TJE Ventures

The Physics: Why the Grid Is Unlike Any Other Infrastructure

Electricity cannot be stored in meaningful quantities at grid scale — not yet, and not cheaply. This single fact determines almost everything about how the grid is designed, operated, and priced. It is what makes the electric grid fundamentally different from every other delivery infrastructure humans have built.

When you open a water tap, you draw from a reservoir that has been filling for days or weeks. When you open a gas valve, you draw from a pipeline network backed by underground storage fields that hold months of supply. When you turn on a light switch, electrons must be generated at that exact moment, somewhere on the interconnected grid, to match your demand. The generation happens before you finish flipping the switch. If more electricity is consumed than is being generated at any instant, the grid’s frequency begins to fall. If that imbalance is not corrected within seconds, the cascade failure that follows can knock out power to millions of people in minutes.

Alternating Current and Frequency

The North American grid runs on alternating current (AC) at a nominal frequency of 60 Hz — the voltage alternates direction 60 times per second. Europe and most of Asia use 50 Hz. The frequency standard was established in the early twentieth century and is embedded in the design of every motor, transformer, and piece of grid-connected equipment on the continent.

Maintaining this frequency at exactly 60 Hz is the grid’s primary moment-to-moment operational challenge. Every synchronized generator on the grid rotates at a speed precisely tied to this frequency — a two-pole generator runs at exactly 3,600 RPM; a four-pole generator at 1,800 RPM. When load increases faster than generation can respond, the excess demand draws kinetic energy from the spinning mass of all the generators on the grid, slowing their rotation slightly. Frequency drops. Grid operators call this “inertia” — the physical flywheel effect of millions of tons of spinning machinery that buys seconds of time for corrective action.

The acceptable frequency range in North America is 59.95–60.05 Hz during normal operations. Below 59.5 Hz, automatic protective relays begin disconnecting generators and loads to prevent equipment damage. Below 57 Hz, the cascade can become uncontrollable. The 2003 Northeast blackout — which left 55 million people without power across eight U.S. states and Ontario — began with a software failure that prevented operators from seeing overloaded transmission lines and progressed to full cascade in 90 seconds once it started.

Why High Voltage for Transmission

Moving electricity over long distances at the voltages used in homes and businesses (120–240 V) would be wildly inefficient. Power loss in a transmission line is proportional to the square of the current flowing through it. The solution, developed by Nikola Tesla and George Westinghouse in the 1880s in direct competition with Thomas Edison’s direct current system, was to step voltage up dramatically for transmission and step it back down for consumption — using transformers that work only with alternating current. This is why AC won what history calls the “War of Currents.”

Transmission line voltages in North America:

• Extra high voltage (EHV) transmission: 345 kV, 500 kV, 765 kV — interstate backbone lines carrying bulk power hundreds of miles
• High voltage transmission: 115 kV to 230 kV — regional transmission connecting major substations
• Sub-transmission: 26 kV to 69 kV — feeding distribution substations within utility service territories
• Primary distribution: 4 kV to 35 kV — the lines on neighborhood utility poles
• Secondary distribution: 120/240 V — the service entrance to your home

A transformer at a transmission substation steps 500 kV down to 115 kV for regional distribution. Another at a neighborhood substation steps it down to 12 kV for the local distribution network. The transformer on the pole outside your house steps it down to 120/240 V for residential use. Each transformation introduces a small efficiency loss, but the combined system moves electricity across a continent with approximately 93–95% efficiency from generator to meter.

The Three Layers: Generation, Transmission, Distribution

Generation

The generation layer encompasses every facility that converts some energy source into electricity and connects to the grid. The diversity of generation types reflects both historical accident and deliberate engineering, and each type plays a different role in the balancing act the grid must maintain at all times.

Baseload generation runs continuously at or near full capacity because it is cheapest to run steadily and expensive or slow to start and stop. Nuclear power plants, large coal plants, and run-of-river hydroelectric dams are archetypal baseload resources. A nuclear plant has very low fuel cost per kWh but enormous capital cost — it must run as many hours per year as possible to recover that investment. The grid depends on baseload plants to provide the stable foundation of supply that variable demand sits on top of.

