# Putting the Agreement into Practice

### The roadmap from 2027 to 2035, and the tools already in hand

*A companion paper in the series "The Great Inversion." Bracketed numbers refer to the source list in Paper III. Sources for the toolbox are listed at the end.*

The program runs on two tracks, and nothing on the state track waits for Washington. Each stage has a test that proves it worked, and the transition protects existing contracts, customers, workers, and utility finances. The tools are already in hand: seventeen technologies that see, use, flex, and build grid capacity, three of them standard practice, and six rules on the roadmap that decide how far the rest spread.

---

## Why this window, and not five years ago or hence

The demand shock is present tense: data center and manufacturing load growth turned grid access into a boardroom problem for the most politically capable companies in the economy, which did not exist as a constituency in 2020. The technologies matured past the argument stage: storage is a commodity, aggregators run real portfolios, and Australian networks operate envelopes at scale, so "unproven" has lost most of its force. The toolbox lists seventeen such tools and where each one runs. The legal groundwork is already laid: Orders 2222, 2023, and 1920 sit on the books waiting for enforcement rather than enactment, which converts multi-year rulemakings into compliance fights. The failure is undeniable and priced: four consecutive cap-bound PJM auctions and bill growth that decides elections gave governors ownership of the problem, and reforms ride on visible failure, as 1907 and 1992 both showed. And capital is waiting on the sidelines for projects with financeable revenue and a predictable connection: infrastructure funds seeking contracted grid assets are the natural financiers of every reform in this program. Windows close. The predictable counter-reformation, incumbents offering rate-based mega-capex as the answer to load growth, is already assembling, and each approved conventional plan extends the old equilibrium's life by an asset's depreciation schedule.

## Implementation roadmap, 2027 to 2035

The program runs on two tracks. Under the Federal Power Act's 1935 division, Principles 1, 2, and 3 sit entirely within state authority, and states own the distribution half of Principle 4. Only Principle 5 requires FERC, though states hold real levers over it through Fixed Resource Requirement elections, large-load tariffs, and their own procurement. **Nothing on the state track waits on the federal track.** A commission that opens a docket in 2027 is not blocked by anything Washington does or fails to do.

### The state track

**2027.** Publish interconnection metrics and machine-readable hosting-capacity maps. Open a performance-regulation docket scoped to totex treatment, peer benchmarking, and shared savings. Approve a flexible-connection tariff so small resources connect inside published limits, and adopt the current inverter standard in interconnection rules. Open a large-load tariff proceeding that assigns capacity and network costs to new large loads.

**2028-2029.** First totex orders take effect, and shared-savings mechanisms make non-wires procurement profitable, which softens utility opposition to everything downstream. Automated screening goes live for small resources. Dynamic rates become the default, with hedged flat products freely available.

**2030-2031.** Where the escalation criteria are met, lead states separate access administration from the wires business; elsewhere ringfencing holds, on its benchmarks. Dynamic operating envelopes deploy at scale in high-DER territories.

**2032-2035.** Standard value-stack tariffs with scheduled, non-retroactive updates replace the net-metering successor fights. Yardstick benchmarks take over allowance-setting with five years of comparable data.

### The federal track

**2027.** FERC opens rulemakings on connect-and-manage as the default interconnection framework and on Order 2222 deadlines with consequences, turns its 2024 notice on dynamic line ratings into a rule, and enforces Order 1920 planning with reconductor-first showings. PJM's obligation-transition stakeholder process begins, with the 2026 auction record in hand \[16\].

**2028-2029.** Connect-and-manage is finalized for bulk generation and large loads. Large-load flexibility tariffs standardize the hyperscaler bargain. Minimum interregional transfer requirements are set.

**2030-2031.** The first hybrid delivery year runs a shrinking centralized residual auction alongside a growing share of demonstrated bilateral obligations, with cleared legacy commitments paid in full.

