Battery-as-a-Service (BaaS)

• Battery-as-a-Service is a model where battery ownership is decoupled from the asset, enabling fast swapping and centralized management.
• It supports diverse applications including EV battery swapping, grid-scale cloud storage, community leasing, and robotic power logistics.
• The framework optimizes battery scheduling, revenue stacking, and degradation economics to enhance operational efficiency and asset value.

Searching arXiv for the specified papers to ground the article in current records. {"query":"id:(Mahoor et al., 2017) OR id:(Chen et al., 2023) OR id:(Holand et al., 2024) OR id:(Pocola et al., 6 Jul 2025) OR id:(Rasouli et al., 2019) OR id:(Vosooghi et al., 2019)","max_results":10} Battery-as-a-Service (BaaS) denotes a service and ownership model in which the battery is separated from the end-use asset and managed by a provider that monetizes battery availability, energy throughput, or reserved capacity rather than transferring battery ownership to the customer. Across the literature, BaaS appears in several operational realizations: battery swapping stations for electric vehicles, cloud storage platforms that sell virtualized storage backed by grid-scale batteries, capacity-rental arrangements between battery operators and energy communities, and centralized battery hubs serving robotic fleets (Mahoor et al., 2017). A common structural feature is that the battery becomes a managed service asset whose charging, availability, degradation, and allocation are optimized across time, users, and applications rather than remaining a fixed component of a single vehicle or premise (Chen et al., 2023).

1. Conceptual definition and service logic

In the battery-swapping literature, BaaS is instantiated by a station model in which the operator owns a pool of batteries, keeps them charged in advance, and provides a rapid exchange service to users. The defining operational distinction is that an electric vehicle with an empty battery can exchange it for a fully charged one and leave within a few minutes, whereas a conventional Battery Charging Station (BCS) requires the vehicle to remain plugged in for hours (Mahoor et al., 2017). The same source explicitly identifies the “company-owned battery model” as the prominent arrangement in which batteries are leased rather than purchased, thereby lowering the upfront price of the vehicle and converting battery access into a service.

The service logic generalizes beyond road vehicles. In a grid-connected setting, BaaS can take the form of third-party ownership and rental of battery capacity, with an energy community renting part of a larger battery owned by an operator that continues to participate in energy markets (Pocola et al., 6 Jul 2025). In cloud electricity storage, end-users subscribe to “virtual batteries” backed by the residual capacity of a multi-service grid-scale battery primarily used for high-priority grid services (Rasouli et al., 2019). In planetary robotics, a centralized hub rover charges replacement modules and swaps them into smaller rover agents on demand, so that mobile units consume battery service rather than own dedicated charging infrastructure (Holand et al., 2024).

These variants suggest that BaaS is best understood not as a single business format but as a family of architectures organized around three principles: centralized battery ownership, decoupling of battery energy logistics from the mobile or end-user asset, and service-oriented optimization of battery availability over time. A plausible implication is that BaaS is most naturally deployed where downtime, capital intensity, or multi-use value of storage dominate the economics of direct ownership.

2. Ownership structures and intermediary roles

The literature consistently places a BaaS provider between end-users and an upstream energy or power system. In the battery swapping station model, the station sits between the EV owner, the station owner, and the power system. The EV owner seeks minimal waiting, extended travel range, lower vehicle purchase cost if the battery is not owned, and reduced need for home high-power chargers. The station owner owns a fleet of batteries, determines how many batteries to buy initially, charges batteries according to time-varying electricity prices, maintains sufficient fully charged inventory to satisfy swap demand, and may use the battery fleet to earn additional revenue through grid services (Mahoor et al., 2017).

A comparable intermediary role appears in cloud storage. There, the grid-scale battery owner retains the battery primarily for congestion management, while a third-party Cloud Storage Operator accesses the residual capacity and resells it as virtual storage contracts to many end-users (Rasouli et al., 2019). The contractual product is explicitly parameterized as

$Q=\{(q_k,c_k,r_k)\}_{k\in K},$

where $q_k$ is the subscription fee, $c_k$ the virtual energy capacity, and $r_k$ the virtual power capacity. End-users then operate the virtual battery subject to the contract constraints.