Peaking generation runs only during periods of high demand and sits idle otherwise. Natural gas combustion turbines — essentially jet engines connected to generators — can start from cold to full output in 10–15 minutes, making them ideal for responding to sudden demand spikes. They are expensive to run (high fuel cost per kWh) but cheap to build relative to their capacity, and their ability to start quickly has made them the grid’s primary tool for managing demand peaks for the past 50 years.

Intermediate or cycling generation runs during mid-range demand periods but not all the time. Combined-cycle natural gas plants — which pair a combustion turbine with a steam turbine that captures the exhaust heat — are more efficient than simple combustion turbines but take 30–60 minutes to start. They occupy the space between always-on baseload and quick-start peakers.

Variable renewable generation — wind and solar — produces electricity when the fuel (wind, sunlight) is available rather than when the grid needs it. This is the defining challenge of the energy transition: renewable resources are increasingly the cheapest source of new generation capacity, but their output is controlled by weather, not by grid operators. The grid must be restructured to accommodate generation that cannot be dispatched on command.

Transmission

The transmission network is the highway system of the grid — the high-voltage backbone that carries bulk electricity from generators to the populated areas that consume it. The U.S. transmission network comprises approximately 600,000 miles of high-voltage lines connecting thousands of generating stations to thousands of substations across the country.

Transmission lines do not just carry power from A to B like a pipeline carries gas. Electricity takes the path of least resistance across the network simultaneously, flowing through every available path in inverse proportion to that path’s impedance. When a line becomes overloaded or trips offline, the power it was carrying instantly redistributes across neighboring lines, which may or may not have the capacity to absorb it. This physical behavior — called loop flows — means that a problem anywhere on an interconnected network can propagate to lines hundreds of miles away, and it is why grid operators must monitor the entire network in real time rather than managing individual lines independently.

The transmission network was largely built between 1950 and 1980, designed for a generation mix dominated by large central power plants located near population centers. It was not designed for the current or projected future generation mix, where the best renewable resources (Texas wind, Southwestern solar, offshore Atlantic wind) are often far from the major load centers on the East and West Coasts. Building transmission to connect those resources to where the power is needed is one of the central infrastructure challenges of the energy transition — and it is moving far more slowly than renewable generation capacity itself.

Distribution

The distribution network is the “last mile” of the grid — the medium- and low-voltage system that takes power from transmission substations and delivers it to individual homes, businesses, and facilities. The U.S. distribution network contains approximately 5.5 million miles of lines — roughly nine times the length of the transmission network — operated by more than 3,000 utilities, cooperatives, and public power agencies.

Distribution systems are radial in design — power flows in one direction, from the substation out to customers, through a branching tree structure. This simplicity made them cheap to build and easy to protect with fuses and reclosers, but it creates single points of failure: if a line between the substation and a group of customers faults, all customers downstream are out until the fault is repaired. This is why most power outages that customers experience are distribution outages caused by equipment failures, fallen trees, and vehicle accidents — not transmission or generation failures.

The distribution network is the part of the grid least prepared for what is coming. Rooftop solar, home batteries, and electric vehicle chargers are being connected to a system designed in the 1950s for one-way power flow from utility to customer. Bidirectional power flow — homes and businesses both consuming and exporting electricity — requires upgraded equipment, smarter controls, and new operating practices that most distribution utilities are still working through.

The North American Grid Structure

North America does not have a single grid. It has four major synchronous interconnections — zones within which all generators operate in synchronized AC frequency — that are connected to each other only through high-voltage direct current (HVDC) ties that allow power to flow between them without requiring synchronization.

• Eastern Interconnection: The largest, covering all of the U.S. east of the Rocky Mountains plus most of eastern Canada and a small portion of Texas. Approximately 700 GW of generating capacity serving roughly 250 million people. This is one synchronized machine: a generator in Maine and a generator in Florida are spinning in exact synchrony at 60 Hz.
• Western Interconnection: The western U.S., western Canada, and the Baja California peninsula of Mexico. Approximately 250 GW of capacity.
• ERCOT (Electric Reliability Council of Texas): Most of Texas operates its own synchronous island, connected to the Eastern and Western Interconnections only through limited DC ties. This isolation is not an accident — it is deliberate, and the reason is regulatory. A grid that crosses state lines falls under federal jurisdiction (FERC). Texas kept its grid within state boundaries specifically to avoid federal oversight. This decision proved consequential during Winter Storm Uri in February 2021, when ERCOT could not import meaningful emergency power from neighboring interconnections because the DC ties connecting them were at maximum capacity.
• Quebec Interconnection: Hydro-Québec operates its own synchronous island, connected to the Eastern Interconnection via DC ties. Quebec’s massive hydroelectric resources allow it to be a large net exporter of power to New England and New York.

