On September 24, 2026, the U.S. Department of Energy selected 31 transmission projects across 26 states for $1.9 billion in federal funding under its Speed to Power through Accelerated Reconductoring and other Key Advanced Transmission Technology Upgrades program, SPARK. Recipients add $3.35 billion of their own money, for $5.25 billion in total. DOE says the work will reconductor or rebuild more than 1,500 miles of transmission line, deploy grid-enhancing technologies (GETs) across nearly 21,000 miles, and make more than 23 GW of additional electricity capacity available to roughly 100 million people.
The bet is that the fastest megawatt is the one already sitting behind an undersized wire. For anyone trying to energize a data center, it matters. It is also easy to misread. The 23 GW is transmission headroom, not generation, and the selections still have to become signed awards with project periods of up to four years.
This piece covers what was funded, what the numbers mean once you do the math, and how a buyer planning power for 2027–2031 should adjust.
Key Takeaways
- DOE selected 31 projects for $1.9 billion in federal funding (nominal 2026 dollars); recipients cover $3.35 billion, or 63.8% of the $5.25 billion total.
- At program level that works out to about $228 of total spend per kW of capacity DOE says it frees up, and about $83 per kW of federal money.
- The 23 GW is transfer capability on existing corridors. It moves power; it does not generate it, and it does not build your substation.
- DOE expects to sign awards between October 2026 and January 2027, with project periods of up to 48 months, so full delivery could run into late 2030 or early 2031.
- For data center developers, SPARK improves the odds in specific constrained corridors. It does not replace a speed-to-power plan for 2027 and 2028.
What DOE funded#
According to DOE's SPARK program page, SPARK is funded through the Grid Resilience and Innovation Partnerships (GRIP) program, which the Infrastructure Investment and Jobs Act authorized at up to $10.5 billion over five years. DOE opened the $1.9 billion funding opportunity on March 12, 2026, with concept papers due April 2 and full applications due May 20.
The solicitation was built around a narrow thesis: reconductoring (restringing existing towers with higher-capacity conductor) plus advanced transmission technologies such as dynamic line ratings and power-flow control. Holland & Knight's summary of the notice lists three topic areas, a 50% cost share (25% minimum for small utilities in Topic 1), and a project period of up to 48 months.
SPARK funding structure by topic area
| Topic | Eligible applicants | Federal pool | Expected award size |
|---|---|---|---|
| 1 — Grid Resilience | Utilities, grid operators, generators, fuel suppliers, transmission and storage owners | Up to $427 million | About $10 million to $100 million |
| 2 — Smart Grid | State and local governments, nonprofits, higher-education institutions, tribes | Up to $614 million | About $10 million to $50 million |
| 3 — Grid Innovation Program | States, tribes, local governments, public utility commissions | Up to $862 million | About $100 million to $250 million |
Pools and eligibility per Utility Dive's coverage of the March solicitation; award ranges per Holland & Knight. The three pools sum to $1.903 billion ($427 million + $614 million + $862 million).
Topic 3 is the one data center developers should read closely. DOE's program page says it focuses on "transmission systems that facilitate development of new large loads," with priority on multi-jurisdictional projects that expand transfer between planning regions. For data center developers, that is the topic written around their problem.
The projects we can name so far#
DOE posted a project list alongside the announcement but did not name recipients in the release body. Early reporting fills in part of the picture.
SPARK selections reported as of September 24, 2026
| Project | Lead | Scope | Reported funding |
|---|---|---|---|
| Three Corners Connection | Oklahoma Office of Management and Enterprise Services | High-capacity link between the Southwest Power Pool in Oklahoma and WECC in Colorado | $250 million federal requested; $832.17 million total |
| Montour Grid Resilience and Advanced Reconductoring | PPL Electric Utilities | Rebuild of about 29.3 miles of existing 230 kV corridor serving the Susquehanna Valley, Lehigh Valley and northeast Pennsylvania | Up to $71.5 million federal |
| Advanced transmission technology deployment | Colorado Energy Office | Interregional transfer capability across Colorado, Texas and neighboring regions | $1.2 billion total; federal share not disclosed |
The Oklahoma figures come from KTUL, the PPL figures from mGrid, and the Colorado figure from Utility Dive, which also names Alabama Power, Duke Energy Carolinas, Eversource Energy and Kit Carson Electric Cooperative among the selected utilities. CenterPoint Energy and DOE scheduled a joint Houston announcement for the same day; its dollar amount and scope were not public at the time of writing.