In the energy-community setting, the business relation is a capacity rental agreement. The community pays a rental fee per kWh of battery capacity per year, while the operator receives that fee but gives up the opportunity to use the rented capacity in the day-ahead market. The contract is feasible only if the community’s rental price is below or equal to the savings it obtains from the battery and the operator’s rental price is above or equal to the foregone market profit (Pocola et al., 6 Jul 2025). This formalizes BaaS as an intermediation problem in which the provider prices flexibility while allocating battery capacity across customers and markets.

The robotic hub architecture extends the same ownership separation into an autonomous system. The hub manages charging, storage, and health or state of standardized battery modules; the rover agents carry only a battery terminal, mating geometry, navigation sensors, autonomy stack, and a removable battery module (Holand et al., 2024). This suggests that BaaS can also be interpreted as infrastructural unbundling: energy production, storage management, and battery servicing are centralized, while the serviced asset is simplified.

3. Operational architectures and asset-state models

A defining technical property of BaaS is that batteries are treated as service inventory rather than static components. In the battery swapping station scheduling model, each battery moves through four states at hour $t$: empty, charging, fully charged, and out of station. Each state is represented by a binary variable, and each battery must be in exactly one state at each time: $$(x^E + x^C + x^F)+ y =1. \ \forall t,\ \forall b \tag{1}$$ The total number of batteries owned by the station is fixed: $$\sum_b (x^E_{bt} + x^C_{bt} + x^F_{bt} + y_{bt}) = N_S \ \ \forall t \tag{2}$$ and the number charging simultaneously is limited by the number of chargers: $\sum_b x^C_{bt} \le N_C \tag{3}$ These relations encode state exclusivity, fixed asset inventory, and charging-resource limits (Mahoor et al., 2017).

The same source emphasizes that the main decision is when each battery should start charging, with scheduling constrained by charger availability, hourly swap demand, minimum charging time, minimum and maximum charging rates, and state-of-charge limits. In this formulation, BaaS is not merely an energy delivery problem; it is a sequencing and dispatch problem over battery inventory.

In cloud storage, the operational architecture is hierarchical rather than physical. High-priority grid service uses the battery first, while cloud storage is served from stochastic residual capacity: $$r_t^{b,l} \le a_t^b \le r_t^{b,h}, \qquad 0\le s_t^b\le c_t^b$$ The low-priority service can be blocked if aggregated user requests exceed physical availability, in which case external resources may be used to fulfill commitments (Rasouli et al., 2019). This introduces a priority-structured BaaS architecture in which customer service quality depends on stochastic residual access.

In planetary robotics, the operational workflow is embodied in an autonomous dock-and-swap sequence: approach and docking entry, docking lift and alignment, battery cache indexing and swap, and exit. The hub contains a battery-management system, a cache of multiple charged modules, docking ports, and swapping hardware; the rover returns to the hub when energy is needed and resumes mission tasks with a fully charged battery (Holand et al., 2024). This is a physically explicit version of the same asset-management principle: spare batteries are pooled, pre-charged, and routed through a servicing process.

4. Scheduling, valuation, and optimization frameworks

The BaaS literature spans feasibility-oriented inventory scheduling, stochastic service design, and intertemporal valuation. In the swapping-station model, the scheduling problem is deliberately framed from the station owner perspective and is feasibility-oriented: charging should be sequenced so that demand is met, chargers are not overloaded, and enough fully charged batteries remain available. The paper notes that an extended objective could minimize electricity payment or cooling cost, or maximize revenue from reserve and flexibility services, but leaves the objective open (Mahoor et al., 2017).

A more explicit life-cycle optimization is developed for battery swapping stations that also provide grid services. There, short-term scheduling of charging, discharging, and swapping is coordinated with long-term battery value through an improved intertemporal decision framework (Chen et al., 2023). The long-term objective is

$$LB^* = \max LB = \max \sum_{t \in T_{life}} \delta_t \, SB_t^* \tag{1}$$

with discount factor

$q_k$0

The framework imposes life-cycle constraints

$q_k$1

and

$q_k$2

so that total degradation remains within the usable degradation budget and operating degradation is not under-accounted relative to calendar aging.

The central economic construct in that model is the marginal degradation cost (MDC), measured in $q_k$3, defined as the opportunity cost of using the battery now rather than preserving future value. The short-term adjusted MDC is

$q_k$4

and the short-term operating objective is

$q_k$5

with

$q_k$6

This formalizes a distinctive BaaS problem: the provider must price battery wear internally while allocating throughput across transportation service, energy arbitrage, and reserve provision (Chen et al., 2023).