Alaska and Hawaii each operate independent grids with no physical connection to the continental system. Alaska has several regional interconnections that are not linked to each other; Hawaii’s islands each operate independently.

Who Runs It: Grid Operators, Utilities, and Markets

The Federal Framework

The Federal Energy Regulatory Commission (FERC) is the primary federal regulator of the wholesale electricity market — the transactions between generators, transmission owners, and utilities that happen above the retail level. FERC sets the rules for interstate transmission access, approves wholesale electricity rates, and oversees the reliability standards developed by the North American Electric Reliability Corporation (NERC). FERC does not regulate retail electricity — what consumers pay for their power is set by state public utility commissions.

NERC is the reliability standards body for North America, operating under FERC oversight in the U.S. NERC develops and enforces mandatory reliability standards (the CIP standards for cybersecurity, and the FAC, TPL, and other standards for physical planning and operations) and assesses the adequacy of the grid through its annual Long-Term Reliability Assessment. NERC’s 2023 assessment flagged more than half of North America as at elevated risk of energy shortfalls during extreme weather — a warning that reflects both the retirement of dispatchable generation and the increasing frequency of weather events that stress supply and demand simultaneously.

Regional Transmission Organizations and Independent System Operators

In deregulated electricity markets, the real-time operation of the bulk power system is managed by Regional Transmission Organizations (RTOs) or Independent System Operators (ISOs) — independent entities that operate the transmission network and run competitive wholesale electricity markets on behalf of generators, utilities, and large customers.

• PJM Interconnection: The largest RTO in the world, covering 13 states from Illinois to New Jersey and the District of Columbia. Manages approximately 180 GW of capacity serving 65 million people.
• MISO (Midcontinent ISO): 15 states in the Midwest and South, ~180 GW.
• CAISO (California ISO): Most of California, ~80 GW.
• SPP (Southwest Power Pool): 14 states in the central U.S., heavily wind-penetrated, ~90 GW.
• NYISO: New York State, ~40 GW.
• ISO-NE: The six New England states, ~30 GW.
• ERCOT: Most of Texas, ~90 GW — the only major U.S. grid operator not under FERC jurisdiction.

Large portions of the U.S. — most of the Southeast, the Pacific Northwest, and parts of the Southwest — are still served by vertically integrated utilities that own generation, transmission, and distribution and are regulated by state commissions rather than competitive markets. In these territories, a single utility (Southern Company, Duke Energy, Pacific Power) makes all the investment decisions about the generation and transmission mix, subject to state regulatory approval of its rates.

How Electricity Is Priced in Competitive Markets

In RTO/ISO markets, generators submit offers stating how much power they are willing to produce at what price. The grid operator stacks these offers from cheapest to most expensive and accepts them in order until projected demand is met — a process called the merit order dispatch. The price paid to all accepted generators is set by the most expensive unit needed to meet demand (the marginal unit). This is the Locational Marginal Price (LMP) — the cost of serving one additional megawatt-hour of load at a specific location at a specific time.

LMP has three components: the system energy price (the marginal cost of generation across the whole market), a congestion component (the cost of moving power across constrained transmission paths), and a loss component (the cost of line losses at that location). A location with plentiful local generation and uncongested transmission has a low LMP. A location where demand is high, local generation is limited, and transmission lines are congested can have an LMP that is 5–10 times the system average — or, when generation is curtailed overnight, can go negative as generators pay to keep running rather than shut down.

Beyond the energy market, most RTOs operate capacity markets that pay generators for the ability to be available during system peaks, regardless of whether they actually run. These capacity payments are what keep peaking plants economically viable during the years when mild weather means they are rarely called. As the generation mix shifts toward zero-marginal-cost renewables that depress energy market prices, capacity markets become increasingly important for keeping backup generation available — and increasingly controversial as debates continue about whether they adequately compensate the right resources.

