Energy markets move megawatt-hours from generators to load. But a power system needs more than bulk energy: it needs that energy delivered at exactly 60 hertz, at stable voltage, on schedule — and it needs a plan for recovering when something big goes wrong. Ancillary services are the markets that buy these capabilities. They are small — typically just a few percent of total wholesale costs — but the grid cannot operate without them, and their design is becoming one of the most dynamic areas in electricity markets.
Why energy alone isn’t enough
Three physical realities create the need. First, supply and demand must balance continuously, not just hourly — the grid has no meaningful buffer, so second-by-second mismatches must be corrected in real time. Second, the system must survive contingencies: the sudden loss of the largest generator or transmission line, an event grid planners call the “N-1” criterion, must not cascade into a blackout. Third, alternating-current networks need voltage support and other technical services that energy dispatch alone doesn’t provide.
Ancillary services are the market’s answer: explicit products, with explicit prices, for each of these reliability functions. The standard taxonomy, defined by FERC, includes regulation, spinning and non-spinning reserves, voltage support, and black start — with newer products like flexible ramping and inertia emerging as the resource mix changes.
Frequency regulation: balancing by the second
Grid frequency — 60 Hz in North America — is the pulse of the system. When generation exceeds load, frequency rises above 60; when load exceeds generation, it falls. Even small sustained deviations damage equipment and, in the extreme, trigger protective relays that disconnect load.
Regulation is the continuous correction. Resources providing regulation sit under automatic generation control (AGC): the operator sends a signal every few seconds instructing them to raise or lower output, and they respond automatically. A regulation resource might move up and down hundreds of times a day around its scheduled output, absorbing the random fluctuations of load and variable renewables.
The key insight of modern regulation markets is that speed and accuracy matter more than size. A battery that can reverse from charging to discharging in milliseconds and follow the signal precisely provides more regulation value per MW than a thermal plant that ramps slowly and overshoots. Markets like PJM’s pioneered performance-based regulation, paying resources based on how accurately they follow the dispatch signal — a design that batteries dominate, and a major reason storage developers target regulation markets as an early revenue stream.
Regulation is procured in two directions: regulation up (ability to increase output) and regulation down (ability to decrease output or absorb surplus). Prices are typically modest — single digits to tens of dollars per MW per hour — but they spike when the system is short of flexible resources, which is precisely when regulation is most valuable.
Reserves: insurance in layers
If regulation handles the continuous small imbalances, operating reserves handle the large sudden ones. They are capacity held back from normal energy dispatch, ready to deploy when a contingency hits. The tiers are defined by response time:
Spinning reserve is synchronized to the grid — turbines literally spinning — and able to ramp to full output within about 10 minutes. It is the first line of defense: when a large generator trips, spinning reserves automatically pick up the slack through governor response within seconds, arresting the frequency decline before load is affected.
Non-spinning reserve is offline but able to start and reach full output within about 10–30 minutes — typically fast-start combustion turbines. It backfills: once spinning reserves have stabilized the system, non-spinning units start up and take over, allowing spinning reserves to be restored.
Supplemental and replacement reserves operate on longer timescales (30–60+ minutes), restoring the reserve cushion so the system is protected against a second contingency.
Consider the textbook contingency, for illustration: a 1,200 MW nuclear unit trips offline without warning. In the first seconds, the grid’s inertia and governor response from spinning reserves automatically inject power, limiting the frequency dip. Within minutes, the operator dispatches additional spinning reserve to its full output. Within half an hour, fast-start units are online and the system is re-secured. Without pre-procured reserves, that sequence fails at step one — and the alternative is involuntary load shedding.
Reserve requirements are set by reliability standards — typically enough to cover the single largest contingency, plus a margin — and reserve markets procure them competitively. Reserve prices are usually low when the system is comfortable and rise steeply as reserves thin, because the value of insurance peaks exactly when you are most likely to need it. Modern operating reserve demand curves (ORDCs) formalize this: they add an administrative price adder to reserves (and energy) that increases as the reserve margin shrinks, ensuring prices reflect scarcity even before an actual shortfall.
Voltage support and reactive power
Real power (megawatts) does the work; reactive power (measured in MVARs) sustains the voltage that lets real power flow. Without adequate reactive support, voltages sag, transmission capacity drops, and in the extreme the system suffers voltage collapse — a cascading failure mode distinct from frequency problems.
Reactive power must be provided locally — it doesn’t travel well over distance — so voltage support is an inherently locational service provided by generators, capacitors, and static VAR compensators near load centers. Most markets compensate it through cost-based or regulated payments rather than competitive auctions, because the local market power concerns are acute: if only two generators can provide voltage support in an area, an auction between them isn’t meaningfully competitive. It is the least “market-like” of the ancillary services, and reformers periodically revisit whether competition can be introduced.