The Oklahoma project stands out. KTUL reports it would create a high-capacity link between the Eastern and Western Interconnections, which makes it an interregional transfer project. The $250 million federal request sits at the top of the Topic 3 award range and equals 13.2% of the entire SPARK federal pool ($250 million ÷ $1.9 billion). The non-federal share is 70% (($832.17 million − $250 million) ÷ $832.17 million = 69.96%).
Doing the math on 23 GW#
$228 per kW — total SPARK spend ($5.25 billion) divided by the 23 GW DOE says the projects make available.
The arithmetic: $5,250,000,000 ÷ 23,000,000 kW = $228.26 per kW. Federal money alone is $1,900,000,000 ÷ 23,000,000 kW = $82.61 per kW. Because DOE says "over 23 GW," the true figure per kW is at or below these numbers.
Do not compare that to what a generator or turbine costs per kW. It is not the same product. A kilowatt of transfer capability lets a kilowatt generated somewhere else reach a load somewhere else. The comparison that matters is against building a new line in a new corridor, and there the case for reconductoring is strong. A 2025 UC Berkeley and GridLab study published in the Proceedings of the National Academy of Sciences found the U.S. could double its transmission capacity by 2035 by reconductoring existing lines with advanced conductors. The same release notes that a new power line takes 10 to 15 years to build.
The headline also blends two different kinds of capacity. More than 1,500 miles get new or rebuilt conductor. Nearly 21,000 miles get GETs, which means sensors, software and power-flow devices that raise how much an existing line is allowed to carry under real conditions. That is a ratio of about 14 miles of GETs for every mile of new conductor (21,000 ÷ 1,500 = 14). mGrid's analysis flags exactly this: new conductor adds firm capacity, while dynamic line ratings and power-flow control release headroom that exists only when conditions allow. DOE published one combined number and has not said how the 23 GW splits between the two.
For a planner, that distinction is the whole story. A line rating that rises on a cool, windy night does not help a 24/7 load at 4 p.m. in August.
Why this matters for data center load#
17–46 GW — the increase in average data center power draw implied by DOE's 2028 range, versus 2023.
The 2024 Berkeley Lab report DOE released put U.S. data center consumption at 176 TWh in 2023, or 4.4% of national electricity, and projected 325 to 580 TWh by 2028, or 6.7% to 12%.
Converted to average load:
Data center electricity use expressed as average power draw
| Year | Consumption | Average draw (TWh × 1,000 ÷ 8,760 h) |
|---|---|---|
| 2023 | 176 TWh | 20.1 GW |
| 2028, low case | 325 TWh | 37.1 GW |
| 2028, high case | 580 TWh | 66.2 GW |
The increment is 37.1 − 20.1 = 17.0 GW in the low case and 66.2 − 20.1 = 46.1 GW in the high case. SPARK's 23 GW sits inside that range. Treat that comparison as a rough scale check: transfer capability and load growth are different quantities, the SPARK corridors are not all where data centers are going, and none of the capacity is dedicated to large loads.
What SPARK does change is the risk profile of specific sites. If your target substation sits behind a thermally constrained 230 kV or 345 kV corridor that just got funded, the utility's answer to "when can you serve 200 MW?" may improve. If it does not, nothing in this announcement moves your date. The SecondWatt data center intelligence library, which covers 5,884 facilities across 26 markets, is a starting point for seeing where existing capacity is concentrated before you overlay the funded corridors.
The timeline problem#
48 months — the maximum SPARK project period, starting from awards DOE expects to sign between October 2026 and January 2027.