The Balancing Act: Reliability in Real Time

Every RTO and utility operates a control room staffed 24 hours a day, 365 days a year, where operators monitor the state of the grid and make continuous adjustments to keep supply and demand in balance. The tools available to them, from fastest to slowest:

• Frequency regulation (seconds): Generators automatically adjust their output in response to frequency deviations — speeding up when frequency drops, slowing when it rises. This automatic response, called governor response, is the grid’s first line of defense and operates faster than any human can act. Grid-scale batteries can respond even faster, injecting or absorbing power in milliseconds.
• Spinning reserves (seconds to minutes): Generators running below full capacity that can ramp up quickly to cover a sudden loss of generation. NERC standards require each grid region to carry spinning reserve equal to the largest single contingency in its footprint — typically the output of the largest generator or transmission line.
• Non-spinning reserves (minutes to 10 minutes): Generation that can start and reach full output within 10 minutes — combustion turbines in hot standby, hydro units with water already in the gate.
• Replacement reserves (10 minutes to an hour): Resources that can be brought online within an hour to replace reserves that have been deployed.
• Economic dispatch (ongoing): The grid operator continuously re-solves the dispatch optimization, adjusting which generators are running and at what output level to meet the current and projected load at minimum cost while respecting all transmission constraints.

The planning standard that underlies all of this is called N-1 contingency: the grid must be able to survive the loss of any single element — any one generator, transmission line, or transformer — without cascading failures. The planning horizon extends to N-1-1 for the most critical facilities: survivability of the loss of any one element while another is already out of service. Meeting this standard requires substantial redundancy — transmission lines and generators that sit underutilized most of the time, available precisely for the emergencies when they are needed.

The Aging Infrastructure Problem

The bulk of the U.S. transmission and distribution network was built between 1950 and 1980 and designed for a 40–50 year service life. Much of it is now operating beyond that design horizon, and the consequences are increasingly visible in reliability statistics and in the cost and complexity of maintaining aging equipment.

• Transmission lines: The average age of U.S. high-voltage transmission infrastructure is approximately 40 years. Transmission lines themselves can operate for 80+ years with maintenance, but the towers, insulators, conductors, and protection systems require periodic replacement. The greater challenge is not the physical infrastructure but the control and protection systems, many of which use digital equipment from the 1980s and 1990s that is no longer supported by manufacturers and cannot be patched against cybersecurity vulnerabilities.
• Power transformers: Large high-voltage transformers are the most critical and most vulnerable component of the transmission system. A 500 kV autotransformer at a major substation may weigh 400–500 tons, cost $3–7 million, take 12–18 months to manufacture, and serve as the sole path for bulk power transfer in its region. The average age of large power transformers in the U.S. is approximately 40 years; about 70% are beyond their design life. There is no domestic spare inventory adequate for a widespread failure event, and lead times from manufacturers (most of which are overseas) make rapid replacement impossible.
• Distribution infrastructure: The 5.5 million miles of distribution lines include wood poles with service lives of 30–50 years. The American Society of Civil Engineers estimates that roughly 70% of U.S. distribution poles are past their design life. Aging distribution infrastructure is the proximate cause of most customer outages and, in dry climates, of electrical ignition of wildfires. Pacific Gas & Electric’s equipment has been implicated in multiple catastrophic California wildfires, including the 2018 Camp Fire that killed 85 people and destroyed the town of Paradise.

The American Society of Civil Engineers gives U.S. energy infrastructure a grade of D+ in its Infrastructure Report Card. The estimated cost to modernize and expand the grid to meet both current reliability needs and the demands of the clean energy transition ranges from $1 trillion to $3 trillion over the next 20 years — a figure that dwarfs any previous infrastructure investment program in U.S. history.

The Renewable Integration Challenge

Adding large amounts of variable generation to a grid designed for dispatchable generation introduces challenges at every level of the system. The challenges are solvable — they are being solved in practice in high-renewable grids around the world — but they are not free, and they are not solved simply by installing more wind and solar panels.

