MethodologyContact usSupportLogin
Recent strength in lithium, copper and aluminium prices is refocusing attention on the cost structure of battery energy storage systems (BESS) just as global demand is accelerating.
These materials are central to BESS deployment: lithium for battery cells, and copper and aluminium for enclosures, cabling, inverters and power conversion systems, making projects increasingly exposed to commodity volatility and supply-chain constraints.
Fastmarkets discussed the rapid change in metals prices and the impacts on BESS and AI data centers with Paul Charles, co-chair of the ESS data center committee for NAATBatt International (The Trade Association for Advanced Battery Technology in North America) and CEO of global advisory group for giga factories and data centers Charles & Associates.
A key driver of demand in BESS is the rapid expansion of AI data centers (AIDCs). Compared with conventional facilities, AI data centers operate at far higher power densities and have limited tolerance for disruption, driving greater reliance on BESS for peak shaving, power quality management, short-duration backup and grid resilience.
In the US alone, AIDC-driven power demand could translate into around 160 GWh of storage requirements, according to Fastmarkets senior analyst Walter Zhang.
But prices for lithium, copper and aluminium have all jumped up over the last few weeks, and influenced project economics, procurement strategies and deployment timelines.
Fastmarkets’ daily assessment of lithium carbonate 99.5% Li₂CO₃ min, battery grade, spot prices cif China, Japan & Korea was $17.50-20.50 per kg on Tuesday January 20, up by around 65% compared with $10.50-12.50 per kg around a month earlier on December 17 last year.
Fastmarkets assessed the weekly copper EQ cathode premium, cif Southeast Asia at $70-80 per tonne on January 13, up by around 36% compared with $50-60 per tonne a month earlier on December 16, 2025.
And Fastmarkets’ daily assessment of the aluminium P1020A premium, ddp Midwest US was 98.00-100 cents per lb on Tuesday January 19, up by around 9% from a month earlier, and the highest level reached since 2003, when Fastmarkets started to assess the price.
While rising raw material costs do not automatically weaken demand — particularly where reliability and uptime are critical — sustained price pressure could reshape deployment priorities, favoring projects with contracted revenues, renewable co-location and optimized storage utilization.
BESS growth is also underpinned by the broader expansion of solar and wind power, where storage is essential for integrating variable generation and maintaining grid stability. Annual BESS installations were forecast by Fastmarkets to rise to around 2,448 GWh by 2035 from roughly 169 GWh in 2024, led by China and the US.
Charles: They’re sensitive, but not equally across the board. Lithium has the biggest impact because battery cells account for roughly 60% of total system cost, and lithium pricing flows directly through that line. Sharp, fast lithium price moves can put projects on pause, particularly for developers buying on spot markets.
Copper and aluminium matter as well, but they sit more in the balance-of-system costs — cabling, busbars, racking and inverters. They move the overall number, but they usually don’t determine whether a project goes ahead.
Duration is what really changes the outcome. Four hours is now the baseline for most utility-scale and data-center specifications, but we’re seeing growing demand for six, eight and even 12-hour systems in some markets. That puts more pressure on lithium and copper budgets, especially for developers without volume supply contracts. I’m currently helping scope a global survey on four- to 24-hour battery durations, and it’s clear the market is shifting longer.
A lot also depends on how the project was structured. If you’re locked into fixed EPC pricing and battery costs jump 20%, there’s no cushion — that’s when projects get delayed or dropped. Developments with flexible terms or pass-through pricing tend to keep moving.
In the US, tax credit eligibility adds another layer of risk. If a system fails to qualify for the ITC or PTC due to sourcing issues — especially under FEOC rules or domestic content requirements — it can lose 30–40% of its value stack. That makes pricing volatility even harder to manage, and it’s pushing some developers to rewrite supply terms or swap equipment just to stay compliant.
Charles: Yes — and not primarily for sustainability reasons. It’s about uptime. AI facilities aren’t like traditional cloud data centers. The loads are much denser, the draw is less predictable, and tolerance for power disruption is effectively zero. AI racks already consume around 10 times the power of CPU-based racks, and that figure is still rising. That fundamentally changes how power availability and stability are designed.
