EVs, AI Storage and Robotics: Mapping Lithium’s Three Demand Waves

Three categories of lithium consumption that were essentially nonexistent fifteen years ago are now scaling simultaneously. Electric vehicles built the foundation, but stationary battery storage for AI data centres grew approximately 71% in 2025, outpacing EV battery demand growth, while robotics is transitioning from isolated pilots toward the kind of S-curve adoption that reshaped the automotive industry’s relationship with lithium over the past decade. On the supply side, Chinese refining capacity is being built at five to ten times current consumption levels, yet feedstock from upstream mines remains the binding constraint. What follows maps the three demand waves shaping lithium’s next decade, examines why they compound rather than substitute, identifies where the real supply bottleneck sits, and draws out the investment and geographic implications for project developers operating in a fragmented trade environment.

EVs set the floor: why the original thesis still holds

Electric vehicles remain the single largest end-use category for lithium and the structural anchor of all serious demand forecasts through at least 2035. Global lithium demand reached approximately 1.54-1.6 million tonnes LCE in 2025, with major producer projections suggesting that figure could rise toward approximately 3.7 million tonnes by 2030, effectively doubling from mid-2020s levels.

The EV thesis is not a peak-and-plateau story. It is a sustained high-volume floor onto which new demand categories are stacking.

Three indicators support the durability of that floor:

• Growth rate: EV adoption continues at double-digit rates globally, with Chinese market expansion notably more robust than North American or European equivalents.
• Geographic breadth: Adoption is broadening across regions, not concentrating in a single market.
• Share of total demand: EVs plus battery storage together are expected to account for the overwhelming majority of lithium consumption by 2035-2040.

Investors who discounted the EV thesis after near-term price volatility risk misreading the structural base. The floor is durable. The question is what builds on top of it.

Stationary storage and AI data centres: the fastest-growing demand segment

The second wave is no longer a forecast. It is visible in 2025 demand data.

Stationary battery storage demand grew approximately 71% in 2025, outpacing EV battery expansion and establishing energy storage as the fastest-growing segment of lithium consumption. Hyperscale AI data centres, which require extremely high, stable power loads, are increasingly paired with large-scale battery storage as a reliability and grid-balancing tool. J.P. Morgan projects energy storage at approximately 30% of global lithium demand in 2026, rising to approximately 36% by 2030.

The AI-specific trajectory sharpens the picture further. According to Benchmark Mineral Intelligence, lithium demand tied specifically to AI-linked data centre storage could follow a steep curve over the next decade:

Benchmark Mineral Intelligence estimates AI-linked data centre storage demand rising from approximately 15,000 tonnes in 2025 to nearly 70,000 tonnes by 2035.

Data-centre-specific storage capacity is projected to grow at approximately 35% CAGR through 2030, with AI as the primary driver. Chinese industry participants are already actively discussing AI storage as a demand driver, per executive commentary from project developers in the sector.

AI-driven commodity demand is not a lithium-exclusive story: copper faces a structurally similar dynamic, where data centre construction and grid infrastructure build-out are creating a demand layer that sits on top of the existing industrial and energy transition base, compressing the available window for new supply to respond.

Analysts and investors still anchoring demand models exclusively to EV penetration rates are working with an incomplete picture. The timescale for this second wave is years, not decades.

What is lithium’s role in the battery economy, and why does it remain irreplaceable

The growth headlines across EVs, storage, and robotics share a common dependency. Each relies on lithium-ion battery chemistry, and the properties that make lithium the preferred choice are not easily replicated.

Three characteristics underpin lithium’s position across all high-performance battery applications:

1. Energy density: Lithium offers the highest energy-to-weight ratio of commercially viable battery chemistries, a property that is critical for EVs (range), robotics (operational endurance), and data centre storage (space efficiency).
2. Cycle life: Lithium-ion cells maintain performance across thousands of charge-discharge cycles, making them suitable for applications where reliability over extended periods is non-negotiable.
3. Established manufacturing infrastructure: China controls approximately 85% of current global lithium chemical conversion and battery manufacturing capacity, meaning the entire production chain, from cathode materials to cell assembly, is built around lithium at industrial scale.