The Duck Curve

California’s grid operator CAISO coined the term “duck curve” in 2013 to describe a phenomenon that has since become one of the defining challenges of the energy transition. As midday solar generation grows, the net load that conventional generators must serve (total load minus solar output) develops a characteristic shape on a daily plot: low and relatively flat in the morning, plunging sharply as solar ramps up through midday, reaching a deep trough in the early afternoon, then rocketing upward as solar fades in the late afternoon just as evening demand peaks. The shape resembles a duck — the trough is the belly, the evening ramp is the neck.

The duck curve creates two distinct operational problems:

• Overgeneration and curtailment. At the bottom of the trough, solar output can exceed total demand. The grid operator must either curtail solar generation (turn it off despite zero fuel cost) or find somewhere to send the surplus. In spring, when air conditioning demand is low and solar output is high, California regularly curtails thousands of megawatt-hours per day. In 2022, California curtailed approximately 2.4 TWh of solar energy — power that was generated but not used.
• The evening ramp. The transition from the midday trough to the evening peak requires the rapid deployment of thousands of megawatts of dispatchable generation in a 3–4 hour window. In 2023, CAISO’s maximum observed evening ramp rate exceeded 17,000 MW in three hours — the equivalent of bringing 17 large power plants online in an afternoon. This ramp requirement is growing every year as more solar is installed, and the grid’s ability to meet it depends on having sufficient fast-ramping generation or storage available.

The Transmission Gap

The best renewable resources are rarely near the people who need the power. The highest-quality onshore wind in the U.S. is in the Great Plains — Wyoming, Montana, the Dakotas, Kansas, Oklahoma. The best solar is in the Desert Southwest and Texas. The major load centers are on the coasts. Moving power from where renewables can be generated cheaply to where it is needed requires transmission that does not yet exist and is being built very slowly.

The interconnection queue — the line of proposed generators waiting for studies and approvals to connect to the grid — contained over 2,600 GW of proposed capacity as of early 2025, approximately 95% of it wind, solar, and storage. The average project has been waiting in the queue for more than four years. Most projects never complete the process: the withdrawal rate from interconnection queues has been roughly 70–80% in recent years, as projects encounter interconnection costs, transmission constraints, or permitting obstacles that make them uneconomical.

FERC Order 1920 (2024) represents the most significant transmission planning reform in decades, requiring transmission owners to plan proactively for the generation mix expected over a 20-year horizon rather than simply reacting to individual generator interconnection requests. Whether this reform translates into actual transmission construction depends on how regulators, utilities, and states implement it — a process that will take years.

Inertia and Grid Stability

As synchronous generators (coal, gas, nuclear, hydro) are replaced by wind and solar, the grid loses a physical property that has always been its primary defense against frequency disturbances: rotational inertia. The massive spinning rotors of synchronous generators store kinetic energy that resists sudden frequency changes, buying the seconds that operators need to deploy reserves. Inverter-based resources — solar panels, wind turbines, and batteries that connect to the grid through power electronics — do not inherently provide this inertia.

Modern inverter technology can be programmed to provide “synthetic inertia” and fast frequency response, and grid-scale batteries can respond to frequency deviations far faster than any conventional generator. But this requires deliberate configuration and compensation, and the industry’s standards and market rules for requiring and paying for these services are still being developed. The South Australian blackout of 2016 — triggered by a transmission fault that cascaded partly due to inadequate inertia as the grid was running on high renewable penetration — provided an early and expensive lesson in what happens when inertia disappears faster than the grid’s control systems can adapt.

The New Load Problem: EVs, Data Centers, and AI

For most of the period from 2007 to 2021, U.S. electricity demand was essentially flat. Efficiency improvements in appliances, lighting (LED replacement of incandescent bulbs), and industrial processes offset the growth in the number of connected devices and customers. Grid planners built their investment cases on flat or declining load forecasts.