**2032-2035.** The centralized auction recedes to a residual backstop wherever obligation coverage passes the reliability-and-cost test, and stays wherever it does not. Distribution-level price formation opens for large participants where the revisit triggers say the substrate is ready.

### What proves each stage worked

Dockets open and data published by the end of 2027. By 2029, a utility earns more by buying a battery service than by building the substation it replaces. By 2031, new data center load carries its own demonstrated capacity. By 2035, queue conversion materially above the historical baseline, with median duration and withdrawal measured separately. A target: above 50 percent conversion for mature cohorts, not for the entire active queue.

Who moves first: three governors and their commissions, with FERC on a parallel track. Who benefits first: whoever is waiting in a queue, which is everyone the current system is failing. And any customer can push the first step along by speaking up at the next rate case.

## What success means

Economically valuable projects connect predictably, on a timeline they can finance against. Customers face lower lifetime costs, which is a different and more defensible claim than cheap power. Reliability holds or improves, measured on the standards already in force. Utilities remain financially healthy. And competition shifts investment toward the least-cost solution, whoever owns it.

The fourth of those is the one most often left unsaid. A well-run utility earns more under a framework that pays for outcomes than under one that pays for spending, which is what the British record shows and what the sequencing here is built around.

## The measures that prove it

By 2030: median small-resource interconnection measured in days, large-load and bulk-generation connection offers inside 18 months, at least five states under totex performance-based regulation, and the first delivery year in which new data center load carries its own demonstrated capacity. By 2035: queue conversion above half for mature cohorts, median application-to-energization under two years for viable projects, interregional transfer capability growing, capacity costs allocated to their drivers, and real retail bill growth in participating states below matched peer states. Each indicator is measurable annually, and Paper II's failure metrics (queue depth, auction outcomes, bill growth) serve as the control chart.

## The toolbox: what we already know how to do

Paper III states the program's test in four verbs: see how much capacity is actually available, use the grid we already have more fully, flex when and where power is made and used, and build only where building is actually needed. Each verb stands for a family of tools. The inventory here lists seventeen of them, with where each one runs and the rule it depends on.

*Status as of October 2026. The numbers on each entry are the principles that supply the rule.*

**1** Pay for performance · **2** Separate the referee · **3** Pay resources for what they do for the grid · **4** Connect by rule · **5** Make big new customers pay their own way

**Standard:** routine at most large utilities or grid operators. **In service:** in daily use in several places, not yet routine. **Early:** trials or first deployments.

### See: measure the capacity that is actually there

**Dynamic line ratings.** Rate a transmission line on measured wind, sun, and temperature, not on a fixed worst case. The Department of Energy puts the average gain at 10 to 30 percent. PPL Electric has fed these ratings into PJM's markets since 2022, on three lines for about $1 million. On one corridor they took the place of a reconductoring PPL had priced at about $12 million. National Grid in New York and Great River Energy in Minnesota have since put them into operation. *In service. Depends on: a payment formula under which a deferred line is not lost earnings, and a showing that cheaper options were weighed before new lines. Principles 1, 4.*

**Dynamic transformer ratings.** Rate a substation transformer on its measured temperature, not its nameplate. British trials from 2014 to 2020 measured ratings that ran from a little below nameplate to well above it. *Early. Depends on: the same payment formula. Principle 1.*

**Grid sensing and telemetry.** Meters, line sensors, and synchrophasors that show flows and voltages across the network. Advanced meters were 84 percent of US meters in 2024, and synchrophasors watch more than 2,500 points on the high-voltage system. By definition the meters report at least daily, which is short of real time. *Standard. Depends on: publication, so that what the sensors show is shared and not held. Principle 2.*

**Hosting-capacity analysis and state estimation.** Calculate, from meter data and a network model, what each local circuit can take. Utilities and agencies in 26 states publish hosting-capacity maps. Colorado ordered monthly updates by the end of 2026 and finer-grained analysis by the end of 2027. Real-time estimation runs inside a distribution management system, which one utility in five had fully working in a 2022 survey. *In service for maps, early for real-time estimation. Depends on: a publication requirement with an update schedule. Principles 2, 3.*

