Multi-State Operator (MSO) Energy Procurement Strategy

Why the site-by-site approach fails

The usual MSO story: each facility was built by a different team, signed whatever the local broker offered, and sends its bill to the local controller. Headquarters sees a consolidated utilities line. Nobody can answer, in one sitting, which sites are on fixed contracts, when they expire, which ISO each sits in, or what share of the total is negotiable at all.

That is a problem because the negotiable share differs by state. In a full-choice state the supply portion, typically half or more of a large cultivation bill, is contractable; delivery is not, though it is still manageable through rate class and demand strategy. In a capped state like Michigan the contractable share may be zero this year because the program is full. In Nevada it is zero unless you are large enough to file an exit application. Treating all of these the same either leaves money on the table in the open markets or wastes effort in the closed ones. What is energy deregulation covers the state-by-state landscape; this page is about running it as one portfolio.

The ISO map: where each site actually buys power

The 17 jurisdictions where cannabis operators can, in some form, choose a supplier fall into six wholesale environments. The rules that matter for a procurement strategy differ in each.

Wholesale market	Jurisdictions among the 17	What is different for a buyer
PJM Interconnection	Illinois (ComEd), Ohio, Pennsylvania, New Jersey, Maryland, Delaware, Washington, D.C.; Michigan's Indiana Michigan Power territory	Serves all or parts of 13 states plus D.C. [pjm-about]. Capacity is bought three years ahead in the Reliability Pricing Model auction, with locational prices that vary by transmission constraint, and the cost flows to customers through their supplier [pjm-learning-capacity]. Each account carries a capacity tag set by its load at PJM's peak hours, so a site's summer behavior sets a year of capacity cost.
MISO	Illinois (Ameren), most of Michigan	Michigan's choice program is capped at 10 percent of each utility's sales and has been full, with a queue of customers waiting for room [mpsc-choice-2025-report]. Sites here plan on utility rates unless and until an allotment opens.
ISO New England	Massachusetts, Connecticut, Rhode Island, Maine, New Hampshire	The Forward Capacity Market holds annual auctions three years ahead of the operating period, with capacity zones reflecting transmission limits [isone-fcm]. Capacity cost is allocated by a customer's share of the single regional peak hour, so one afternoon per year matters.
NYISO	New York	The Installed Capacity market procures resource capability to meet peak load plus a reserve margin, with capability-period, monthly, and spot auctions [nyiso-icap], and locational requirements that make downstate capacity more expensive than upstate.
ERCOT	Texas (choice areas)	No capacity market. Transmission cost is allocated by 4CP: the average of a customer's demand at the four ERCOT system-peak 15-minute intervals in June through September sets the following year's transmission charges for 4CP-billed classes [oncor-tariff]. A site that curtails at the right four moments avoids a year of cost; the state has been reviewing whether to change the method.
CAISO and non-RTO	California, Nevada	California's Direct Access is limited to non-residential customers under a statewide load cap of roughly 28,800 GWh, entered through a lottery with notices filed during the second full business week in June [cpuc-direct-access]. Nevada has no retail choice program; a customer averaging 1 MW or more (8,760,000 kWh a year) can apply to the PUCN to leave NV Energy's bundled service and pay an impact fee [nac-704b].

Those are six different answers to "what does a supplier contract control." In PJM and New England a contract can fix or pass through capacity, and peak management on a handful of hours lowers next year's cost. In ERCOT the lever is transmission via 4CP, and the contract is energy only. In California the question is whether you have a Direct Access slot at all. In Nevada a single large site might justify a 704B filing, but a dispensary chain never will. State pages on this site, such as Illinois, Pennsylvania, Massachusetts, and Texas, go through each jurisdiction's mechanics.

Prices differ as much as rules. In June 2026 the average commercial price ranged from 8.66 cents per kWh in Texas and 9.83 in Nevada to 23.56 in New York, 24.52 in Massachusetts, and 27.33 in California [eia-epm-5-6-a]. Wholesale volatility differs too: EIA reported that in June 2026 the PJM Mid-Atlantic hub had the widest monthly range of any region, 36 to 216 dollars per MWh [eia-wholesale-markets]. A one-cent saving on a Texas site is worth a fraction of a one-cent saving on a Massachusetts site, and a New England site's exposure to a bad month is larger than a Texas site's. Effort should follow the dollars.

Staggered expirations: the core of the portfolio approach

A single-site operator picks a contract term, usually 12 to 36 months, and reprices at the end of it. An MSO with every site on the same calendar is doing the same thing with the whole company: all of its supply cost reprices in one month, at whatever the forward curve shows that month. If that month is one of the bad ones, the entire portfolio carries it for the next two or three years.

Staggering fixes this the way dollar-cost averaging does. Split the portfolio into tranches that expire in different years, and each year only a slice reprices. The company's blended supply price then moves toward the multi-year average of the market instead of jumping to wherever the market sat on one signing date.