That assumption has been shattered. U.S. electricity demand is now growing at rates not seen since the mid-twentieth century, driven by three converging forces:

• Data centers and AI. As discussed in the data center article, AI infrastructure alone could add 10–30 TWh of new demand annually by 2030. The buildout of hyperscale AI clusters is creating concentrated new loads of 500 MW to 1,000 MW in single locations — the equivalent of adding a medium-sized city to the grid at one point of interconnection. Northern Virginia, which already hosts the densest data center concentration in the world, is facing transmission constraints that are delaying new data center connections by 5–10 years despite the economic pressure to connect faster.
• Electric vehicles. The U.S. fleet has approximately 3 million EVs as of 2024. Under projected adoption scenarios, that could grow to 40–50 million by 2035. A fleet of 40 million EVs charging primarily in the evening could add 100–150 TWh of annual demand and create local distribution grid congestion in neighborhoods with high EV adoption. The good news is that EV charging is highly flexible: most charging can shift to off-peak overnight hours or to midday when solar is abundant, turning what could be a grid problem into a grid asset if managed through smart charging programs.
• Building and industrial electrification. Heat pumps replacing gas furnaces, electric water heaters replacing gas units, and electric arc furnaces replacing blast furnaces in steel manufacturing represent large but diffuse additions to electric load. The combination of these efficiency improvements and fuel substitutions is expected to roughly double total U.S. electricity demand by 2050 relative to today.

Utilities that spent a decade planning for flat load growth are now scrambling to develop long-range plans for load doubling. Transmission and generation investment decisions made today take 5–15 years to come online. The mismatch between planning horizons and the speed of load growth is one of the grid’s most acute near-term challenges.

The Smart Grid: What Is Actually Deployed

“Smart grid” has been an industry buzzword for two decades, often used so broadly as to be nearly meaningless. Setting aside the hype, several specific technologies have been deployed at material scale and are genuinely changing grid operations:

Advanced Metering Infrastructure (AMI)

Roughly 115 million “smart meters” have been installed in the U.S. as of 2024 — approximately 80% of all residential and commercial meters. Unlike traditional meters that require a physical read, smart meters communicate electricity consumption data (typically in 15-minute intervals) back to the utility automatically. This enables time-of-use (TOU) rates that charge more during peak hours and less during off-peak hours, creating a price signal that encourages customers to shift flexible loads (dishwashers, water heaters, EV chargers) away from system peaks.

The penetration of smart meters has been high; the penetration of TOU rates has been lower. Many utilities with full AMI deployment still offer flat-rate pricing to most customers, leaving the demand flexibility potential of smart metering largely unrealized.

Grid-Scale Battery Storage

Battery Energy Storage Systems (BESS) have grown from a novelty to a major grid resource in the span of five years. The U.S. had approximately 20 GW of grid-scale battery storage installed as of mid-2024, with another 30–40 GW in the interconnection queue for the next three years. California alone has deployed over 7 GW of BESS, primarily to address the duck curve ramp challenge — storing midday solar and discharging it during the evening peak.

BESS provides three distinct grid services:

• Energy arbitrage: Charging during low-price periods (midday solar surplus) and discharging during high-price periods (evening peak), earning the price spread
• Frequency regulation: Responding to frequency deviations within milliseconds — faster than any conventional generator and therefore highly valuable for the ancillary services market
• Capacity: Providing reliable peak capacity for resource adequacy requirements, subject to duration limits (most current BESS is 2–4 hours of storage)

The current limitation of grid-scale batteries is duration. A 4-hour battery fully charged can meet 4 hours of discharge at rated power, then it is empty. For the daily duck curve, 4 hours is often sufficient. For multi-day calm weather events when solar and wind output are both low — the “dark doldrums” that challenge high-renewable grids — 4-hour batteries provide no help. Longer-duration storage technologies (pumped hydro, iron-air batteries, hydrogen) are being developed but are not yet commercially deployed at the scale needed.

Demand Response

Demand response programs pay electricity customers to reduce consumption during grid stress events — the demand-side equivalent of calling up a peaking power plant. Large industrial customers have participated in demand response for decades. Residential demand response has scaled significantly with smart meters and connected devices:

• Smart thermostats (Nest, Ecobee) enrolled in utility programs can receive a signal to raise the cooling setpoint by 2–4°F during peak events, reducing load across millions of homes simultaneously without customers noticing
• Water heaters, pool pumps, and EV chargers can be cycled off for 15–30 minute intervals with minimal customer impact but significant aggregate load reduction
• Virtual Power Plants (VPPs) aggregate distributed resources — rooftop solar, home batteries, flexible loads — into a single dispatchable resource that a grid operator can call on like a conventional power plant

The Rocky Mountain Institute estimated in 2022 that the U.S. has 200 GW of technical demand response potential — more than the capacity of the entire U.S. nuclear fleet. The amount actually enrolled in utility programs is a fraction of that.

