From EV Slowdown to AI Power Play: Korea's Battery Giants Rewrite Their Growth Story with ESS
The electric vehicle market did not collapse in 2024 and 2025 — but it slowed sharply enough to expose the risks of a single-market strategy. For LG Energy Solution and Samsung SDI, two of the world's most capable battery manufacturers, the deceleration in EV demand created a pressure that demanded a strategic response, not just a quarterly adjustment. The response they chose was not defensive. It was a deliberate pivot toward one of the fastest-growing and highest-margin segments in the entire energy storage market: battery systems for AI data centers. The timing, as it turns out, could not have been better calibrated.
![]() |
| The new power behind AI — inside the energy storage systems keeping the world's most demanding data centers online. |
Why AI Data Centers Have Become the Most Demanding Energy Customers on Earth
Training and running large AI models requires continuous, uninterrupted power at a scale that strains conventional grid infrastructure. A hyperscale data center running thousands of GPU accelerators draws power in the range of hundreds of megawatts — comparable to a mid-sized city. More critically, the power must be delivered without interruption. Even a brief outage during an AI training run can corrupt hours or days of computation, costing millions in wasted compute time and delayed product timelines.
This is where Energy Storage Systems become essential infrastructure rather than optional redundancy. ESS units installed at data center facilities serve multiple functions simultaneously: they provide instantaneous backup power in the event of grid instability, they buffer peak demand spikes that would otherwise require expensive grid upgrades, and they enable facilities to participate in energy arbitrage — storing cheap off-peak power and deploying it during high-cost demand periods. For hyperscalers spending billions annually on electricity, the financial case for advanced ESS is straightforward. The operational case is non-negotiable.
LGES JF2 DC Link 5.0: Engineering for the AI Infrastructure Standard
LG Energy Solution's flagship data center ESS product for 2026 is the JF2 DC Link 5.0 — a system engineered specifically for the power density, thermal management, and reliability requirements of AI infrastructure environments. The DC Link architecture delivers direct current output that integrates more efficiently with the internal power distribution systems of modern data centers than conventional AC-coupled alternatives, reducing conversion losses and simplifying the electrical architecture of large installations.
The 5.0 generation brings significant improvements in energy density per rack unit, allowing data center operators to store more usable energy within the same physical footprint — a critical consideration in facilities where floor space is priced at a premium. Thermal management has been redesigned around the heat profiles generated by high-density GPU clusters, which operate at power densities considerably higher than conventional server hardware. LGES has also emphasized cycle life and reliability metrics that align with data center SLA requirements: operators running AI workloads need ESS systems that deliver consistent performance across tens of thousands of charge-discharge cycles without capacity degradation that would compromise their power backup guarantees.
The commercial traction is real. LGES has secured supply agreements with data center developers and hyperscalers across North America, Europe, and Asia-Pacific, with total ESS order backlog figures that have become a meaningful contributor to the company's revenue visibility in a period when EV demand forecasting has become considerably less predictable.
![]() |
| Engineered for the demands of always-on AI infrastructure — precision energy storage at industrial scale. |
Samsung SDI's SBB Platform: The Modular Approach to Data Center Scale
Samsung SDI has approached the data center ESS market with its Samsung Battery Box platform, known as SBB — a modular architecture that allows installations to scale from single-cabinet configurations to multi-megawatt deployments using standardized building blocks. The modularity is intentional and commercially significant. Data center operators building new facilities often do not know their final power requirements at the time of initial construction. A modular ESS platform allows them to deploy what they need today and expand capacity incrementally as their AI workload requirements grow.
The SBB platform incorporates Samsung SDI's prismatic cell technology, which offers advantages in energy density and long-term cycle stability compared to cylindrical cell configurations typically used in EV applications. The thermal management system uses liquid cooling to maintain cell temperatures within the narrow operating range that maximizes both performance and longevity — a design priority driven directly by data center customers whose operational continuity requirements leave no tolerance for battery degradation-driven capacity surprises.
