Storage economics: what actually gets paid for

The short version

Battery hardware is cheap and the technology is proven. That stopped being the question for most of 2026. What decides whether a storage project gets built is whether some rule pays for the job the asset does: a frequency product, a capacity contract, a duration-weighted accreditation, an offtaker willing to sign, or a discount rate the project can survive while it waits in the queue. Where that rule exists, storage gets built quickly. Where it doesn't, you get unpaid batteries, solar that can't be financed, and gas peakers paid to sit idle.

The mechanism

Storage earns money in three ways: it sells energy across a price spread, it sells grid services, or it gets paid simply for being available. Only the first of these is open everywhere, and it erodes as more batteries chase the same spread. The other two exist only where a regulator or a buyer has written them down. A two-hour battery and a twelve-hour battery look identical on a capacity spreadsheet, and only one of them carries a grid through a still, cloudy week. If no rule tells those two assets apart, capital funds the cheaper one.

Financing has its own version. Connection queues now run for years, so the interest rate during that wait becomes part of the project's cost.

How it developed

• April 23 post, Tesla's storage miss: I argued it was commoditization, not collapse. Once cells are cheap and plentiful, the edge moves away from chemistry and toward whatever makes the revenue bankable.
• May 9 post, California's battery milestone: batteries took over the evening shift, but the gas peakers stayed online and kept collecting capacity payments as insurance. This was the first clear case of the payment rule, rather than the physics, deciding the mix.
• May 11 post, Megapacks in Memphis: the first big buyer of cheap batteries at data centers was gas turbines that can't ramp fast enough. Lenders understand gas, and the credit case for off-grid solar plus storage is still being written.
• June 29 and June 30 posts: long-duration storage is technically proven but has no revenue model in most US markets. The US-China gap reads better as a constraint stack (queue, EPC, cells, revenue stack) than as a race.
• July 14 post, Spain's solar glut: negative prices bring household bills down while merchant returns fall apart. A market priced for scarcity was never repriced for abundance.
• August 2 and August 19 posts: a New York behind-the-meter project gave up grid export revenue to avoid a permit, and home batteries scaled in New York with no subsidy. Market structure did what the tax credit didn't.
• August 25 and 26 posts: storage queues of over five years met Treasury yields at a 20-month high, so every extra year waiting adds rate exposure. And 42 GW of Indian solar couldn't move forward without a battery attached.
• September 11 post, Spain: frequency reserve was mandatory and unpaid, so the fastest responder on the system earned nothing for its best service.
• September 18 and October 5 posts: California's no-blackout summer was a portfolio result, not a battery result. Solar has won on cost; the fight has moved to grid, storage and offtake.

Where it stands now

• Spain: the European Commission approved a capacity market worth roughly €9 billion, open to generation, demand response and storage (energy-storage.news, May 2026). The ministry announced the order on September 16, 2026, with contracts of up to 15 years for new investments (ESS News, September 17, 2026). I could not confirm a date for the first auction. Analysts say de-rating factors will decide what batteries actually earn from it.
• Britain: Ofgem's minded-to decision on June 26, 2026 shortlisted 16 long-duration projects, about 7.65 GW with 8 to 22 hours of storage, for cap-and-floor support. Final awards are expected this autumn (Hill Dickinson, Foot Anstey). As of today I have not seen final determinations.
• US rules: Treasury's interim FEOC guidance (Notice 2026-15, February 2026) sets a 55% material-assistance threshold for storage that begins construction in 2026. Tax counsel say it leaves open questions on recapture and contract-level control (Morgan Lewis, March 2026).
• US build: Wood Mackenzie and ACP reported a record 18.9 GWh deployed in Q2 2026, and withdrew an earlier contraction forecast because projects were rushing to secure tax-credit safe harbor (ESS News, October 1, 2026). The pace of the build is again being set by a rule's calendar.
• Cost of money: the 10-year Treasury went from 4.66% on August 26 to 5.31% on October 5, then eased to 4.99% on October 27 (FRED, DGS10). Any pro forma underwritten in summer is running on a discount rate that is now out of date.
• California: the first state-run procurement of up to 2 GW of long-duration storage is scheduled to open in late 2026 (CPUC). I could not confirm it has launched.

The pushback worth taking seriously

Across these posts, the pressure commenters came back to most was firmness. Batteries move energy in time; they don't make it. On the California posts, commenters kept pointing out that gas carried a large share of supply in late August, that imports do real work, that four-hour batteries cover the evening ramp but not the night, and that winter is the binding season.

I take most of that as correct. Gas and imports cover firmness because firmness is what capacity and resource-adequacy payments already reward, and they reward it in a form gas was built to supply. Whether storage can take a larger share of that job depends on whether duration gets accredited and paid. The batteries themselves aren't the limit.

Two other patterns sharpened the read. Several commenters pointed out that storage erodes its own merchant revenue, because each new battery narrows the spread the others earn on. That pushes value toward contracted floors. On the queue post, commenters said standalone and behind-the-meter storage connects much faster than the national median. Fair: configuration is a lever.

What it means if you sell into this

• Underwrite by rule, not by spread. For each market, write down which services are paid, which are contracted, and how your duration is accredited. Treat the merchant tail as something that shrinks as more batteries arrive.
• Price the wait. Re-run queue time at today's discount rate. Configurations that skip co-location queue inheritance, or that sit behind the meter, are financing levers as much as engineering choices.
• Treat FEOC as a bankability item. Supplier certification and contractual risk allocation now sit next to the cell price in diligence. US EV battery plants now being rerouted to grid storage are a supply option worth pricing.
• Answer the firmness question before it's asked. Say which hours you cover and what accreditation you hold, and claim nothing you aren't paid for.
• Show up while rules are written. Spain's auction design, Britain's second window and California's solicitation are being set now. Nobody gets to design the market after they need it.

The question I'd want answered first: If your next financing close lands before Spain sets its de-rating, what are you underwriting duration on?

What would change my read

• Merchant-only storage financing at scale in a market with no new floor. For example, if Spain builds out on arbitrage alone before its first capacity auction, then cost and spreads matter more than I'm giving them credit for.
• A major ISO or FERC adopting duration-weighted accreditation without long-duration projects following. That would put the constraint back on technology and cost.
• Queue times shortening materially while yields fall. The financing half of this thesis would weaken.

If you're already seeing one of these, I'd rather hear it now than find it in the next filing.

What to watch

• Do Ofgem's final Window 1 cap-and-floor awards confirm the 16 shortlisted projects on the minded-to terms, and do they land before year-end 2026?
• When Spain sets its first transitional capacity auction, what de-rating does it give two- and four-hour batteries? That number is the first test of whether the new market pays storage for availability.