### Use: the grid we already have, more fully

**Topology optimization.** Software that reroutes power around a bottleneck by opening and closing breakers. MISO, the grid operator for the central states, takes reconfiguration requests from market participants and put the 2025 market benefit of those it carried out at $113 million. FERC approved a similar process at SPP, its neighbor to the southwest, in August 2026, and ERCOT plans one for 2027. *In service. Depends on: a neutral operator with a standing process that any participant can use. Principles 2, 4.*

**Advanced power-flow control.** Devices that push power off an overloaded line and onto neighboring lines with room. National Grid reports more than 2 GW of capacity freed at three substations in northern England. Georgia Power put 21 units on two 230 kV circuits in 2025, and Central Hudson freed 185 MW with 15. *In service. Depends on: a payment formula that does not penalize the cheaper fix, and a showing that these devices were weighed before new lines. Principles 1, 4.*

**Capacity analysis software.** Search the network hour by hour for room that a single worst-case study misses, and answer a connection request from data. Portland General Electric used it in 2025 to find more than 80 MW for data centers in Hillsboro in 2026, with batteries and on-site generation covering the tight hours. It expects more than 400 MW by 2029. MISO has validated automated software for its first-phase queue studies. *Early. Depends on: an access administrator that owns the data and the answer. Principle 2.*

**Distribution and resource management systems.** The control-room software that runs a local network and dispatches the devices connected to it, known as ADMS and DERMS. In a 2022 survey of sixty distribution utilities, 21 percent had a distribution management system fully working and another 49 percent were installing one. Twelve percent had a working system for distributed resources. PG&E and ComEd run flexible connections through theirs. *In service. Depends on: rules for who operates the platform and who can use its data. Principle 2.*

**Flexible interconnection.** Connect now, with a limit that tightens in the few hours the local network is full. Britain has done it since 2009, and its regulator standardized the terms in a 2022 decision. California's scheduled export limits took effect in July 2025. ComEd has approved more than 50 MW of community solar on flexible terms since November 2025, and Colorado has ordered a standard tariff for generation and load. *In service. Depends on: a standard tariff with published limits and a stated curtailment expectation. Principle 4.*

**Dynamic operating envelopes.** Limits recalculated for each device through the day, so that small systems share a circuit's capacity as conditions change. South Australia has required the capability in essentially all new exporting solar systems since July 2023, and more than 88 percent of eligible customers choose a flexible limit over a low fixed one. The nearest US example is a PG&E pilot with five customers. *Early in the United States. Depends on: the flexible-connection tariff, and a device standard that requires the capability. Principle 4.*

**Smart inverter controls.** Inverters that hold voltage steady, so a circuit can host more solar and storage. California's utilities and Hawaiian Electric require inverters certified to the 2018 national standard, and eight states had adopted it statewide by mid-2024. *In service. Depends on: adoption of the standard in each state's interconnection rules. Principle 4.*

### Flex: when and where power is made and used

**Virtual power plants.** Home batteries, thermostats, chargers, and other devices run together as one resource. The Department of Energy counts about 30 GW nationally, a figure that includes conventional demand response. Green Mountain Power's 110 MW virtual power plant, more than half of it home batteries, delivered 90 MW in Vermont's July 2026 heat wave. Puerto Rico's battery program was called 49 times in summer 2025, in many cases to avoid or shrink rolling blackouts. *In service. Depends on: payment for measured performance, by place and time. Principle 3.*

**Managed EV charging.** Shift vehicle charging out of the hours when a neighborhood circuit is full. A national laboratory survey reviewed more than 100 utility programs in 2024. In Baltimore, managed charging cut the peak of one group of 880 vehicles by 72 percent against unmanaged charging, and the program is approved to grow to 30,000 vehicles by 2027. *In service. Depends on: payment by place and time, starting with time-varying default rates and a hedged flat option. Principle 3.*

