Chinese lithium carbonate prices have surpassed CNY 175,000 per tonne, up roughly 50% year-to-date, while cobalt trades near $56,400 per tonne as DRC export quotas bite into global supply. SQM is forecasting 25% market growth in 2026, with battery energy storage systems displacing EVs as the marginal demand driver. Together, these developments describe a structural market realignment in which AI data center infrastructure, grid-scale storage deployment, and resource sovereignty policies are converging to produce the tightest critical minerals environment since 2022.
Introduction
The lithium market spent 2023 and 2024 in a prolonged correction. From its 2022 peak, lithium carbonate lost more than 80% of its value, with spot prices in China cratering to a historic low of approximately $8,259 per tonne by mid-2025, well below the cash cost of production for high-cost lepidolite operations in Jiangxi province and several Australian hard-rock mines. That correction is now definitively over.
As of early May 2026, SMM spot data places battery-grade lithium carbonate at CNY 190,500 per tonne including VAT, with UBS raising its full-year 2026 average price forecast by 18% to CNY 200,000 per tonne and projecting spot prices could touch CNY 250,000 in the May-to-June window. The move represents a round trip of extraordinary scale: the market has absorbed a surplus of more than 300,000 tonnes of lithium carbonate equivalent accumulated between 2023 and 2024, and analysts at SC Insights now estimate 2026 demand could grow 24% while supply expands only 19%.
What is structurally different about this cycle is not the magnitude of the price recovery but its origin. The demand vector driving the 2026 rally is not, primarily, passenger electric vehicles. It is battery energy storage systems, and behind BESS lies a demand source that the lithium industry's traditional price models were not built to accommodate: the AI data center.
Simultaneously, the cobalt market is navigating its own structural disruption, as the Democratic Republic of Congo's quota system removes nearly half of the country's 2024 export volume from the market and Indonesia races to fill a gap that its current HPAL output cannot yet close. These two metals, governed by different chemistry and different supply geographies, are converging on a common problem: policy-driven supply constraints meeting a demand profile that has quietly been redesigned from below.
BESS and the AI Demand Vector: A New Architecture for Lithium Consumption
For most of the EV era, lithium demand modeling was essentially an exercise in tracking automotive sales forecasts. Battery energy storage was a secondary consideration, useful for baseline projections but rarely the swing factor. That framework is no longer adequate. SQM now estimates energy storage's share of total lithium carbonate equivalent demand will rise from approximately 23% in 2025 to roughly 31% in 2026. UBS, in its base case, projects energy storage battery demand growing 60% year-on-year in 2026, far outpacing 12% growth in EV battery demand. In the bank's upside scenario, ESS demand growing 80% year-on-year would push the global supply-demand gap to 123,000 tonnes LCE and drive average prices to CNY 250,000 per tonne.
The mechanism connecting AI infrastructure to lithium markets is more direct than it might appear. AI data centers operate at power densities that conventional facilities do not approach, and they have minimal tolerance for supply interruption. This drives aggressive deployment of on-site battery storage for peak shaving, power quality management, and short-duration backup. Fastmarkets senior analyst Walter Zhang estimates that AIDC-driven power demand in the United States alone could translate into approximately 160 GWh of storage requirements. Data center operators including Microsoft, Google, and Amazon have announced over 100 GWh of new capacity with integrated battery backup requirements through 2026, with individual facilities deploying between 5 and 50 megawatt-hours of battery backup and consuming between 600 and 9,000 tonnes of lithium carbonate equivalent depending on specifications.
At the system level, the numbers are substantial. Global BESS shipments reached 421.2 GWh in 2025, a 75.5% increase year-on-year, with 600 GWh projected for 2026. Benchmark Mineral Intelligence estimates new operational BESS capacity will exceed 450 GWh in 2026, against 315 GWh in 2025. Critically, as Benchmark's head of research Iola Hughes has noted, this demand stream grows on its own logic: it is tied to grid investment policy and AI infrastructure capital expenditure cycles, not to consumer automotive purchasing behavior. This independence from EV sales cycles is precisely what SQM's CEO Ricardo Ramos identified as structurally significant when he described demand fundamentals as strong not only for electric vehicles but also for energy storage systems.