In the energy-community BaaS model, both community operation and operator trading are formulated as linear programs. The community control model minimizes the electricity bill: $q_k$7 subject to state-of-charge dynamics, power balance, cycle constraints, and end-of-day constraints. The operator’s market model maximizes day-ahead profit: $q_k$8 with analogous battery constraints (Pocola et al., 6 Jul 2025). The paper identifies two practical defects of daily-horizon LP control—too many charge-discharge cycles and excessive end-of-day emptying—and introduces regularizers: $q_k$9 with $c_k$0, and

$c_k$1

with $c_k$2. The best performance is reported for $c_k$3 together (Pocola et al., 6 Jul 2025).

Cloud storage introduces a two-stage stochastic optimization problem in which platform design precedes stochastic operation (Rasouli et al., 2019). The CSO maximizes

$c_k$4

and the paper proposes an effective capacity approximation to replace the hard blocking cost with a convex approximation. The blocking-related metric is linked to aggregated user actions, and the resulting formulation becomes computationally tractable. This is important because it connects BaaS economics to statistical multiplexing rather than to one-to-one asset allocation.

5. Economic mechanisms and revenue stacking

A recurring theme in BaaS research is revenue stacking: the same battery asset can support multiple applications whose relative value changes over time. In the battery-swapping valuation framework, the station simultaneously provides battery swapping to EV users and flexibility service to the power grid, including energy arbitrage and reserve (Chen et al., 2023). The model decides how battery throughput should be allocated between transportation application and energy application, with the MDC internalizing the inter-application tradeoff.

The energy-community study embeds BaaS within a stacked revenue model in which the operator earns from multiple services, including day-ahead market arbitrage and community leasing revenue; imbalance and ancillary services are identified as additional possible services (Pocola et al., 6 Jul 2025). The main tradeoff modeled is between day-ahead market revenue and community rental revenue. The operator is assumed to allocate the most profitable part of the battery to market use and the least profitable marginal capacity to community rental, so the minimum acceptable rental price is the opportunity cost of diverting that capacity.

Cloud storage similarly uses residual capacity from a battery devoted primarily to another service. The high-priority service is congestion management, while the low-priority cloud storage layer sells virtual battery contracts to end-users (Rasouli et al., 2019). This is an especially explicit example of layered BaaS: a battery can first satisfy system-level obligations and then monetize stochastic residual capacity through a secondary service market.

These revenue structures differentiate BaaS from simple battery leasing. In the literature considered here, BaaS is not just a financing device but an operational and market design framework in which battery capacity, throughput, or service priority is sold while the provider retains the option to redeploy the battery across applications. A plausible implication is that BaaS becomes more attractive when battery assets are sufficiently flexible to generate value in multiple non-coincident markets.

6. Performance characteristics, service benefits, and design constraints

The principal operational advantage of swapping-based BaaS is sharply reduced service time. In EV applications, a battery swap is described as taking only a few minutes and as being “as fast and easy as refuelling a gas station,” in contrast to hours at a BCS (Mahoor et al., 2017). In shared autonomous electric vehicle simulation, battery swapping stations are modeled with 20 simultaneous swap spaces and a 5-minute swapping process; under this configuration, in-vehicle passenger kilometers traveled increase to $c_k$5–$c_k$6 million km, fleet usage rises to about $c_k$7–$c_k$8, and total queue time becomes almost negligible, around 1–3 minutes (Vosooghi et al., 2019). The same study reports that swapping significantly improves SAEV performance indicators relative to charging-based operation and emphasizes that swapping restores fleet availability by decoupling recharge time from vehicle downtime.

In the planetary multi-agent system, the prototype achieves an average servicing time of 98 seconds in integrated indoor testing and completes 15 consecutive autonomous dock-and-swaps without human interference; the battery swap mechanism alone achieves a 100% success rate over 50 consecutive swaps (Holand et al., 2024). Docking was successful on a $c_k$9 pitch but not on $r_k$0, and swapping failed on both $r_k$1 and $r_k$2 roll conditions due to lateral slip during transfer. The optimized guide-rail geometry increased the valid docking configuration space by 258%, and yaw tolerance improved from $r_k$3 with purely mechanical compensation to $r_k$4 with autonomous navigation plus mechanical compensation (Holand et al., 2024). These results show that BaaS performance depends not only on energy economics but also on robust mechanical and control design where autonomous exchange is required.