**Flexible commercial and industrial load.** Large customers that cut or shift their use when the system is tight. Demand response enrolled in the seven organized markets totaled 33 GW in 2024, 6.5 percent of peak. For data centers it is new: Google signed agreements in 2025 to curtail computing load, the first to target machine-learning work \[13\], and SPP's curtailable service for large loads took effect in July 2026. *Standard as demand response, early for data centers. Depends on: a tariff that trades faster connection for verified curtailment. Principles 4, 5.*

**Batteries placed at the constraint.** Charge when the network has room and discharge at the bottleneck, in place of a larger wire. National Grid's 6 MW battery on Nantucket, paired with a combustion turbine, was built to defer a third undersea cable. An 850 MW battery in New South Wales, in full service since September 2026, lets existing transmission lines carry more. Arizona Public Service used a 2 MW battery at Punkin Center in place of rebuilding about 20 miles of line, and disconnected it in 2019 after a fire elsewhere on its system \[20\]. *In service. Depends on: a payment formula under which a battery contract competes with a substation, and posted deferral values. Principles 1, 3.*

### Build: only where it is actually needed

**Advanced conductors and reconductoring.** Replace the wire on existing towers with one that carries more: up to double the capacity at about half the cost of a new line, where the corridor allows \[11\]. AEP reconductored 240 miles in the Lower Rio Grande Valley with the lines energized \[21\]. In September 2026 the Department of Energy selected projects in 26 states to reconductor or rebuild more than 1,500 miles. *In service. Depends on: a reconductor-first showing before new corridors, and the payment formula. Principles 1, 4.*

**Voltage upgrades.** Rebuild a line in its existing corridor to run at a higher voltage. New York's Smart Path Connect rebuilt about 100 miles from 230 kV to 345 kV, almost entirely within existing rights-of-way, and opened about 1,000 MW of capacity in 2026. Construction took three and a half years, so a rebuild follows the faster tools. *Standard. Depends on: the same showing. Principle 4.*

### What the inventory shows

**Seventeen tools depend on six rules.** Each tool is listed under the rule it depends on most. Several need a second.

| The rule | Tools that depend on it | Principle |
| --- | --- | --- |
| A total-expenditure allowance with shared savings | Line ratings, transformer ratings, power-flow control, batteries at the constraint | 1 |
| Published grid data, and a neutral operator that answers requests from it | Sensing, hosting-capacity analysis, capacity analysis, management systems, topology optimization | 2 |
| A standard flexible-connection tariff, with published limits and device standards | Flexible interconnection, operating envelopes, smart inverters | 4 |
| Payment for measured performance, by place and time | Virtual power plants, managed charging | 3 |
| A large-load tariff that trades connection speed for verified curtailment | Flexible commercial and industrial load | 4, 5 |
| A showing that cheaper options were weighed before new corridors | Reconductoring, voltage upgrades | 4 |

The roadmap's 2027 list starts five of the six: the docket on total expenditure, published data, the flexible-connection tariff, the large-load tariff, and the reconductor-first showing. Payment by place and time begins with default dynamic rates in 2028 and 2029. That is why the program is organized around the rules and not the hardware.

**Availability is rarely the constraint.** Most of these tools have been for sale for years. The Department of Energy reported that of the nearly $90 billion investor-owned utilities put into transmission and distribution in 2023, about $6 billion, 7 percent, went to what it calls advanced grid solutions, a list that overlaps this one. It named the payment model among the reasons: traditional cost-of-service business models "have not sufficiently incentivized these solutions" to justify the effort of deploying them at scale. Massachusetts regulators reported in 2025 that five such transmission tools "are now commercially available but remain underutilized," and that further deployment also depends on technical progress and on federal and regional rules.

The rules have moved slowly too. FERC required the simplest step, line ratings adjusted for forecast air temperature, in 2021. The first regional market switched them on in March 2026, and two have until the end of 2028. A federal notice on the fuller, sensor-based ratings has sat since 2024, with its next step listed as undetermined.