Samsung SDI is also leveraging its existing relationships with Samsung Electronics and Samsung C&T — the construction arm that builds data center facilities globally — to create integrated project delivery capabilities that competitors cannot easily replicate. A customer working with Samsung on a new AI data center facility can engage a single vendor ecosystem for building design, construction, and ESS installation, with system integration handled internally rather than across multiple contractors. This reduces both procurement complexity and the technical risk of integration failures between components designed by different teams.
The Battery Chasm Strategy: Turning a Market Transition into a Structural Advantage
Within Korea's battery industry, the period of EV demand deceleration has been described internally using the term "battery chasm" — the gap between the first wave of mainstream EV adoption and the next phase of accelerated growth expected as charging infrastructure matures and vehicle prices decline. The strategic question for every battery manufacturer was how to maintain revenue growth, sustain R&D investment, and preserve production workforce capability during a period when the primary market was underperforming its earlier growth trajectory.
LGES and Samsung SDI's ESS pivot is the most coherent answer to that question in the industry. Data center ESS demand is growing at rates that more than offset the EV shortfall in absolute revenue terms, and the margin profile of B2B infrastructure contracts is significantly more attractive than automotive supply agreements. Where automotive battery supply involves thin margins, intense price competition, and customer negotiating leverage that grows with volume, data center ESS contracts are structured around performance specifications, reliability guarantees, and long-term service agreements that support premium pricing and recurring revenue streams.
The customer base is also fundamentally different. Hyperscalers and data center developers are making infrastructure investments with ten to twenty year planning horizons. They are not shopping for the lowest-cost battery cell — they are selecting a technology partner whose product roadmap, financial stability, and engineering support capability will remain aligned with their operational requirements for years after installation. This is a procurement dynamic that rewards exactly the capabilities LGES and Samsung SDI have spent decades building.
![]() |
| What looks like a facility is actually a strategy — Korea's battery giants are building the energy backbone of the AI era. |
The Global Market and Korea's Competitive Position
The data center ESS market is not without competition. Chinese battery manufacturers — CATL most prominently — are aggressively pursuing the same hyperscaler customers with competitive pricing and improving product specifications. The geopolitical dimension of this competition is significant. Many Western data center operators, particularly those with US government contracts or operating under national security-adjacent regulatory frameworks, face explicit or implicit pressure to avoid Chinese battery systems in critical infrastructure deployments. This creates a structural opportunity for Korean suppliers that is entirely separate from product performance comparisons.
LGES and Samsung SDI have invested in US and European manufacturing capacity specifically to address supply chain localization requirements. LGES operates production facilities in Michigan and Ohio, with ESS product lines integrated into those operations. Samsung SDI is expanding its US manufacturing footprint through partnerships with automotive and utility customers that include ESS production capabilities. This geographic presence is not just a logistics convenience — it is a qualification requirement for a growing portion of the data center ESS market.
The AI infrastructure buildout is expected to continue accelerating through the remainder of the decade, driven by hyperscaler capital expenditure commitments that have grown larger with each successive earnings cycle. The energy storage requirements of that buildout are vast, and the window for establishing long-term supply relationships with the facilities being designed and constructed right now is open today, not indefinitely. For Korea's battery giants, the ESS pivot is not a temporary detour around an EV slowdown — it is the opening of a second major growth chapter. As data centers become the most power-intensive infrastructure category on earth, who controls their energy storage layer controls something considerably more valuable than market share in a single product category. Which battery partner your AI data center trusts with its power continuity may matter more than any spec sheet comparison.
Continue your journey into Korean life below:
- insight / Korea Economy / korea-robotics / ktodayApr 12, 2026
- insight / Korea Economy / mediaApr 12, 2026
- insight / Korea Economy / korea-venture-capital / ktodayApr 12, 2026
.webp)
.webp)

.webp)
.webp)
.webp)
0 Comments