Here is a problem we are quietly running for a dozen commercial clients right now. A building owner installed a well-designed solar array in 2019, locked in a legacy rate at Permission to Operate, and has been happy ever since. The system zeroes out most of the energy charge. Then we model what their bill looks like once PG&E moves them onto a current B rate — and the projected monthly cost jumps 20% to 40%, even though nothing about the building or the array has changed. The culprit isn’t energy. It’s demand charges, and a peak window that has marched three hours later into the evening.
That migration isn’t hypothetical, and it isn’t optional. PG&E’s legacy commercial rate grandfathering ends July 31, 2027 for most commercial customers (December 31, 2027 for public schools). When it ends, every grandfathered commercial and industrial account transitions to one of the current B-series rates — B-1, B-6, B-10, B-19, or B-20. If your facility carries a meaningful load, this is the single biggest line-item change coming to your utility bill, and it rewards anyone who plans for it twelve months early.
What actually changes in 2027
Three structural shifts matter more than the headline rate numbers. First, the summer season shrinks from six months to four (June through September instead of May through October), which concentrates the most expensive pricing into a tighter window. Second, the on-peak period moves to 4:00–9:00 PM — the evening hours — where it used to sit around noon to 6:00 PM. Third, and most consequential for anyone with solar, the demand charge structure on the medium and large rates (B-19 and B-20) is layered: a standard maximum demand charge applies year-round, and during summer an additional maximum demand charge applies for every hour except 9:00 AM to 2:00 PM, plus time-of-use demand charges tied to the peak and part-peak periods.
Read that exemption window again: the additional demand charge is waived only from 9:00 AM to 2:00 PM. That is precisely the window where a south-facing solar array is at full output and naturally suppresses your demand. PG&E has, in effect, designed the demand charge to bite hardest at the exact hours solar can’t help you — late afternoon and evening.
The demand charge mechanics most people miss
Demand charges are billed on your single highest 15-minute average power draw in the billing period, measured in kilowatts, and the rate runs in the rough neighborhood of $20 to $28 per kW per month on B-19/B-20 secondary service. (Verify the current figure against your own tariff sheet — the exact $/kW depends on voltage level and vintage.) The important engineering point is that a demand charge doesn’t care how much energy you used. It cares about one bad spike.
A 400 kW peak at $25/kW is a $10,000 monthly charge, or $120,000 a year, set by a few 15-minute intervals. Solar panels alone do almost nothing about that peak if the spike happens at 6:30 PM, because the array is winding down or already dark. This is the same dynamic we wrote about for homeowners in NEM 3.0 Realities: the value of midday solar collapses once the rate structure pushes the expensive hours into the evening. The commercial version is just larger and is dominated by demand charges rather than export credits.
Why a 2018-era array no longer covers the bill
Most commercial arrays installed before 2020 were sized and oriented to crush the energy charge under the old noon-to-six peak. South-facing, fixed tilt, no storage — a perfectly rational design for the rules at the time. Under the 2027 structure, that same array produces a textbook duck curve mismatch: full production at 11:00 AM when the additional demand charge is waived anyway, and near-zero production at 7:00 PM when the on-peak demand and energy charges are at their worst.
We see three predictable failure modes when we run the migration. The array offsets energy it didn’t need to offset (the 9 AM–2 PM hours). It leaves the evening peak fully exposed. And because the building’s demand spike often lands in the early evening — HVAC recovery, EV charging, kitchen or production loads ramping — the demand charge is essentially untouched by the solar. The owner did everything right in 2018 and still gets a worse bill in 2027.
The engineering fix: size storage to the demand profile, not the energy bill
The lever that actually works is battery storage, sized and dispatched against the demand charge rather than against kilowatt-hours. The control logic is straightforward: the system watches the building’s real-time demand and discharges to shave the 15-minute peaks, then shifts stored midday solar into the 4:00–9:00 PM window to cut both the on-peak energy and the time-of-use demand charge. We model this the same way we model commercial battery paybacks generally, which we walked through in The Hidden ROI of Commercial Batteries.
A worked example: a facility peaking at 400 kW that can reliably shave 120 kW with a correctly sized battery saves roughly 120 kW × $25/kW = $3,000 a month on the standard demand charge alone, before counting the additional summer demand charge and the energy arbitrage. That is $36,000+ a year from peak shaving, on top of the solar energy savings. The battery has to be sized to the shape of the load — power rating (kW) for the shave, energy capacity (kWh) for how long the peak lasts — not to a rule of thumb.
Option S is worth modeling
PG&E offers Option S, a storage-oriented rate available on B-19 and B-20 that swaps the monthly maximum demand charge for a lower daily demand charge — structurally friendlier to a battery that can guarantee a daily shave. Enrollment is capped (150 MW program-wide, with sub-caps per rate class), so it is first-come. For storage-heavy sites it can materially change the payback. We model standard B-19/B-20 against Option S side by side before recommending either.
What to do in the next twelve months
The customers who come out ahead in 2027 are the ones who treat the next year as design time, not a fire drill in July. Concretely: pull your interval data and identify when your real demand peaks actually occur — most owners are surprised it’s the evening, not midday. Model your bill on the B rate you’ll be forced onto, not the legacy rate you’re leaving. If you’re adding storage or a new array, file interconnection early, because the queue is the bottleneck, not the install — a point we made for general contractors in Navigating PG&E Interconnection Delays. And if you operate multiple buildings, sequence the upgrades so the worst-exposed meter goes first.
None of this requires guessing. The tariff is published, the sunset date is fixed, and your own interval data tells you exactly where the money leaks. The only real variable is whether you model it now or absorb the increase in 2027.
A demand charge is set by your worst fifteen minutes of the month. Solar alone rarely touches it once the peak moves to the evening — but a battery sized to the load profile can, and the 2027 restructure is what finally makes that math obvious.