For SQM specifically, the strategic implications are concrete. The company produced 233,000 tonnes LCE in 2025 under its new Nova Andino Litio joint venture with Codelco, with a record Q4 of 66,000 tonnes LCE. It has contracted approximately 80% of its 2026 volumes and raised production guidance to 260,000 tonnes LCE for the year, leaving roughly 52,000 tonnes exposed to spot prices. That deliberate structure reflects a calculated bet that the market tightening SQM's CEO forecast in late 2025 is materializing. SQM's VP of Strategy Pablo Hernandez places 2026 BESS demand growth at 40-50% year-on-year, consistent with or above most sell-side estimates. Bank of America, which has raised its price target for SQM to $53 per share and lifted EBITDA estimates by 41% to $3.6 billion, appears to share that confidence.
Supply Under Siege: Zimbabwe, Jiangxi, and the Structural Supply Deficit
The demand story would be consequential on its own. Combined with simultaneous supply disruptions across three distinct geographies, it describes a market entering deficit territory with significant conviction. As I examined in detail in my April 2026 analysis of Zimbabwe's spodumene ban, Harare's February 25 suspension of all raw mineral exports removed an estimated 46,000 metric tonnes of concentrate from the 2026 market and exposed structural vulnerabilities in China's vertically integrated lithium processing model. Zimbabwe is expected to produce approximately 124,000 metric tonnes LCE in 2026, roughly 7% of global supply, and supplies around 15% of China's spodumene imports. That pipeline is now disrupted, with at least one cathode manufacturer reported to have turned to the spot market for feedstock.
The Jiangxi disruption adds a further layer of domestic supply pressure. CATL's Jianxiawo lepidolite mine, China's largest such operation with annual capacity of 100,000 tonnes LCE, had its mining permit expire in August 2025 and remains halted pending extension approval. The mine represents roughly one-third of Jiangxi's lepidolite capacity and approximately 10% of total Chinese domestic lithium mining. Beyond Jianxiawo, eight lepidolite mines in Jiangxi responsible for 160,000 tonnes LCE of production have been asked to submit resource reserve verification reports by September 2025 following discrepancies between licensed mining rights and actual extracted minerals. In aggregate, SC Insights estimates that the regulatory actions in Jiangxi and the Zimbabwe ban together affect approximately 17% of global lithium supply in 2026.
The spodumene market has already priced in these constraints. As noted in my earlier reporting on the Zimbabwe situation, prices have climbed above $2,000 per metric tonne for the first time since late 2023, a threshold that is beginning to make previously shuttered Australian hard-rock operations economically viable again. Fastmarkets analyst Lusty described the ban as adding fuel to the bull case, with elevated spodumene prices likely to accelerate the restart of mothballed Australian mines. That supply response, if it materializes at scale, sits at the core of the bear scenario for 2026 prices: UBS's downside case, which assumes faster-than-expected resumption at Jianxiawo and rapid ramp-up of Zijin Mining projects, implies a 198,000-tonne LCE surplus and an average price of CNY 130,000 per tonne. Morgan Stanley, by contrast, forecasts a deficit of around 80,000 tonnes LCE, while UBS's base case sits at approximately 65,000 tonnes. The range is wide, and the outcome will be determined primarily by how quickly new supply enters the market, not by demand.
Cobalt's Parallel Crisis: DRC Quotas, Indonesian MHP, and the Cathode Chemistry Reckoning
The cobalt market is running a structurally similar script to lithium, with one important difference: the supply disruption in cobalt is governed by a formal government quota system that has already demonstrated its capacity to reduce actual export flows far below allocated theoretical volumes. The DRC produced 74.5% of global cobalt supply in 2024 according to S&P Global Market Intelligence. When Kinshasa replaced its export ban with a quota system on September 21, 2025, the 96,600-tonne annual cap it established for 2026 represented 48.2% of the country's 2024 production. As I outlined in my May 2026 analysis of the broader sovereign resource turn, this is resource sovereignty as an operational market force, not a policy aspiration.
The gap between allocated and actual export flows has been particularly severe. Fastmarkets reporting indicates the DRC may have exported less than half of even the quota volumes allocated for Q4 2025, extended into Q1 2026, due to regulatory complexity around royalty prepayment and compliance certification. No material left the country via the quota system through late December 2025; Glencore became the first company to export under the new framework, followed shortly by CMOC subsidiary Tenke Fungurume Mining. Cobalt prices responded accordingly, with the Platts-assessed cobalt hydroxide CIF China price surging 69.9% to $54,674 per metric tonne by December 9, 2025, from $32,187 at the time of the quota announcement. Current prices near $56,414 per metric tonne reflect a market that S&P Global has warned could see Chinese cobalt stocks hit dangerously low levels.
