BatteryJoy Weekly — 2026-W42 (Oct 12–18)

PJM just answered its own data-center power crunch with a number worth sitting with: of the first 2.1 GW it fast-tracked onto the grid this week, 1.7 GW — 81% — is battery storage, not gas or nuclear, even as WTI climbed ~4% and lithium prices stayed dark for a seventh straight week and two Chinese graphite-related export-control suspensions tick toward lapsing within the next month.

WTI crude rose from $97.89/bbl on Oct 2 to $101.91/bbl on Oct 14, a roughly 4% climb across the window (EIA, direct API pull), even as U.S. commercial crude stocks excluding the SPR fell 8.6 million barrels to 415.6 million for the week ended Oct 9 — a sharper draw than the prior week's 3.19-million-barrel pullback and the second consecutive weekly draw (EIA Weekly Petroleum Status Report, via secondary republication).

Natural gas told a steadier, supplied story. Henry Hub firmed from $2.90/MMBtu on Oct 2 to $3.22/MMBtu by Oct 14 — breaking through the sub-$3.20 band that had held since mid-September — while Lower-48 working gas storage still added 67 Bcf to reach 3,567 Bcf for the week ended Oct 9, a build that keeps inventories 45 Bcf above the five-year average even as the pace slowed from the prior week's 85 Bcf addition.

• Crude's second straight weekly draw lines up with a ~4% WTI climb — oil's supply-and-price signals are now moving the same tightening direction after weeks of mixed data.

• Gas keeps building above its five-year average (+45 Bcf) even as Henry Hub pushes to a multi-month high — comfortably supplied, not tight, unlike the oil side.

No fresh in-window lithium carbonate or spodumene print surfaced this week — Benchmark Mineral Intelligence's own pricing page still shows Sep 2's $19,750/tonne CIF Asia carbonate and $2,254/tonne FOB Australia spodumene as the latest posted assessment, now nearly seven weeks stale. In the absence of a price move, the real upstream story is two separate regulatory clocks in China: a battery-and-graphite-anode export-control suspension set to lapse Nov 10, 2026, and a broader graphite/critical-minerals suspension set to lapse Nov 27, 2026 [NEED: source] for any 2026 update confirming either still stands unchanged.

Both dates sit three to six weeks out — a near-term window for cell makers and BESS pack suppliers to plan around a possible tightening in the anode material most directly tied to midstream cylindrical-cell and sodium-ion cost, the lane's clearest causal link this week.

• Lithium pricing is stale for a seventh straight week — any absolute figure in circulation should be read as a September print, not current.

• Two Chinese export-control suspensions (batteries/graphite anode, Nov 10; graphite/critical minerals, Nov 27) both lapse within the next month — the next real upstream data point is a policy date, not a price print.

A September Hebei storage tender put real numbers on sodium-ion's 2026 commercialization story: sodium-ion systems bid at an average of ~$153/kWh (1.0322 yuan/Wh) against ~$79/kWh (0.5318 yuan/Wh) for LFP in the same project — sodium-ion at roughly double LFP's cost even as it wins real tenders. That hasn't slowed deal-making: CATL and HyperStrong signed a 60 GWh, three-year sodium-ion supply agreement on Apr 27, 2026, while Alsym/Juniper (500 MWh) and ESS Tech (8.5 GWh) both committed to US-made sodium-ion product.

All-solid-state stayed the slower story by comparison — EVTank's July 2026 white paper counted 8.0 GWh of global solid-state shipments in 2025 (up 51.9% y/y), all of it semi-solid; full all-solid-state mass production still isn't expected before 2027. No fresh gigafactory-capacity announcement landed in-window this week, so this section stays compressed to the chemistry-cost read rather than padding with a stale capacity item.

• Sodium-ion's real-tender price (~$153/kWh) is roughly 2x LFP's (~$79/kWh) in the same Hebei project — the chemistry is scaling on supply deals, not yet on cost parity.

• CATL-HyperStrong's 60 GWh sodium-ion agreement (Apr 2026) is the lane's largest confirmed forward commitment, landing the same year all-solid-state remains semi-solid-only.

PJM selected its first batch of projects under its Expedited Interconnection Track on Oct 8, 2026 — 2.1 GW total, confirmed by PJM's own announcement: two Engie IR Holdings standalone battery-storage projects (860 MW at Crawford, Pennsylvania, and 800 MW at Morrow, Ohio — 1.7 GW combined) plus a 455 MW gas-fired uprate to LS Power's existing Hunterstown plant in Adams, Pennsylvania. All three are expected online by mid-2029. The EIT itself — FERC-approved June 9, 2026 — lets PJM fast-track up to 10 interconnection requests a year for resources of at least 250 MW that can come online within three years, running through the end of 2027; PJM says its broader generation push is "mainly driven by planned data centers," citing thin new-generation growth against rising demand forecasts.

The storage-heavy mix is the number worth sitting with: 1.7 of the first 2.1 GW PJM fast-tracked toward its data-center-driven capacity shortfall is battery storage, not gas or nuclear — a direct, primary-sourced data point for the compute-buildout-to-power-demand-to-storage-response chain this newsletter has been tracking all year. That's still a small fraction of the demand it's answering: Dominion Energy's Virginia pipeline alone had roughly 51 GW of contracted data-center capacity as of Q1 2026, climbing further since, by one company report [NEED: source] for an October-dated update specifically.

• PJM's first EIT batch (2.1 GW, Oct 8) is 81% storage (1.7 GW) — the grid operator's own emergency mechanism for data-center load is leaning on BESS as the fastest-build answer.

• PJM explicitly ties its fast-track program to data-center demand outstripping new generation — one of the clearest primary-sourced statements of the causal chain this newsletter tracks.

• The gap is still wide: 2.1 GW fast-tracked against Dominion's ~51+ GW contracted pipeline alone shows how far fast-track mechanisms are from matching the scale of demand already signed.

No new October rule landed, but the standing schedule keeps compounding: Treasury's FEOC material-assistance thresholds for battery storage sit at 55% non-prohibited-foreign-entity cost share for projects starting construction in 2026, stepping up to 60% (2027), 65% (2028), 70% (2029), and 75% (2030+) — with several contested guidance questions still open as of the most recent tracking located this session [NEED: source] for an October-specific update. Falling below threshold makes a project ineligible for its underlying tax credit, a binding near-term constraint on sourcing decisions BESS and EV-pack developers are locking in now.

Separately, the Section 301 tariff on Chinese non-EV lithium-ion batteries stands at 25% as of Jan 1, 2026 (with one estimate of an effective combined rate near 58%) — no new October change identified this session [NEED: source] for primary USTR confirmation. Layered against the two upstream-flagged Chinese export-control suspensions lapsing within the next month, the policy picture is several standing constraints compounding rather than any single new action.

• FEOC's 55%→75% non-PFE cost-share ramp (2026–2030) remains a bigger near-term constraint on BESS/pack sourcing than any single new rule this week.

• The China tariff stack (25% Section 301, ~58% effective per one estimate) and the two upstream export-control suspensions lapsing in the next month are converging on the same supply chain from different directions.

This week's clearest number is also its starkest gap: PJM's own emergency fast-track mechanism chose battery storage for 81% of its first approved batch — proof the grid operator sees BESS as the quickest real answer to data-center-driven demand — yet that whole batch (2.1 GW) is a rounding error against Dominion's Virginia pipeline alone (~51+ GW contracted). If storage is the fastest tool PJM has, what has to change — financing, permitting, or the fast-track program's own 10-project-a-year cap — for the fast lane to actually keep pace with the demand it's racing against?