California Multi-Site Cannabis Operators: One ISO, Three Utilities, Two Dozen CCAs, and One Lottery

What a California portfolio looks like on paper

Start with the map. A mid-size California operator with five to fifteen licensed sites will typically have accounts in two of the three investor-owned utility territories, and sometimes a store inside a municipal utility. Each account has a delivery schedule set by its demand, a default generation provider that is either the utility or a local CCA, and a Direct Access status of bundled, waitlisted, or enrolled.

Site	Likely utility	Likely delivery class	Default generation	DA odds after 2023 lottery
Central Valley or Emerald Triangle grow	PG&E	B-10, B-19, or an AG schedule [pge-sched-b19]	PG&E bundled or one of twelve CCAs [pge-cca]	Low: 185 GWh open, 836 customers waiting at year end [cpuc-da-lottery-2023]
Inland Empire or desert grow or lab	SCE	TOU-GS-2, TOU-GS-3, or TOU-8 [sce-rate-options-2025]	SCE bundled or one of twelve CCAs [sce-cca]	Good: 1,206 GWh open, waitlist cleared [cpuc-da-lottery-2023]
San Diego store or lab	SDG&E	TOU-A under 20 kW, AL-TOU family above	San Diego Community Power or Clean Energy Alliance [sdge-cca]	Low: 45 GWh open, 194 waiting [cpuc-da-lottery-2023]
Los Angeles city or Sacramento store	LADWP or SMUD	Muni schedule	Muni only	None: outside DA and CCA

The whole state, for the three IOUs, is one grid operator. There is no PJM-versus-MISO seam to price across, no capacity auction with different zonal prices, and no second set of settlement rules. That makes a California portfolio easier to price than a Midwestern one, and it means the strategy questions are almost all about the retail layer: which sites can get a supply contract, which are stuck on a CCA, and how to keep the data comparable.

The lottery decides what you can aggregate

Direct Access is capped at roughly 28,800 GWh statewide, split among the three utilities, and new load enters only by winning a spot in the June lottery [cpuc-direct-access]. PG&E's share is an Overall Load Cap of 11,393 GWh, and its 2026 enrollment window ran from 9 am on June 8 to 5 pm on June 12 [pge-direct-access]. SDG&E's DA Load Allowance is 3,942 GWh [sdge-direct-access]. SCE runs the same second-full-week-of-June process [sce-direct-access].

For a multi-site operator, the mechanics of the filing matter. PG&E's Six-Month Notice is Form 79-1117 plus an Excel list of every Service Agreement ID you want included, emailed during the window [pge-direct-access]. That means every PG&E account you own can go in as one filing. But the randomizer runs per utility, against that utility's headroom. In 2023, SCE had 1,206 GWh available for 326 filers and cleared all of them, while PG&E had 185 GWh for 560 filers and SDG&E 45 GWh for 267 [cpuc-da-lottery-2023]. A portfolio with grows on PG&E and labs on SCE will very likely end a lottery cycle with the SCE sites eligible and the PG&E sites on a waitlist that expires on the last business day of December [pge-direct-access].

The practical rule: aggregate within a utility, never assume across utilities, and be ready to contract for a partial portfolio. A supplier will price the accounts that cleared. Structure that contract so additional accounts can be added at the same terms when they clear a later lottery, and keep the term end aligned to a June window so the next cycle's winners can be folded in.

Sites that were already on DA when you acquired them are the exception. Load that is already inside the cap stays there as long as it stays on DA. When you buy a facility, ask whether its accounts are DA-enrolled; that status is worth real money in PG&E and SDG&E territory, and it is lost if someone returns the account to bundled service, which triggers an 18-month minimum commitment before it can go back [sce-direct-access] [pge-direct-access].