Grid-facing BaaS architectures emphasize flexibility and load shaping. The battery swapping station is modeled as a flexible load that can shift charging to off-peak or nighttime hours, reducing peak demand, feeder congestion, and the need for grid upgrades (Mahoor et al., 2017). Cloud storage can coexist profitably with congestion management, and a slight leasing-price discount can restore the CSO profit from a stochastic residual battery to the level of a deterministic one (Rasouli et al., 2019). In the energy-community case study, a community of 200 houses with a 330 kW wind turbine can save up to 12,874 euros per year by renting 280 kWh of battery capacity after subtracting battery rental costs (Pocola et al., 6 Jul 2025).

The literature also identifies important requirements and limitations. Swapping requires consistent battery standards across vehicles, sufficient charger capacity, sufficient battery inventory, and coordinated demand forecasting (Mahoor et al., 2017). In SAEV systems, station geography matters: optimized placement that minimizes access distance while maintaining dispersion performs better than overly centralized layouts, and battery capacity must be chosen to avoid charging during demand peaks (Vosooghi et al., 2019). The planetary system remains an Earth-based proof of concept using commercial off-the-shelf parts, without dust shielding, thermal insulation, or radiation hardening (Holand et al., 2024). These findings counter a common misconception that BaaS is solely a financing mechanism; in practice it is tightly coupled to physical standardization, inventory management, scheduling, and infrastructure siting.

7. Degradation, reliability, and second-life implications

Because BaaS providers centralize battery management and increase asset utilization, degradation modeling becomes a first-order design variable rather than a background maintenance issue. The intertemporal valuation paper is explicit that frequent battery swapping accelerates degradation and shortens physical battery life, but can lengthen economic life because the additional service revenue justifies continued operation (Chen et al., 2023). In its case study, physical end-of-life is defined by 80% capacity after 2000 cycles at 100% DOD plus calendar aging of 1% capacity loss per year; with fixed O&M costs, economic end-of-life occurs in Year 6 while physical end-of-life occurs in Year 7 (Chen et al., 2023). This divergence is central to BaaS because providers optimize over economic utility of the battery asset, not merely physical survivability.

The same study further shows that battery swapping price affects not only profitability but also degradation economics. Life-cycle revenue rises as swapping price increases from about \$r_k$5160/MWh and drops sharply beyond \$r_k$630/MWh-throughput to \$r_k$780 to \$r_k$845 to \$r_k$9200/MWh (<a href="/papers/2302.14291" title="" rel="nofollow" data-turbo="false" class="assistant-link" x-data x-tooltip.raw="">Chen et al., 2023</a>). When swapping price exceeds about \$180/MWh, demand tends toward zero and EVs revert to charging or discharging without swapping. This indicates that service pricing in BaaS is also a battery-life management instrument.

Reliability in shared-storage BaaS is often treated probabilistically. Cloud storage introduces multiplexing gain,

$t$0

which measures how much virtual capacity can be oversold relative to physical capacity, and probability of blocking,

$t$1

which measures the likelihood that the physical battery cannot follow virtual commands (Rasouli et al., 2019). In the empirical analysis, access to external resources at retail price yields an optimal battery of $t$2 MWh and $t$3 MW, multiplexing gain of 10%, probability of blocking of 9.6%, and annual CSO profit of \$t$41.5M/yr when no external resources are available (Rasouli et al., 2019). This demonstrates that BaaS reliability and profitability can be traded off explicitly through overbooking and external balancing.

The degradation literature also extends BaaS beyond first-life operation. The intertemporal swapping framework states that its decision support can be used for battery trading, renting, and secondary use, and recommends second-hand market use or less cycling-intensive services such as contingency reserve, backup, and black-start support when economic end-of-life arrives before physical end-of-life (Chen et al., 2023). A plausible implication is that mature BaaS ecosystems may depend on life-cycle asset routing, in which batteries migrate between service tiers as their residual value and capability change.

Battery-as-a-Service therefore comprises a set of technical and economic arrangements in which batteries are pooled, centrally managed, and allocated as time-varying service resources. Whether implemented as vehicle battery swapping, cloud storage, community capacity rental, or autonomous robotic power logistics, the defining challenge is the same: to optimize battery availability, throughput, degradation, and revenue across heterogeneous users and competing applications while preserving fast service and operational reliability (Mahoor et al., 2017).