**The verbs are an order of work.** A utility's first ratings, switching, or power-flow project typically takes one to three years, and each one after that takes months. Reconductoring takes one to three years, and a voltage rebuild longer. A commission that starts with see and use gains capacity while the larger projects are still in design, and learns where the building needs to go.

### What the tools do not do

The inventory is also a list of limits. Dynamic ratings add capacity when the weather allows and withdraw it when it does not, so they relieve congestion more reliably than they firm up a connection. Operating envelopes need a control path to every device, which is why South Australia wrote the capability into its inverter requirements. Reconductoring depends on the conductor, the towers, and the corridor, and not every line qualifies.

Flexibility delivers only what is measured. Puerto Rico's batteries answered 49 calls in one summer, at an average of 33 MW from about 500 MW enrolled. In California's September 2022 emergency, utility demand response delivered most of what was scheduled, and third-party providers delivered about 195 MW of the 350 MW dispatched.

Flexibility also defers building without ending it. Britain's regulator found that about 75 percent of the capacity its distribution companies added in the first year of the current price control came from flexibility, and traced the emphasis on deferral to the total-expenditure incentive. It also found the companies falling behind on planned upgrades, and decided that the next price control, from 2028, will invest ahead of demand it expects to rise. None of these limits is an argument against a tool. Each is an argument for the published limits, the verification, and the output measures that connect by rule and pay for performance already require.

## What America should copy, and from whom

**From Britain:** the regulatory machinery. Totex allowances, output incentives, and DSO separation, plus the caution: RIIO's early price controls over-rewarded outperformance, so calibrate sharing factors tightly from the start \[1\]. A second caution: under the incentive most new distribution capacity came from deferral, and the regulator is now steering investment ahead of demand, so pair the incentive with an output measure for capacity delivered.

**From Australia:** the grid-edge engineering. Dynamic operating envelopes and aggregator interfaces at rooftop scale, plus the caution: envelopes without market signals waste the flexibility they create; pair them with the locational offers of Principle 3 \[7\].

**From Texas:** the speed. Connect-and-manage and an entry-friendly bulk market, plus the caution: an energy-only construct passes scarcity through to retail, so pair speed with the hedging obligations of Principle 5 instead of importing the volatility \[9\].

**From Hawaii:** the politics. Performance-based regulation adopted in the state where distributed penetration made the old model untenable first, proof that a commission can rewrite the compact when the legislature stakes out the destination \[2\].

## Transition fairness

Nothing here is retroactive: resources that cleared past auctions get paid through their delivery years, and obligations phase in over three cycles. Stranded value that survives independent review has a tested instrument, the ratepayer-backed securitization states used for restructuring-era stranded costs and use today for retired plants \[19\]. Public power, roughly a quarter of US load, adopts through different doors: board action at TVA (whose 2025 Google agreement shows the appetite \[13\]), financing conditions at the Rural Utilities Service, and joint-action agencies supplying DSO functions small systems cannot staff. And the open engineering questions Paper II names, inverter-dominated operation above all, argue for the sequenced deployment; every operating year in South Australia and ERCOT shrinks them.

### The advocate's capacity

The swing constituency in most of these dockets is the consumer advocate, and the program should arm them. Four provisions do it: funded intervention so participation does not depend on donated expertise; a standing right of access to the utility data the case turns on, under the same rules the operator follows; standardized cost-benefit and incidence templates so every mechanism is scored the same way in every docket, including this program's own proposals; and the authority to commission independent modeling against the utility's counterfactual rather than merely cross-examining it. A different regulatory bargain requires giving the regulator and the public the analytical capacity to negotiate it.