Generators are still part of the stack, but batteries are essential for covering the first seconds and minutes of any disruption. They’ve become core infrastructure. The swing from idle to full load can be enormous, and you need fast, stable power to handle it.
We’ve entered the era of Grid Law — where the pace of innovation is determined not by transistor density, as in Moore’s Law, but by gigawatt-scale power delivery. In many cases, batteries are also what allow projects to move forward on schedule. If the grid can’t deliver capacity in time, developers will install 20 MW or 50 MW of battery capacity to cover peak loads or bridge delays. For large operators, BESS is no longer optional — it’s on the critical path, not the wish list.
Charles: Yes, although interest doesn’t mean immediate substitution. Sodium-ion is attracting attention because it eliminates lithium, cobalt and nickel — the most problematic materials from both cost and supply-chain perspectives. It’s not ready to replace LFP across the board, but for two- to six-hour systems where energy density is less critical, it’s likely to start appearing.
Flow batteries and iron-air systems are also back in the conversation, mainly for long-duration applications. There’s growing interest in covering the eight- to 24-hour use case with technologies that are cheaper or more stable than lithium. We’re increasingly seeing developers think in terms of mixed portfolios: lithium for fast response and shorter durations, and alternative chemistries for deeper storage.
That said, bankability is the real hurdle. If a technology can’t be financed or insured with confidence, its theoretical advantages don’t matter. Sodium-ion and other alternatives still need to prove performance at scale. Until then, LFP will continue to carry most of the load.
Policy is also starting to influence these decisions. Projects using sodium, flow or other FEOC-compliant chemistries may qualify more easily for full tax credits in the US. Between PFE adders and domestic content rules, tax considerations are increasingly shaping procurement choices. For many developers, the tax tail is beginning to wag the procurement dog.
Charles: Without question. Large developers can absorb cost swings, lock in long-term supply and manage delays. Smaller players generally can’t. If battery prices move 20% after bid submission and you don’t have volume contracts or hedges in place, margins disappear.
Vertical integration makes a significant difference as well. Developers that control EPC, design and sourcing internally are in a much stronger position. The market is sorting by scale: projects are still getting built, but it’s the teams with procurement leverage and balance-sheet strength that can commit with confidence.
Charles: The cost of building more infrastructure is real — but the cost of being offline is far higher. AI workloads simply can’t tolerate downtime. If a model crashes mid-run, you’re not just losing power; you’re losing compute time, data processing and potentially weeks of work. A single power event can cost millions. That’s why BESS and backup systems are seen as essential risk mitigation. The question isn’t “Can we afford this?” It’s “Can we afford the failure if we don’t?” And the answer is no.
These AI campuses already involve massive capital expenditure. GPUs, interconnects and cooling systems often dominate the budget. Batteries aren’t competing with that spend — they’re protecting it. When hundreds of millions of dollars are tied up in silicon, cutting corners on power infrastructure doesn’t make sense.
Grid constraints reinforce this logic. Many sites can’t secure enough grid capacity quickly, so they’re turning to microgrids. Around a quarter to a third of new AIDCs are using behind-the-meter systems — batteries, generators, solar — to move ahead. In those cases, BESS isn’t just backup; it’s what enables the project to be built and delivered on time.
Rising lithium, copper and aluminium prices are clearly tightening margins and exposing weak points in BESS project structures. But for AI-led demand, cost inflation alone is unlikely to derail deployment.
For AIDC operators, power reliability is non-negotiable, and BESS is increasingly embedded in core infrastructure planning. While higher material costs may accelerate interest in alternative chemistries, longer-duration storage and more sophisticated system designs, lithium-ion — particularly LFP — is set to remain dominant in the near term.
“The bigger shift may be structural rather than technological. Scale, procurement power and policy alignment are becoming decisive advantages, favoring larger, vertically integrated developers,” Fastmarkets analyst Zhang said. “As AI infrastructure continues to expand globally, BESS is less a discretionary add-on and more a prerequisite — reinforcing demand for both storage systems and the raw materials that underpin them, even as the cost curve tightens.”
Interested in how increasing use of AI is driving up metals prices? Learn more about energy storage systems and gain crucial market insights and forecasting.