Selections are not awards. DOE announced its "intention to help fund" these projects, and negotiations follow. Using DOE's schedule, a project awarded in January 2027 with a full 48-month period finishes in January 2031. Some reconductoring jobs may finish sooner, since the towers and easements already exist. But the planning-grade assumption for 2027 and most of 2028 is that SPARK capacity is not yet in service.
Compare that with the interconnection queue. Berkeley Lab's Queued Up 2026 edition found the median time from interconnection request to commercial operation was over five years for projects built in 2025, and only 13% of capacity that requested interconnection from 2000 to 2020 had reached operation by the end of 2025. Natural gas capacity in the queue grew 86% in 2025, to 253 GW.
SPARK shortens the transmission leg. It does not shorten the generation queue, the large-load study process, or the substation build. SecondWatt's interconnection roadmap for data center power walks through each of those legs, and our coverage of FERC's large-load interconnection rulemaking covers the parallel regulatory track.
What reconductoring does not fix#
Three constraints sit outside SPARK's scope and still set most data center energization dates.
Substations and transformers. A reconductored line still terminates at a substation that needs capacity, breakers, and large power transformers. Wood Mackenzie projected a 30% supply deficit for power transformers in 2025, with demand up 116% since 2019 and about 80% of U.S. supply imported. Our transformer sizing and lead-time guide covers what that means at the site, and the transformer equipment catalog lists specific units.
Generation. More wire moves more power only if there is power to move. In regions short on firm capacity, a bigger corridor raises the ceiling without raising the supply.
Ratings rules already in motion. FERC Order 881, issued in December 2021, already requires ambient-adjusted ratings for near-term transmission service, while leaving dynamic line ratings voluntary. When a utility cites new headroom on your corridor, ask how much comes from SPARK hardware and how much from ratings changes that were already required.
How to plan around SPARK#
A practical sequence for a developer, EPC or operator evaluating a site today:
- Get the project list and map it. Identify whether any funded corridor feeds the substation you are targeting. If one does, ask the utility directly how the upgrade changes its large-load study assumptions.
- Ask which kind of capacity it is. Reconductoring on your corridor is closer to firm. A dynamic line rating is conditional. Get the utility to say which.
- Keep the date you would have had without it. Treat SPARK as upside on a 2029–2031 grid date and keep your base case unchanged.
- Price the gap. The years between site readiness and firm grid service are where bridge and on-site power earn their keep. That decision is technology-neutral: aeroderivative turbines, reciprocating engines, solid oxide fuel cells, diesel generators, and solar plus storage all fit some sites and fail others.
The SecondWatt Power Intelligence Tool scores all five technologies against your project inputs, including revenue per kW-month, so the bridge decision rests on your numbers, not on a vendor's pitch. For deeper context on the options, see our pieces on fuel cell time-to-power benchmarks against gas turbines and 2026 backup power lead times, or browse the gas turbine catalog.
Frequently asked questions#
What is the DOE SPARK program?#
SPARK stands for Speed to Power through Accelerated Reconductoring and other Key Advanced Transmission Technology Upgrades. It is a $1.9 billion funding round from DOE's Office of Electricity, paid for through the GRIP program, that funds reconductoring and grid-enhancing technologies on existing transmission corridors.
How much funding did DOE announce on September 24, 2026?#
DOE selected 31 projects for $1.9 billion in federal funding. Recipients contribute $3.35 billion, for $5.25 billion in total. These are selections; final awards are expected between October 2026 and January 2027.
Does SPARK add 23 GW of new power generation?#
No. The 23 GW is transmission capacity DOE says will become available on existing lines. It lets more power move between generators and loads. It does not build generating plants.
When will SPARK projects be finished?#
Project periods run up to 48 months from award. With awards expected through January 2027, the latest projects could finish in late 2030 or early 2031. Individual reconductoring jobs may finish sooner.
Will SPARK speed up data center interconnection?#
Only where a funded corridor relieves the specific constraint limiting a site. It does not change generation queue timelines, large-load study processes, or substation and transformer lead times.
What is reconductoring?#
Reconductoring replaces the wires on existing transmission towers with higher-capacity conductors, often composite-core designs that run hotter with less sag. It raises line capacity without acquiring new right-of-way.