Grid Security: Physical and Cyber

Physical Attacks

On April 16, 2013, snipers opened fire on the Metcalf transmission substation in California, shooting out 17 high-voltage transformers with rifle fire over 52 minutes. The attack cut fiber optic communication lines first, then systematically targeted transformer cooling systems. The substation was disabled for 27 days. The attack caused no blackout only because grid operators rerouted power before critical thresholds were reached. The perpetrators were never identified.

The Metcalf attack exposed a vulnerability that has not been adequately addressed in the decade since: the U.S. transmission system has a small number of critical substations whose loss would cause cascading blackouts across large regions and whose physical protection is inadequate. FERC issued a physical security standard (CIP-014) following the attack, but the adequacy of its implementation has been questioned by NERC, the Department of Energy, and congressional oversight committees.

The December 2022 attacks on substations in North Carolina — where rifle fire on two substations left 45,000 customers without power for days in December cold — demonstrated that the vulnerability documented in 2013 persists. The FBI has investigated multiple additional substation attacks and plots in the years since.

Cyber Vulnerabilities

In December 2015 and December 2016, Russian military hackers caused the first confirmed cyberattack-induced power outages in history, attacking Ukrainian electricity distribution companies with malware that allowed remote attackers to open breakers and cut power to hundreds of thousands of customers. The 2016 attack used specialized industrial control system malware (Industroyer/Crashoverride) designed specifically to communicate with and manipulate grid control systems.

U.S. grid operators have NERC CIP (Critical Infrastructure Protection) cybersecurity standards that are more rigorous than Ukraine’s pre-attack posture, but the same classes of vulnerabilities exist. The Cybersecurity and Infrastructure Security Agency (CISA) has documented persistent access by state-sponsored threat actors in U.S. critical infrastructure networks, including energy facilities. These intrusions are typically characterized as pre-positioning — establishing access that could be activated during a conflict rather than immediately disruptive.

The cybersecurity challenge in the power sector is compounded by the industry’s legacy of operational technology (OT) systems — industrial control systems, SCADA systems, and protective relays from the 1980s and 1990s — that were designed when they were air-gapped from the internet and are now connected to corporate networks in ways their designers never anticipated. Patching these systems is difficult because they cannot be taken offline without disrupting grid operations, and many run software that is no longer supported.

Solar Storm Risk

In September 1859, a solar coronal mass ejection (CME) struck the Earth with a force that, if repeated today, could induce geomagnetically induced currents (GICs) in long transmission lines powerful enough to destroy large power transformers at scale. The 1859 Carrington Event is the most powerful solar storm in the historical record; a repeat at similar magnitude is estimated to have a 1-in-150 to 1-in-100 year probability — a low-probability, catastrophic-consequence risk.

A 1989 solar storm (far smaller than the Carrington Event) caused a nine-hour blackout across all of Quebec, damaging transformers across North America. The Hydro-Québec system was particularly vulnerable because its long high-voltage lines from remote hydro generation acted as antennas for the induced currents. The U.S. has similar configurations in its long transmission corridors.

NERC has developed geomagnetic disturbance (GMD) standards requiring utilities to assess their transformer fleets for GIC vulnerability and implement protective measures. Progress has been incremental. The fundamental challenge is that truly adequate protection against a Carrington-scale event would require replacement or protection of hundreds of large power transformers at a cost of billions of dollars — a bill that utilities are reluctant to pay for a risk that has not materialized in the modern grid era.

What Needs to Happen

The gap between the grid as it exists and the grid required to support the energy transition, growing electricity demand, and adequate reliability is large. The technology to close most of that gap exists; the primary barriers are permitting, financing, and the coordination of dozens of independent actors with conflicting interests.

Transmission Expansion

The U.S. needs to roughly double its high-voltage transmission capacity over the next 20 years to connect renewable generation to load centers, provide the redundancy needed for N-1 reliability under a more variable generation mix, and create the interstate power flows that allow surplus generation in one region to serve deficits in another. Current transmission construction rates are nowhere near this pace.