Each humanoid robot, autonomous system, and EV carries its own battery pack. Aggregate demand therefore scales with unit volume rather than usage intensity. When any of these categories follows an S-curve adoption pattern, the demand impact on lithium becomes non-linear.

Where substitution risk is real and where it is not

Alternative chemistries are advancing. Sodium-ion batteries are competitive in short-range, cost-sensitive applications where energy density is less critical. Solid-state technology holds longer-term potential. Neither, however, is expected to materially displace lithium-ion in high-performance, high-energy-density applications, including long-range EVs, data centre storage, and robotics packs, within the 2025-2035 window.

Sodium-ion energy density comparisons with lithium-ion put the gap in concrete terms: lithium-ion cells currently achieve 120-270 Wh/kg versus 100-160 Wh/kg for sodium-ion equivalents, a performance differential that analysts expect to persist across high-demand applications, including long-range EVs, data centre storage, and robotics packs, through at least 2035.

Refining capacity is not the bottleneck: where the real supply constraint sits

A common reading of Chinese capacity expansion leads to a bearish conclusion: if refining capacity is being built at five to ten times current consumption levels, oversupply should follow. That reading mistakes the nature of the constraint.

Chinese refining and conversion capacity is indeed expanding aggressively. China controls approximately 85% of global lithium chemical conversion and battery manufacturing capacity, and that share is growing. European refining and conversion infrastructure investment is advancing faster than North American equivalents, though Tesla is cited as one of the few North American companies to have developed meaningful domestic refining capacity.

The bottleneck sits upstream.

Management estimates from project developers illustrate the scale of the mismatch: multiple Bandeira-scale operations would be needed to supply the capacity expansion plans of a single major customer such as Yawah. Processing overcapacity without sufficient ore leads to underutilised refining assets and tight feedstock markets.

Upstream supply disruptions in copper during 2025 offer a direct historical parallel for what analysts project in lithium feedstock markets: when mine development timelines compress the available new-production window, even modest demand acceleration relative to consensus can produce feedstock shortages that refining overcapacity cannot resolve.

Why permitting timelines make the gap structural, not cyclical

Mine development timelines of seven to twelve years mean upstream supply cannot respond quickly to demand signals. Permitting challenges, resource concentration, and the sheer time required to bring new hard-rock or brine operations into production make any near-term feedstock shortage durable rather than transient. The supply gap is structural.

How robotics adoption could reshape long-run lithium consumption

The third wave is earlier in its trajectory, but the pattern is familiar. Robotics demand for lithium is structurally analogous to where EV demand sat fifteen years ago: confined to isolated pilots, approaching an inflection point where per-unit battery consumption becomes aggregate-material.

The EV precedent is instructive. The shift from niche technology to mass adoption changed the demand calculus entirely. Robotics, including humanoid and service robots, industrial automation, and autonomous systems, appears positioned for the same transition.

Analyst projections already incorporate robotics into baseline demand models. Emerging technology applications, spanning AI backup systems, humanoid robots, industrial robotics, aviation, and defence batteries, are projected to add:

• Approximately 105,000 tonnes LCE per year by 2035
• Approximately 303,000 tonnes by 2040
• Approximately 720,000 tonnes by 2050 under a baseline scenario

In an optimistic scenario where several lithium-intensive technologies scale concurrently, additional demand from these uses could reach approximately 2.8 million tonnes LCE annually by 2050.

Lithium Ionic’s CEO has explicitly identified robotics as the next anticipated wave of battery demand following AI storage infrastructure build-out, framing the progression as EVs first, AI storage second, and robotics third.

Projects and companies entering production by 2030-2035 are not only capturing EV and AI storage demand. They are entering the market as robotics demand begins its own curve, creating a longer and more resilient demand runway than any single-wave thesis implies.