### The utility transition

A utility executive's first questions are operational, and they have answers. **Data:** grid data becomes a regulated asset with access rules, on the model of open banking; the utility keeps custody and cost recovery for the systems, and loses the right to withhold. **IT systems:** the ADMS, GIS, and metering platforms utilities are already deploying are the DSO's toolset; separation transfers administrative control and API obligations, not the software estate, and Britain executed this as a licence change rather than a systems rebuild \[4\]. **People:** the interconnection engineers, planners, and data teams move with their functions, as transmission staff moved into ISOs in the 1990s; the DSO is a destination for the workforce, not a layoff. **Planning processes:** the integrated distribution plan survives, produced by the DSO with the utility as builder-of-record for wires solutions, which changes who holds the pen and nothing about whether planning happens. **Capital plans:** in-flight projects grandfather into the totex baseline at approved values, and the reconductor-first showing applies prospectively. **Earnings expectations:** the multi-year revenue cap is set to be earnable at current-plan performance, with upside from outperformance, which is precisely the trade British networks accepted and then profited from; the transition risk investors should actually price is calibration of the sharing factors, and the RIIO record gives them a decade of data to price it with \[1\]. None of this is free, and the administrative cost is real. It is also one-time, small against a single deferred substation, and largely spending the utilities would incur anyway.

## Sources for the toolbox

Across the inventory: US Department of Energy, *Pathways to Commercial Liftoff: Innovative Grid Deployment* (April 2024), including its 2022 survey figures on management systems and its deployment timelines; Massachusetts Department of Public Utilities, *Advanced Transmission Solutions Report* (September 2025); FERC Order No. 1920 (2024).

See: PPL Electric Utilities, dynamic line rating results (October 2024), and FERC's advance notice on dynamic line ratings, Docket RM24-6 (June 2024); National Grid and Great River Energy deployment reports (2024, 2025); FERC Order No. 881 (2021), with its compliance orders for MISO and NYISO, and PJM's implementation notice (March 2026); FERC regulatory agenda entry for Docket RM24-6 (August 2026); Electricity North West, Celsius closedown report (2020); US Energy Information Administration, *Electric Power Annual*, Table 10.5 (2024 data); FERC staff, *2025 Assessment of Demand Response and Advanced Metering*; Department of Energy Office of Electricity on synchrophasors (2023); Department of Energy, *US Atlas of Electric Distribution System Hosting Capacity Maps* (2024); Colorado PUC Decision C25-0903 (December 2025).

Use: MISO Reliability Subcommittee minutes (February 2026) and quarterly update (August 2025); FERC order accepting SPP's reconfiguration tariff, ER26-2592 (August 2026); National Grid Electricity Transmission (May 2024), Georgia Power (September 2025), and Central Hudson on power-flow control; Portland General Electric (October 2025); MISO benchmarking of interconnection study software (April 2025); reporting on PG&E's and ComEd's flexible connections (2026); Scottish and Southern Electricity Networks on Orkney, and Ofgem's access reform decision (May 2022); California PUC Resolution E-5296 (2024); South Australia Office of the Technical Regulator (2023) and SA Power Networks (September 2025); California Rule 21 and Hawaiian Electric Rule 14H; Interstate Renewable Energy Council tracking of IEEE 1547-2018 adoption (2024).

Flex: Department of Energy, *Pathways to Commercial Liftoff: Virtual Power Plants* (2023, updated January 2025); Green Mountain Power (July 2026); Institute for Energy Economics and Financial Analysis on Puerto Rico's battery program (December 2025); California ISO Department of Market Monitoring, demand response report (February 2023); National Renewable Energy Laboratory survey of managed charging (2024) and Baltimore Gas and Electric program results (2025); FERC order on SPP's curtailable service, ER26-1323 (June 2026); Pacific Northwest National Laboratory on the Nantucket battery (2019); EnergyCo NSW on the Waratah battery (September 2026).

Build: GridLab and the University of California, Berkeley, *2035 Report: Reconductoring* (2024); Department of Energy, SPARK project selections (September 2026); New York Power Authority, Smart Path Connect (2026).

Britain: Ofgem, *ED3 Framework Consultation* (November 2024) and *ED3 Framework Decision* (April 2025).