The barriers are primarily permitting and siting, not technology. A major interstate transmission line requires approvals from every state and county it crosses, often facing local opposition, environmental reviews, and landowner litigation that extend timelines to 10–20 years from conception to energization. The SunZia transmission project in New Mexico and Arizona — a 550-mile HVDC line to carry wind power to load centers — took 17 years from proposal to groundbreaking. The Grain Belt Express, crossing Missouri, took years of legislative battles before a single tower was erected.

Federal permitting reform and the development of designated national transmission corridors that streamline multi-state approvals are the most direct available interventions. Neither has been implemented at the scale the problem requires.

Long-Duration Storage

Four-hour lithium-ion batteries solve the duck curve. They do not solve the multi-day low-generation events that will become the critical reliability challenge as the grid approaches 80%+ renewable penetration. Technologies that can store energy for days to weeks at grid scale include:

• Pumped hydro: Mature, cost-effective, but geographically constrained and requiring decades to permit and build. The U.S. has approximately 22 GW of pumped hydro; the development pipeline is thin.
• Iron-air batteries (Form Energy): Multi-day storage at significantly lower cost than lithium-ion, but still pre-commercial at utility scale.
• Green hydrogen: Electrolysis converts surplus electricity to hydrogen for long-term storage, re-converted to electricity through fuel cells or turbines when needed. High round-trip efficiency losses (~40–60%) are a significant economic challenge.
• Compressed air energy storage (CAES): Stores energy as compressed air in underground caverns; limited geographic availability.

No single technology is ready to provide seasonal storage at the required scale and cost today. A portfolio approach — deploying multiple storage technologies at their points of greatest effectiveness while pushing down costs through deployment — is the most realistic path.

Grid Modernization and Interconnection Reform

FERC Order 1920 on transmission planning and Order 2023 on interconnection reform are the most significant regulatory changes to U.S. grid governance in a generation. Their implementation will determine whether the 2,600 GW of projects sitting in interconnection queues can actually get connected in a reasonable timeframe. Early indications suggest that processing speed has improved modestly; whether the backlog can be meaningfully reduced requires sustained regulatory attention and utility investment in study resources.

Upgrading aging protection and control systems — replacing 1980s digital relays with modern, cybersecurity-hardened equipment — is necessary for both reliability and security, and it is a continuous program of work rather than a one-time project. Reconductoring existing transmission lines with advanced high-temperature low-sag (ACSS/TW) conductors can increase the capacity of existing rights-of-way by 50–100% at a fraction of the cost of new construction, and it avoids the permitting challenges of new corridors. This is one of the most cost-effective near-term investments available and is receiving growing attention from grid planners.

Bottom Line

The electric grid is aging infrastructure facing simultaneous demands it was not designed to meet: a generation mix shifting rapidly toward variable renewables, a load profile growing after decades of stagnation, new threats — cyber and physical — that legacy infrastructure was not built to resist, and a clean energy transition that requires it to double in capacity while replacing much of what already exists.

None of this means the lights are about to go out. The grid’s resilience is real; it survives thousands of component failures every year through redundancy and real-time operator skill. NERC’s reliability standards, imperfect as their implementation sometimes is, represent a genuine and globally respected engineering framework. The U.S. grid’s average customer interruption duration has been essentially flat for two decades despite increasing weather severity — a quiet testament to the maintenance and operations workforce that keeps the machine running.

What is not tenable is the current rate of investment and reform relative to the demands being placed on the system. The mismatch between load growth and generation/transmission development timelines is the most acute near-term risk. The mismatch between renewable integration requirements and the pace of transmission permitting reform is the most acute medium-term risk. And the mismatch between the economic value at stake — the entire digital economy runs on grid reliability — and the transformer spare inventory and cyber protection standards currently in place is the most acute tail risk.

The technology to address all of these exists. What has been slower to develop is the regulatory coordination, financing structures, and political will to deploy it at the pace the problem requires.

The grid is not a passive piece of infrastructure. It is a real-time balancing act performed continuously by thousands of engineers, operators, and automated systems, across a machine spanning a continent. Understanding it is the starting point for understanding every other energy story in this series.