August 2026.
- Why it matters
- Wholesale price volatility measures the grid's margin in time, as dispersion measures it in space: a system with spare capacity absorbs a heatwave without the price moving; a system at its limit sees the price rise tenfold in a day. The measure is the standard deviation of PJM West's daily prices within each month, relative to the month's mean. PJM is the largest US market and the one where Virginia's data centres are concentrated.
- What's driving it
- Thinner reserve margins in PJM, retiring coal plants and weather swings that a grid at its limit cannot absorb.
- What to watch
- An annual average volatility above 50% would mean the wholesale price has become a scarcity price, set by the last available megawatts rather than by the cost of gas. PJM's capacity auctions — whose price rose ninefold between 2024 and 2025 and has since sat at the $325 per megawatt-day cap for three auctions, the latest in July 2026 — are the other reading of the same phenomenon: a market paying for reserves it no longer has.
Methodology
1Sources
EIA (retail sales, ICE wholesale hub prices), Census via FRED (construction spending on power), Argonne National Laboratory (plug-in vehicle sales).
2Normalisation
Demand and investment as year-on-year growth, dispersion and volatility as coefficients of variation, EV share on trailing twelve months; monthly.
3Z-scores
Each constituent minus its mean since 2017, divided by its standard deviation; capped at ±3.
4Weighting
Five constituents, equal weights.
5Index
Average of the five z-scores × 20 points, rebased to 100 in January 2017. Updated monthly.
6Interpretation
Higher means faster electrification and a tighter grid; neither is a verdict on its own. Read demand and investment against dispersion and volatility.