Energy storage transforms into a financial asset at 100 GW worldwide
The global energy storage market crossed a historic milestone in 2025 with 100 GW installed for the first time. This new capacity no longer merely stabilizes the electrical grid. Artificial intelligence now optimizes charging and discharging according to market prices, transforming batteries into financial trading instruments. This shift toward financialization of storage redefines energy as a speculative asset, accelerating deployment while raising unprecedented questions about the speculation of an essential commodity.
124 GW installed: the infrastructure of energy trading
Global utility-scale battery capacity has grown from 10 GW in 2019 to 124 GW in 2024, representing a 12-fold increase in four years according to the International Energy Agency. Annual investments in energy storage now exceed 60-70 billion dollars, with installed global capacity surpassing 200 GWh and annual additions growing 25-30% per year over the past five years.
Lithium-ion battery prices reached $115/kWh in 2024, then $108/kWh in 2025 according to BloombergNEF. This 58% decline since 2019 — from $511 to $213/kWh — transforms the economic equation for storage.
The United States leads this expansion with more than 50 GW installed representing 140 GWh of cumulative capacity. The American administration projects growth from 28 GW at the end of 2025 to 64.9 GW by the end of 2026. Europe follows with 20 GW installed and the ambition to approach 100 GW by 2030.
AI transforms batteries into automated traders
This massive infrastructure becomes the foundation for a revolution: algorithmic optimization. Optimization platforms integrate billions of data points, AI, machine learning, and reinforcement learning to anticipate market developments. Artificial intelligence enables storage systems to learn the best trading strategies from market data while accounting for physical constraints such as battery degradation.
The results are spectacular. British studies show that machine learning improves energy arbitrage profits by 58.5% compared to traditional linear programming methods. High-frequency trading algorithms generate 14% additional revenue compared to re-optimization every minute, highlighting that profits depend crucially on trading speed.
AI-based energy management systems adjust charge-discharge cycles in real time according to weather forecasts, electrical demand, grid conditions, and market prices. This automation transforms batteries into precise traders that no human team could match.
Virtual power plants: 5 GW of aggregated assets
Aggregation multiplies this financialization. Virtual power plants represent one of the fastest-growing segments in energy storage, with global capacity projected at 5 GW in 2026. California dominates with more than 42 GW of VPP capacity registered in March 2026, aggregating more than 95,000 batteries through partnerships with Sunrun, Tesla, and Lunar Energy.
One thousand residential batteries of 10 kWh represent 10 MWh of dispatched storage — equivalent to a small peak-power plant. The aggregator builds and manages this fleet, navigates market complexity, and shares revenues with owners. Typical European revenues are set at 100-500 euros per year for a 10 kWh battery: 150-400 euros in Germany via Sonnen Community, 100-300 euros in Italy with Enel X, with the United Kingdom trending higher thanks to active markets.
Germany illustrates this maturity: aggregators manage more than 15,000 decentralized units within a single network, transitioning from simple frequency response to “cross-market optimization” allowing residential batteries to trade in real time on the wholesale market.
From public service to speculative asset
This evolution reveals a fundamental transformation. Electricity becomes a commodity traded on traditional financial rails. New assets such as battery storage systems have the opportunity to hedge the volatility of electricity markets, but bring additional financial risks to manage.
In the wholesale market, optimizers ensure revenue diversification and real-time value capture with continuous intraday trading. By participating in ancillary services, they help network operators balance the system on day-ahead and intraday markets.
This financialization follows the model observed in agricultural commodities. The influx of institutional investors led to significant changes in futures markets along two dimensions: gross positions grew dramatically from 2004 to 2006, and CFTC data shows that open interest in many commodities increased dramatically since 2004. Energy commodities now exhibit characteristics resembling financial products.
The risks of energy speculation
This transformation raises legitimate concerns. Energy prices have experienced excessive volatility, speculative bubbles, and price co-movements. The consequence of such changes is that traditional forecasts of energy prices based on fundamentals tend to fail.
The California example illustrates these tensions. CAISO’s implementation of FERC Order 2222 represents a textbook case of malicious compliance: by maintaining the 24/7 settlement obligation, requiring measurement at the resource level ($1,000-$3,000 per residential site), and limiting aggregations to single sub-LAP units, they created a program that looks good on paper but is economically impossible for most battery operators.
Energy demand increasingly becomes a political issue, with consumers blaming data centers for higher electricity bills. This tension reveals the central challenge: how to ensure that financialization of storage benefits consumers rather than speculators alone?
The emergence of batteries as a financial asset marks a turning point in energy history. VPPs offer a short-term, low-cost solution for network operators to manage the grid and make electricity more affordable for Americans. But this financialization of stored energy fundamentally questions our relationship with electricity: essential commodity or speculative merchandise? The answer will determine whether this technological revolution serves the energy transition or the enrichment of algorithms.