Indonesia is the most credible partial offset. Domestic cobalt production reached 38,324 tonnes in 2025 and is forecast to climb 39.1% to 53,318 tonnes in 2026, primarily as a byproduct of HPAL nickel operations through facilities including Zhejiang Huayou's Huafei Cobalt-Nickel Project and Ningbo Lygend's PT Halmahera Persada Lygend. Fastmarkets analysts estimate Indonesian cobalt-in-MHP production could reach approximately 67,500 tonnes in 2026. However, analysts are consistent on one point: Indonesian MHP cannot fully substitute DRC hydroxide in chloride processing routes because of nickel contamination thresholds, meaning the offset trims rather than erases the supply gap. Fastmarkets projects a structural shortfall of approximately 10,700 metric tonnes against 2026 demand of near 292,300 metric tonnes.
The policy response from Washington has been direct. The Pentagon's Defense Logistics Agency issued a five-year tender for up to 7,500 tonnes of alloy-grade cobalt, its first cobalt purchase since 1990, seeking to commit between $2 million and $500 million to rebuild the national strategic stockpile. Separately, Project Vault, a $12 billion critical mineral stockpile initiative, has been announced alongside a memorandum of understanding between government-backed Orion CMC and Glencore covering a 40% stake in two DRC copper and cobalt mines. The strategic logic is explicit: every other major cobalt producer in the DRC is Chinese-controlled, with the exception of Eurasian Resources Group.
The cathode chemistry dimension of the cobalt story is where the longer-term structural response plays out. LFP adoption, which eliminates cobalt entirely, has been accelerating: LFP was the fastest-growing battery chemistry in 2025 with demand rising 48% year-on-year, driven precisely by BESS deployment and Chinese EV expansion. As cobalt prices remain elevated, the economic incentive to move toward cobalt-free cathode architectures intensifies. CMOC's own CEO has warned that prices at current levels risk demand destruction and substitution, a dynamic that has been a long-running trend but which the DRC quota system has materially accelerated.
The AMG Signal: Industrial Capital Reads the Cycle
AMG Advanced Metallurgical Group's Q1 2026 results, released May 7, offer a useful real-world calibration of how industrial capital is reading the current cycle. The Netherlands-based critical materials producer reported revenue of $446.1 million, up 15% year-on-year, with its lithium operations as the standout performer: lithium segment revenue surged 89% year-on-year to $60.6 million as the company's Bitterfeld, Germany battery-grade lithium hydroxide refinery began commercial operations. Spodumene concentrate prices were up 149% year-on-year in Q1, and lithium hydroxide prices nearly doubled. Net income attributable to shareholders reached $12 million, more than double the $5 million in the prior year, aided by a write-up of lithium inventories.
The forward signals from AMG are arguably more informative than the reported numbers. CEO Heinz Schimmelbusch characterized the geopolitical environment as offering significant opportunities to continue growing the company's critical materials footprint, a framing consistent with a management team that sees supply-side disruption as structurally favorable to processors with secured feedstock. AMG raised $127 million in a 10% capital increase in April 2026, specifically to finance expansions into lithium carbonate, high-purity molybdenum, and vanadium. The Bitterfeld site is also commencing engineering on a 5,000-tonne lithium carbonate to lithium hydroxide conversion plant designed to accept recycled lithium carbonate, a deliberate positioning for the circular supply stream that will grow materially as early-generation EV batteries reach end-of-life.
The broader market data that AMG's results illuminate is mixed in ways that matter for analysis. Global EV sales reached 21.6 million in 2025 and are forecast at 22.7 million in 2026, with year-on-year growth slowing from 31% in 2025 to approximately 15% in 2026. AMG's Q1 data noted a roughly 3% YoY dip in global EV sales for the period. Yet lithium revenues surged 89%. The disconnect makes the structural point that SQM and UBS are both making: the price narrative for lithium in 2026 is no longer being written by automotive demand. It is being written by grid operators, data center developers, and the policy frameworks governing sovereign resource access.
