US Data Center BESS: From Backup to Grid Asset

LFP cells now price at $90–$130 per kWh from Tier-1 suppliers, giving the integrated system a 2026 cost floor of $380–$520 per kWh at 1–4 MW scale

Decision Lens

The US data center battery market sits at a structural inflection. Installed costs have dropped roughly 30% since 2022, and LFP chemistry now dominates new deployments. But falling unit economics are running into hard supply-side friction: domestic cell production covers less than 15% of demand, UL 9540A certification backlogs add 8–14 weeks to project timelines, and qualified integrators number fewer than 20 nationally. For Global Heads of Data Center Energy, the core tension is straightforward — battery storage is becoming a genuine grid-services asset, but the procurement path to get there is longer, more constrained, and more regulatory-dependent than the falling price curve suggests.

90-Second Brief

As the week closes, the US data center battery market is projected to expand from roughly $2.8, $3.5 billion in 2026 to $9.5, $12.0 billion by 2035, driven by hyperscale expansion, rising backup duration requirements, and active grid services participation. LFP chemistry captured over 70% of new US data center deployments as of 2025, displacing legacy lead-acid systems that still represent a large retrofit backlog. Hyperscale cloud operators and large colocation providers together account for roughly 65, 70% of procurement by value. Supply chain concentration and certification bottlenecks are the primary constraints on deployment velocity through at least 2028.

What’s Actually Happening

Three structural shifts are converging. First, the chemistry transition is essentially complete at the front end: LFP has displaced NMC and VRLA in new builds due to superior thermal stability, 6,000–10,000 cycle life, and lower cobalt exposure — all materially aligned with 10–15 year facility lifespans. LFP cells now price at $90–$130 per kWh from Tier-1 suppliers, giving the integrated system a 2026 cost floor of $380–$520 per kWh at 1–4 MW scale.

Second, behind-the-meter grid services are moving from pilot to procurement specification. Colocation and hyperscale operators in PJM, ERCOT, and CAISO are deploying DC-coupled BESS to generate $50–$120 per kW-year in wholesale market revenue while maintaining backup readiness. Extended backup durations — now routinely specified at 15–30 minutes versus the historical 5–10 minutes — are widening the addressable use case for each installed system.

Third, the supply chain remains structurally import-dependent. An estimated 80–85% of lithium-ion cells consumed domestically are imported, with Chinese-origin cells subject to 25% Section 301 tariffs. South Korean and Japanese suppliers carry a cost advantage under existing free trade agreements, but cell allocation for mission-critical, long-warranty applications (10+ years) is constrained regardless of origin.

Why It Matters for Global Heads of Data Center Energy?

Battery strategy is no longer separable from grid interconnection strategy. If your team is specifying BESS purely as UPS backup, you are leaving behind-the-meter revenue on the table and underutilizing assets that already carry $380–$520 per kWh in capital cost. The more immediate operational risk is timeline: UL 9540A large-scale fire test reports for new system configurations face 6–9 month wait times, and NFPA 855 compliance adds 8–14 weeks to project schedules on top of that. Combined with 20–35 week lead times for high-voltage switchgear and PCS units, a BESS procurement decision made today for a facility targeting a 2027 commissioning date is already running tight.

Supply chain concentration adds a second pressure point. CATL, BYD, and Samsung SDI collectively supply an estimated 60–70% of cells entering US data center BESS. Tariff exposure on Chinese-origin cells, combined with allocation constraints for mission-critical-qualified supply, means procurement teams need diversified cell sourcing strategies — not just preferred vendor agreements. The 45X Advanced Manufacturing Production Tax Credit offers a $35/kWh incentive for domestically produced cells, which will matter for long-term cost modeling once domestic capacity scales, but coverage remains below 15% of demand through at least 2028.

The Forward View

The period from 2026 to 2029 is likely to be defined by supply constraint rather than demand constraint. Annual data center battery deployments are expected to grow from 12–15 GWh today toward 35–45 GWh by 2035, but qualified LFP cell supply and high-voltage PCS availability will periodically limit execution velocity. Domestic gigafactory output — from facilities in Ohio, Georgia, Texas, Michigan, and Arizona — is projected to raise domestic cell coverage to 25–30% of demand by 2030, though the fraction qualifying for data center warranty and reliability requirements will be smaller still.

The grid services opportunity will intensify procurement complexity. As more hyperscale and colocation operators configure BESS for PJM and ERCOT participation, demand for energy management software capable of simultaneous backup readiness and market dispatch optimization will create a new vendor tier. Operators who lock in framework agreements with capable software-plus-hardware integrators in 2026–2027 will have scheduling and certification advantages over those entering the market later, when integrator capacity is more constrained.

What We’re Uncertain About?

Tariff trajectory for Chinese-origin cells. Section 301 tariffs currently impose a 25% cost premium on Chinese LFP cells. Whether these tariffs expand to battery components or are modified under future trade policy is unresolved. The cost differential between Chinese and South Korean supply could widen or narrow materially, affecting long-term economics for BESS-backed grid services.
Domestic cell qualification timelines. New US gigafactories are targeting data center segments, but the fraction of output that will meet 10–15 year warranty and mission-critical reliability screening requirements is not yet established. What would resolve this: published qualification data from Samsung SDI, ONE, or Kore Power for data center-grade cells by 2027–2028.
Grid services revenue durability in PJM and ERCOT. The $50–$120 per kW-year revenue range for behind-the-meter participation reflects current market conditions. How frequency regulation and demand response revenues evolve as more storage capacity enters these markets — and how utilities respond with interconnection rule changes — is a material uncertainty for BESS business case modeling.
End-of-life liability exposure. Less than 5% of decommissioned data center lithium-ion batteries are currently recycled through certified processors. As first-generation systems installed between 2020 and 2025 approach end-of-life, the regulatory and ESG reporting exposure for operators without certified recycling pathways will grow. What would resolve this: RCRA enforcement guidance specific to data center battery operators and the emergence of certified recycler capacity at scale.

One Question to Bring to Your Team

Given that BESS certification, switchgear lead times, and integrator availability can collectively add 9–18 months to project execution — and that domestic qualified cell supply covers less than 15% of demand — does our current BESS procurement pipeline reflect the real commissioning timeline, or are we planning against a cost curve that our supply chain cannot yet deliver?