CCAs: the sites you cannot aggregate but must still manage

Most of a California portfolio will sit on Community Choice Aggregators. Enrollment is automatic and opt-out; the CCA sends at least two notices in the 60 days before service starts, opting out is free during that window, and after the first 60 days of service an opt-out may carry charges and requires a one-year stay on bundled service before rejoining [cpuc-cca-faq]. PG&E lists twelve CCAs in its territory, SCE twelve, and SDG&E two [pge-cca] [sce-cca] [sdge-cca]. A statewide operator can easily have accounts on six or eight different CCAs, each with its own rate schedule for commercial customers, its own default product (many offer a base tier and a 100 percent renewable tier at a premium), and its own board that sets rates on its own calendar.

You cannot negotiate with a CCA the way you negotiate with a supplier. What you can do is decide, account by account, whether the CCA's commercial rate plus the PCIA beats the utility's bundled generation rate, and whether the CCA's renewable tier is worth paying for as a brand claim. Make that decision annually after the October PCIA reset [cpuc-pcia], and keep the opt-out rules in mind: leaving a CCA is a one-year commitment to the utility.

One procurement calendar for the state

Because the DA lottery and the PCIA reset are both annual, a California portfolio runs on a fixed calendar. Everything else fits around it.

When	What	Why
January	Pull twelve months of interval data for every account; reconcile which accounts are on which utility, schedule, CCA, and DA status	Waitlists from the previous June expired at the end of December [pge-direct-access]; you are starting clean
February to April	Model each site against its CCA, bundled, and a supplier indicative price; decide which accounts go in the June notice	Suppliers can give indications, but cannot contract for load that has not cleared
Second full business week of June	File Six-Month Notices with each IOU, listing every SAID you want in [pge-direct-access] [sce-direct-access]	The only window all year [cpuc-direct-access]
July to August	Learn which accounts cleared; go to market for those	Winners' six-month clock starts; service begins the following January
October	Review each utility's PCIA re-forecast and its effect on each vintage in the portfolio [cpuc-pcia]	Changes the CCA-versus-bundled answer for every account
November to December	Finalize supply contracts for cleared accounts; decide on CCA opt-outs or product changes; DASR submissions ahead of January service	Waitlisted accounts expire on the last business day of December and must be refiled [pge-direct-access]

Staggered contract terms, which are a hedging tool in open-choice states, are less useful here because you cannot freely re-enter DA. If you win a slot, a longer term with extension options is usually better than a series of short ones, since the alternative to renewing is not another supplier but the utility's bundled rate. See fixed versus index versus block-and-index for how to structure the product once you have the right to buy it.

Standardizing data across sites

Comparing a B-19 grow in Fresno with a TOU-GS-2 lab in Ontario and a TOU-A store in Chula Vista is hard because the bills do not use the same words. Build one table with, for every account: utility, service agreement ID, delivery schedule, CCA or bundled or DA, DA vintage year, PCIA rate, maximum demand by month, on-peak demand by month, kWh by TOU period, and total cost. All three IOUs put the schedule on the bill and all three have interval meters on demand-billed accounts, so the raw material exists; see interval data for how to request it.

Two normalizations make the sites comparable. First, express every account's all-in price in cents per kWh, so it can be checked against the June 2026 statewide commercial average of 27.33 cents [eia-epm-5-6-a]; sites far above it are the first to investigate. Second, express every demand-billed account's demand charges as a share of total cost. A grow at 35 percent and a lab at 12 percent are behaving normally; a store at 30 percent is on the wrong schedule. The tenant-and-landlord version of this problem, where several licensees share one building and one meter, is covered in submetering.

Where the savings are, honestly

In most deregulated states a multi-site operator's largest energy lever is the supply contract. In California it is usually not, because most sites cannot get one. The levers that are always available are schedule optimization (the right B, TOU-GS, or AL-TOU option for each site's shape), on-peak demand management on the grows, the CCA-versus-bundled decision after each PCIA reset, and utility efficiency rebates. The supply contract is the upside case for whichever accounts clear the lottery, and it is worth pursuing every year because it is free to file.

For a portfolio that also operates in other states, the multi-state procurement guide covers ISO seams and capacity markets. For contrast with a state where every site can contract, compare New York multi-site, and for a capped-choice market with a queue rather than a lottery, Michigan multi-site.