Where a project is located now determines who it can sell to

The macro demand and supply dynamics narrow to a single, increasingly material variable at the project level: jurisdiction.

Geopolitical fragmentation and trade policy are creating a two-tier project landscape. The U.S. Inflation Reduction Act, the EU Critical Raw Materials Act (which entered into force in 2024), and various national strategic minerals programmes are generating explicit incentives for domestic or allied-nation supply. North American projects may face preferences or constraints directing sales toward specific regional customers or away from Chinese buyers.

Allied-nation supply agreements are evolving faster than most project financing models assume, with frameworks like the Argentina-US arrangement demonstrating how bilateral trade structures can expand the viable customer set for projects located outside traditional Western supply corridors.

The EU Critical Raw Materials Act entered into force in May 2024, identifying lithium as a strategic raw material and setting 2030 benchmarks requiring that at least 10% of annual consumption be sourced from domestic extraction, 40% from domestic processing, and 25% from recycling, benchmarks that directly shape the customer landscape for projects outside the EU seeking European offtake agreements.

Brazil offers a case study in the open-framework category. Characterised as having over a century of established trade relationships with global partners, it provides project developers with flexibility to supply customers across multiple regions without the jurisdictional restrictions affecting some North American operations.

The economics reinforce the geographic argument. Spodumene prices around $2,000 per tonne are described as significantly accretive given an approximately $600 per tonne all-in sustaining cost structure.

A floor price of $1,000 per tonne is sufficient to demonstrate positive margins to prospective lenders, with spodumene pricing stability improving lender confidence relative to the prior three to four years.

Week-to-week trade policy instability, particularly between the US and its partners, has made jurisdictional flexibility an active consideration in project financing and offtake negotiations. Project location is no longer only a geological and logistics consideration; it is a commercial and financial one that directly affects available customer sets, capital sources, and ultimately project valuation.

The structure of the next decade: compounding waves, not a single curve

The three demand waves are not sequential replacements. They are additive layers. EVs provide the durable floor. AI-linked stationary storage is the fastest-growing current segment. Robotics is the emerging horizon. Each stacks on the others.

Total lithium demand by 2030-2035 will reflect the simultaneous maturation of EVs, the rapid scaling of AI-linked battery storage, and the early emergence of robotics as a new consumption category. Ganfeng’s chairman has spoken of 30-40% demand growth in 2026 alone, with the possibility of prices doubling from current levels, though this remains unverified executive commentary.

A pattern common in resource sectors is also visible: commodity prices stabilising or rising while equities lag. Even producing lithium companies have seen share price weakness relative to improving commodity and demand fundamentals. That divergence tends to close, though timing varies.

For investors evaluating lithium exposure, four variables separate high-quality positioning from generic sector bets:

Supply chain risk as a portfolio variable has moved from a specialist concern to a mainstream allocation consideration, with rare earth and lithium exposure increasingly evaluated not just on commodity price sensitivity but on jurisdictional concentration, refiner dependency, and policy regime durability.

1. Upstream feedstock position: Does the project control its own ore supply, or is it dependent on purchased concentrate?
2. Jurisdictional trade optionality: Can it sell to customers across regions, or is it constrained by policy alignment?
3. Project economics relative to conservative price floors: Are margins positive at floor prices, not just at spot?
4. Offtake structure with floor protections: Do agreements provide downside protection that satisfies lender requirements?

The lithium demand story has moved beyond a single EV curve. Investors who frame their exposure as a single-sector bet are systematically underestimating the addressable demand base. The multi-wave thesis is not a speculative overlay; it is increasingly supported by 2025 demand data, refinery capacity planning decisions, and executive commentary from major producers.

This article is for informational purposes only and should not be considered financial advice. Investors should conduct their own research and consult with financial professionals before making investment decisions. Forward-looking demand projections are subject to change based on market developments, technological adoption rates, and policy shifts.