Sodium-Ion and LFP: The Chemistry Variables That Could Shift the Calculus
No analysis of the current lithium and cobalt market environment is complete without accounting for the chemistry transitions that represent the primary downside risk to the bull case. As I detailed in my May 2026 coverage of CATL's Super Technology Day, the company committed to GWh-scale industrialization of its Naxtra sodium-ion platform and followed that announcement six days later with a 60 GWh sodium-ion supply agreement with Beijing HyperStrong Technology. Sodium-ion batteries eliminate lithium, cobalt, and nickel entirely, the three most strategically problematic materials in the current supply chain.
For the BESS segment specifically, sodium-ion is a credible substitution candidate in the two-to-six-hour storage window where energy density is a secondary consideration to cost and cycle life. If manufacturers accelerate the sodium-ion transition in stationary storage applications faster than current projections assume, the lithium demand uplift from BESS that SQM, UBS, and Benchmark are all pricing into their 2026 outlooks could be partially offset. The timeline for meaningful displacement is not 2026, but it is not 2035 either.
LFP's trajectory, however, cuts in the opposite direction. LFP demand rose 48% in 2025, driven by BESS deployment and Chinese EV growth, and the chemistry is directly favorable to lithium carbonate producers like SQM whose core product is carbonate rather than hydroxide. LFP growth simultaneously reduces cobalt intensity in the overall battery market, compounding the demand destruction dynamic that persistently elevated cobalt prices are already encouraging. The interplay between LFP expansion, sodium-ion emergence, and the relative cost curves of DRC-sourced hydroxide versus Indonesian MHP represents the principal chemistry uncertainty layered on top of the supply and demand fundamentals.
Recycled cobalt supply adds a further variable. Battery recycling is projected to contribute a growing share of cobalt feedstock through the late 2020s, with hydrometallurgical processes capable of recovering over 95% of cobalt in spent NMC batteries. Recycled cobalt salts already trade at 15-20% premiums when certified as low-carbon, and the EU Battery Regulation mandates at least 95% cobalt recovery from recycled batteries by 2031. Current recycled supply represents approximately 17% of feedstock and is projected to grow 43% in 2026. That growth rate is meaningful but insufficient to materially offset the DRC structural gap in the near term.
Conclusion: A Market Reorientation, Not a Super-Cycle
The lithium market's move from CNY 55,000 per tonne in mid-2025 to SMM spot quotes of CNY 190,500 per tonne in early May 2026 is striking in velocity. But the more analytically significant development is structural: the demand architecture that is driving this recovery is categorically different from the one that drove the 2022 peak. In 2022, lithium prices were carried almost entirely by EV sales growth and the downstream inventory building that accompanied it. The correction that followed was brutal precisely because that single demand vector proved insufficient to absorb the supply that had been commissioned at peak prices.
In 2026, the recovery is supported by at least three demand streams operating on different cycles: EV adoption, which continues to grow but is decelerating toward 15% year-on-year; BESS deployment, which Benchmark projects at over 450 GWh of new operational capacity this year, growing independently of automotive cycles; and AI data center infrastructure, which is injecting a novel and high-intensity demand signal into the storage system in ways that standard lithium demand models did not anticipate. SQM's identification of BESS as the marginal demand driver, UBS's 60% year-on-year growth estimate for energy storage batteries, and the AI data center storage estimates from Fastmarkets all point in the same direction.
The supply side is simultaneously experiencing a structural reorganization driven not by geology or project economics but by sovereign policy. Zimbabwe's export ban, Jiangxi's mining permit enforcement, the DRC's cobalt quota system: these are not temporary disruptions waiting for regulatory resolution. They represent a sustained assertion of resource sovereignty that is unlikely to reverse even as prices rise. As Carlos Diaz of SQM told the World Lithium Conference in Santiago, the $7 to $8 per kilogram nightmare is over. The new floor is somewhere between $15 and $18 per kilogram, and the structural forces now in play suggest that floor is defensible.
The risk to that floor is real but is more likely to come from chemistry evolution (sodium-ion scaling faster than expected in stationary storage) or macroeconomic disruption to data center capital expenditure than from a return of the surplus conditions that defined 2023 and 2024. Morgan Stanley's deficit estimate of 80,000 tonnes LCE and UBS's base case of 65,000 tonnes LCE bracket a market that is tighter than it has been in three years, supported by demand vectors that are structurally independent of each other. For critical minerals professionals modeling supply chains into 2027, the working assumption should be that both lithium and cobalt remain in structurally constrained supply environments, that BESS and AI infrastructure represent durable incremental demand, and that sovereign resource policy has become a first-order input into commodity price formation rather than a secondary risk factor.