BatteryJoy Weekly — 2026-W31 (Jul 27–Aug 2)

The same week Eolian broke ground on the largest battery project to ever clear a PJM capacity auction — a 1.06 GWh plant built specifically to buy time for an overloaded data-center grid — Washington reset the tariff math underneath every battery import the industry needs to build the next one, expiring one duty and imposing another within days of each other.

WTI spot held almost flat at $84.25/bbl (Jul 27) versus $84.38 the prior week, after spiking to $93.08 mid-week (EIA). Henry Hub kept easing, down to $2.63/MMBtu (Jul 27) from $2.93 the prior week's close — comfortable fuel costs on both sides of the ledger.

But the number that matters sits downstream of fuel price entirely: BloombergNEF's July 21 report puts US data centers at 5.9% of electricity today, rising toward 20% by 2035, while new grid-connection capacity adds only 7.1 GW a year against a 194 GW target — a 19 GW annual shortfall by decade's end. Cheap gas and oil don't close a connection-capacity gap; only new capacity, wherever it can be sited fastest, does.

• WTI $84.25/bbl (flat wk/wk); Henry Hub $2.63/MMBtu (▼ from $2.93) — fuel costs aren't the constraint.

• BNEF: data centers reach ~20% of US power by 2035 against a 19 GW/year connection shortfall.

• The gap between demand growth and grid buildout is exactly what routes new load toward storage sited closer to the point of need.

Lithium eased again to $21.57/kg globally (Jul 29) as restarted mine capacity keeps feeding oversupply fears; China's domestic benchmark slipped 0.36% to ¥146,000/ton (Carbon Credits, Jul 29). Cobalt, by contrast, sat dead flat — $25.53/lb as of Jul 20, 56,290 RMB/t on Jul 28 — unchanged from a month earlier, while graphite stays weighed down by weak battery demand and idle capacity.

The read: this isn't one commodity story, it's a divergence. Lithium is still working off a supply overhang that keeps pushing lower; cobalt and graphite are simply stalled, waiting on a demand catalyst neither has yet. Cheap lithium lowers cell input costs right as storage buildout (see Downstream) needs every margin point it can get — but a stalled cobalt/graphite market is a signal that the demand pull isn't broad-based yet.

• Lithium $21.57/kg (Jul 29), still drifting lower on mine-restart oversupply.

• Cobalt flat at $25.53/lb / ¥56,290/t for a full month — no catalyst either direction.

• [NEED: source] a citable battery-grade Li carbonate $/t level (Benchmark/Fastmarkets paywalled).

No fresh cell-chemistry or manufacturing story cleared the 7-day window this week. The standing candidate — China's GB 38031-2025 "no fire, no explosion" traction-battery safety standard — took effect Jul 1, but its only available coverage is a month-old article, too stale to carry as this week's news.

That's worth naming rather than papering over: a quiet midstream week doesn't mean the chemistry race paused, it means nothing new enough to verify broke this specific week. [NEED: source] a fresher chemistry/manufacturing item, if one surfaces before publish.

• No in-window midstream story this week — compressed rather than padded.

• [NEED: source] a fresher cell-chemistry/manufacturing development for this window.

Eolian began construction Jul 29 on Flint Grid — 200 MW / 1.06 GWh, 5+ hour duration — in Licking County, Ohio, sited directly against the New Albany data-center corridor. It's the largest project to clear PJM's 2027/28 capacity auction, over half of all new battery capacity cleared for that year, and the first grid-battery project permitted under Ohio's Power Siting Board (PR Newswire).

This is this week's macro gap (Section 2) showing up as a shovel in the ground: when new transmission and generation can't be sited or built fast enough for a data-center corridor's load growth, a multi-hour battery plant is the fastest thing that can legally and physically close the gap in time. Meanwhile Europe's BEV market posted a record June — 275,060 units, ▲39.5% YoY, crossing 25.6% share across 17 key markets — while China hit a record 63% NEV share (43% BEV + 20% PHEV) on ~1.6M vehicles sold, growth driven by high gas prices and PHEV-incentive timing rather than any policy pause. Two different demand stories, one throughline: whatever chemistry and format, batteries are absorbing load growth faster than the grid around them can expand.

• Flint Grid: 200 MW / 1.06 GWh, first Ohio Power Siting Board-approved grid battery, built for a data-center corridor.

• It's >50% of all new PJM battery capacity clearing the 2027/28 auction — a single project setting the regional pace.

• Europe BEV: record 275,060 units in June (▲39.5% YoY, 25.6% share); China hit a record 63% NEV share.

USTR imposed a new 12.5% tariff on all Chinese goods over a forced-labor enforcement finding, effective Jul 24 — the same date the existing 10% Section 122 global tariff was set to expire. For Chinese EV-battery imports already stacking a 25% Section 301 duty on top of Section 122 (35%+ combined), that's two rate changes landing in the same window, reshuffling the import-cost math every buyer just finished planning around.

For BESS developers mid-project — like the one breaking ground in Ohio this week — that kind of tariff churn is exactly the variable FEOC/PFE sourcing rules were already forcing them to price around. Non-Chinese cell supply (South Korea, Japan, Southeast Asia assembly) becomes relatively more attractive with each stacking change, reinforcing the supply-chain diversification already underway rather than reversing it.

• New 12.5% China tariff (forced-labor finding) effective Jul 24, same week Section 122's 10% was due to expire.

• Chinese battery-cell imports already stack 35%+ in combined duties before this change.

• Net effect: further incentive to diversify cell sourcing away from China, layering onto existing FEOC pressure.

Ohio just proved a multi-hour battery plant can go from permit to groundbreak fast enough to matter for a data-center corridor that can't wait on transmission — but the tariff floor under every cell in that plant moved twice in one week. When the fastest-to-build grid asset also has the least stable input-cost picture, which risk wins: the speed advantage, or the sourcing uncertainty?

• Flint Grid shows storage can outrun the grid buildout timeline.

• The same week, tariff changes reset the cost floor under every cell it needs.

• Reply and tell me which risk you'd underwrite first.