Black start: the doomsday product
Black start is the capability to restart the grid after a total blackout — generating units that can start without external power (typically hydro plants or diesel-backed combustion turbines) and energize transmission paths so other plants can restart in sequence. It is procured through long-term contracts with designated units, tested periodically, and — mercifully — almost never used. Like all insurance, its value is proven only in the disaster it prevents. System restoration plans, which black start units anchor, are among the most security-sensitive documents grid operators hold.
The emerging products: ramping and inertia
Two physical changes are creating demand for new ancillary products. First, variable renewables create net-load ramps — the “duck curve” phenomenon where solar floods the midday market and then vanishes as evening demand peaks, requiring the remaining fleet to ramp thousands of MW in a few hours. Flexible ramping products, pioneered by CAISO and MISO, explicitly procure ramp capability in the day-ahead and real-time markets, paying resources to hold headroom for expected ramps rather than discovering the shortage in real time.
Second, inertia — the stabilizing momentum of massive spinning turbines — has historically come free as a byproduct of thermal generation. Inverter-based resources (wind, solar, batteries) provide no inherent inertia. As synchronous generation retires, some markets are studying or implementing inertia as a procured product, with grid-forming inverters and synchronous condensers as potential providers. It is a vivid example of a physical property the old grid got for free becoming a priced service in the new one.
Co-optimization: buying energy and reliability together
A subtle but important design feature: most RTOs co-optimize energy and ancillary services in a single dispatch. The optimization decides simultaneously which units provide energy and which hold reserves, recognizing the tradeoff — a megawatt held as spinning reserve is a megawatt not selling energy. Co-optimization finds the genuinely least-cost combination, and it produces consistent prices: the reserve price reflects the opportunity cost of withholding energy. Markets that procure reserves separately from energy leave money on the table and produce distorted incentives.
Who provides what: the resource mix
Different resources naturally gravitate to different ancillary products, and the mapping reveals the economics of each technology:
Batteries dominate fast regulation and increasingly provide spinning reserve — their millisecond response and precise controllability are unmatched, and regulation’s performance-based pay rewards exactly those traits. A battery’s ancillary revenue often exceeds its energy arbitrage revenue in early years.
Hydro plants (where available) are superb reserve providers — they can go from zero to full output in minutes — and many provide black start capability inherently.
Gas combustion turbines are the traditional spinning and non-spinning reserve backbone: fast-starting, dispatchable, and already sited across the grid. Their reserve revenue is a meaningful supplement to energy and capacity income.
Demand response — aggregated commercial and industrial load willing to curtail — competes directly in reserve and regulation markets in most RTOs, and FERC policy requires that it be allowed to participate on comparable terms to generation. A cold-storage warehouse that can pause refrigeration for 30 minutes is, to the grid, functionally a fast-start generator.
Nuclear and coal provide inertia and voltage support as byproducts of simply running, but rarely compete in the fast markets — they’re too inflexible. Their retirement removes these byproduct services, which is part of why inertia and ramping are becoming explicit products.
This mapping is why the resource transition reshapes ancillary markets so profoundly: the fleet that’s best at bulk energy (inflexible baseload, variable renewables) is not the fleet that’s best at flexibility services, and markets must price the latter well enough to attract it.
How ancillary prices are actually set
Most ancillary services clear in co-optimized auctions alongside energy: the operator’s optimization simultaneously decides which units generate energy and which hold reserves, minimizing total cost. The reserve price emerges as the opportunity cost of withholding capacity from energy — for illustration, if a gas plant could earn $50/MWh generating but is held as spinning reserve, the reserve price must compensate roughly that foregone margin.
This produces a characteristic price pattern: ancillary prices are near zero when the system is long on flexible capacity (reserves are nearly free because no energy margin is sacrificed), and spike when the system is tight (withholding energy is expensive, and the insurance value is highest). Some products clear day-ahead with hourly prices; regulation and reserves also clear in real time. Settlement follows the same two-settlement logic as energy: day-ahead awards provide certainty, real-time deviations settle at real-time prices.
Why this matters for investors and analysts
Ancillary services are a small share of system cost but a meaningful share of revenue for the right assets. A battery project’s business case often stacks regulation, spinning reserve, and energy arbitrage — with regulation frequently the anchor revenue in early years. As renewable penetration rises, the demand for flexibility services grows faster than the demand for bulk energy, which means the value pool for fast, flexible resources is structurally expanding even as average energy prices face downward pressure from zero-marginal-cost renewables.
Watch for: new product introductions (ramping, inertia) that create fresh revenue streams; performance-pay reforms that reward speed; and ORDC enhancements that steepen scarcity pricing. The ancillary services nobody thought about a decade ago are becoming some of the most strategically important markets on the grid.