A workable design, with every number an assumption you should adjust:

• Group by zone, not by site. Sites in the same utility zone should share a contract so they can be priced and administered together; a supplier will price ComEd load as ComEd load whether it is one meter or nine.
• Three tranches, roughly equal in annual MWh. Tranche A expires in year one, B in year two, C in year three, then each rolls to a new three-year term as it comes up. After the first cycle, a third of the load reprices every year.
• Layer within a tranche where products allow. Block-and-index lets you fix 50 or 70 percent of a zone's load and float the rest, and to add blocks when forward prices dip, which is a second, finer stagger inside the tranche.
• Keep a rolling 18-month window. Anything expiring inside 18 months is actively priced; anything outside is monitored. Suppliers will quote 12 to 18 months ahead in most markets, and locking a renewal early is often possible.
• Match tranches to site life. A cultivation site with a 10-year lease and stable load belongs in a long tranche; a dispensary in a lease with a two-year option does not. Termination fees, described in contract terms, early termination fees, and evergreen clauses, are the cost of getting this wrong.

The tradeoff is administration: three expirations a year instead of one. That is what the data and governance sections below are for.

Aggregation: when many meters become one account

Aggregation means presenting multiple meters to a supplier as a single load so they price and contract it together. It works within a utility zone and market, and it has real advantages: larger volume gets a supplier's attention and often a lower margin, one contract replaces a dozen, and a portfolio with mixed profiles (flat cultivation load plus daytime retail plus peaky processing) has a better combined load shape than any of its parts, which suppliers price into the offer.

It has limits. A supplier cannot aggregate a PJM account with an ISO-NE account into one price; it can offer a master agreement with a schedule per zone, which is worth asking for. Small accounts, such as individual dispensaries at 40,000 to 80,000 kWh a year, are often below the size a supplier will price individually, and aggregation is what makes them contractable at all. And a few states impose product restrictions on small non-residential accounts that do not apply to large ones, which is another reason to know each account's class before assuming it can join the group. Energy broker vs. direct supplier explains who does the aggregation work and how they are paid.

Aggregation also changes the demand-side picture. Utility demand charges and ratchets remain per meter, always, so a site with a bad ratchet does not get help from being in a group contract. Those are fixed site by site, through the methods in demand charges explained.

Data standardization: the unglamorous prerequisite

None of the above works without a single dataset covering every meter. The minimum:

Field	Why it matters
Utility, rate class, service voltage	Determines the delivery tariff and the class rules on contracting
ISO and zone	Determines what a contract can control (capacity, transmission, energy only)
12 months of kWh and peak kW by month	Baseline for pricing and for spotting rate class errors
Interval data, where the meter records it	Needed for load-shape pricing, capacity tag management, and 4CP planning; see interval data and AMI meters
Capacity tag or PLC, in ISOs that use one	The number that sets next year's capacity cost
Supply contract: supplier, product, price, start, end, notice date, termination terms	The tranche map
Site plan: expansions, closures, lease end	Changes the load and the term you should sign

Get it into one place, in the same units, refreshed monthly from bills and portals. Submetered sites inside a shared building need their own row and a note on which utility meter they roll into; submetering multi-tenant cannabis buildings covers the mechanics. The supplier contract comparison worksheet is a starting template for the contract fields.

Governance: who decides, and by what rule

The decision to fix three years of supply cost for a 30-site company is a hedging decision, not a facilities decision, and it should be governed like one. A short written policy usually covers it:

1. Ownership. One person, typically in finance, owns the dataset and the calendar. Site managers feed it; they do not sign contracts.
2. Allowed products. Fixed, block-and-index within stated fixed percentages, and index with a cap, for example. Products outside the list need approval.
3. Term limits and stagger rule. Maximum term, target tranche structure, and the rule that no more than a stated share of annual MWh expires in any twelve months.
4. Signing authority. Who can execute, at what dollar threshold, with what second signature.
5. Review cadence. Monthly data refresh; quarterly review of forward prices against the rolling 18-month window; annual review of the policy itself.
6. Peak-day protocol. For sites in PJM, ISO-NE, NYISO, and ERCOT, a written procedure for what each site does when a peak alert is issued, since a few hours of curtailment set a year of capacity or transmission cost.

A worked example of the stagger rule, with assumed numbers: an operator with 60,000 MWh a year across PJM, ISO-NE, and ERCOT sites sets a rule that no more than 40 percent expires in any twelve months. The PJM tranche (30,000 MWh) is split into two contracts expiring a year apart; the ISO-NE tranche (20,000 MWh) expires the following year; the ERCOT tranche (10,000 MWh) is on a shorter energy-only term that can float because it carries no capacity exposure. In any given year, between 15,000 and 24,000 MWh reprices, never the whole book.

A checklist to start with

1. List every meter and fill the data table above, even if some rows are blank at first.
2. Map each meter to its ISO and confirm whether it is contractable this year with the state deregulation eligibility checker.
3. Build the expiration calendar and identify the cluster: the month in which the most MWh currently reprices.
4. Fix rate class and ratchet problems site by site; these are independent of supply.
5. Design tranches by zone, set the stagger rule, and write the one-page policy.
6. Aggregate within zones and go to market on the rolling window.