Publishing Disclaimer: In all of its publications and products, Military Review presents professional information. However, the views expressed therein are those of the authors and are not necessarily those of the Army University, the Department of the Army, or any other agency of the U.S. government.

---

 

The Energy Requirements for Processing Data on the Future Battlefield

 

2nd Lt. Aaron Lawrence, U.S. Army
Dr. Vikram Mittal

 

Download the PDF

 

Military technologists frequently assert that the future battlefield will be “transparent” since the widespread use of autonomous platforms will allow military forces to have complete knowledge of their area of operations. Coupled with this proliferation of autonomous systems is an increased demand for computer processing to handle the large amount of data being received by units. Indeed, a common axiom holds that data will soon be as vital as bullets or diesel fuel.¹ However, this data processing requires energy that forward-deployed units must now supply. This issue is becoming larger as units are fielding increasing numbers of advanced autonomous systems, which will demand far greater energy to transmit, store, and process information.²

This analysis estimates the energy requirements needed to store and process the data stemming from these autonomous systems to help inform the design of logistical frameworks for warfare in 2040. It assesses the data processing needs and associated energy requirements at the brigade combat team (BCT) level and explores viable energy solutions, including solar power and modular nuclear reactors, alongside conventional diesel fuel.

Composition of the Future Army Brigade Combat Team

The estimated composition of an Army BCT in 2040 is based on future force structure projections developed by Army Futures Command (now U.S. Army Transformation and Training Command) and prior research by Army professionals on emerging technologies.³ For this analysis, autonomous systems are broadly grouped into three categories: unmanned aircraft (UA), unmanned ground vehicles (UGV), and unmanned ground sensors (UGS).

The assessment begins at the platoon level, estimating the type and number of assets assigned to a standard infantry platoon. Both light rifle and heavy infantry platoons are considered, with asset allocations differing between the two.⁴ These platoon-level estimates are then scaled to the company level, which consists of three light platoons, one heavy platoon, and additional company-level assets. The battalion level includes three infantry companies and relevant battalion-level assets. Unlike current doctrine, which includes a separate heavy company, this analysis assumes the integration of heavy weapons at the company level through a dedicated heavy platoon. This approach aligns with current efforts to increase organic capabilities at lower echelons and reduce reliance on higher-level support. It also reflects evolving doctrine that emphasizes distributed lethality and network-based operations, enabling smaller units to operate independently.

The brigade-level structure, consisting of seven battalions, is expected to remain consistent within the current organization. Three of these will be infantry battalions, while the remaining four are projected to include reconnaissance, field artillery, engineer, and support battalions. Each is assigned an estimated number of autonomous assets based on its combat role. These estimates are represented in the brigade row of the table. The table provides the estimated number of autonomous assets at each echelon but does not include lower-echelon assets already nested within those totals. For example, the three UAs listed at the company level do not include the two UAs assigned to each of the company’s platoons.

Figure 1 presents a high-level illustration of the relationship between autonomous data-collecting systems and power requirements. The information gathered by UAs, UGVs, and UGSs must be processed, which requires energy. In figure 1, a centralized data center receives input from these systems, with a small power plant supplying the necessary energy for data storage and processing. Because this scenario assumes a BCT is forward deployed, the unit must be self-sufficient and generate its own power. Whether processing occurs at centralized locations, is distributed across multiple data centers, or takes place onboard the sensing platforms themselves, the energy demands remain the responsibility of the brigade.

Data Processing Requirements

While the exact design specifications of future assets are unknown, their data transmission and processing requirements are expected to be similar to current systems. Each of the three categories of autonomous assets is likely to include a variety of subtypes, each with unique specifications and energy needs. However, for the purposes of this analysis, each category is represented by a single proxy system. These proxies, drawn from both military and civilian sources, were selected based on publicly available design specifications. Civilian systems were used in some cases due to the limited availability of unclassified military data. While these proxies provide similar capabilities to what is expected of future military systems, actual energy demands may be higher due to added security requirements not reflected in the available data.

UA proxy. The UA proxy is the Parrot ANAFI USA, a quadcopter drone with thermal imaging, GPS navigation, and AES-256 encryption for secure data transmission. Though originally designed for law enforcement, its capabilities make it a suitable stand-in for military UAs expected to conduct reconnaissance, target acquisition, and communications relay missions. These missions will require day and night surveillance, secure communications, GPS tracking, and low acoustic signatures. The ANAFI’s data throughput (shown in figure 2), including video and GPS transmission, is used in this analysis.⁵

UGV proxy. The UGV proxy is the Squad Multipurpose Equipment Transport (S-MET), a military system designed to support small units through logistics transport and mobile power generation. Capable of carrying up to one thousand pounds and operating for seventy-two hours without resupply, S-MET operates remotely and transmits GPS location, logistical inventory updates, and movement data. While future UGVs may have broader missions, such as route clearance or fire support, S-MET represents the likely most common use case: autonomous logistics support. Its throughput data is summarized in figure 2.⁶

UGS proxy. The UGS proxy is the Ouster OS2 Long-Range High-Resolution Imaging Lidar. This sensor produces 3D environmental maps and can detect moving targets at distances up to four hundred meters. It was selected due to the availability of its technical specifications and its representative data output. UGS systems are expected to be widely used for persistent surveillance; detection of vehicles and personnel; multidomain integration; and chemical, biological, radiological, or nuclear monitoring. Although data output may vary across sensor types, the OS2 provides a reasonable average for this category.⁷

Figure 2 summarizes each proxy system’s data throughput in megabits per second and estimates daily operational use as a percentage of a full twenty-four-hour day. These values are used to estimate the total amount of data in gigabytes that a BCT must process daily. The daily usage is estimated based on the environmental and operational conditions associated with deploying each sensor. For example, the UA would require to be recharged for a portion of the day.

Altogether, the different autonomous systems will generate approximately 53,370 gigabytes of data per day that the BCT must process. Although much of this data will be discarded and not stored, it must undergo a degree of initial processing to identify changes on the battlefield that will be relevant to the warfighter. This substantial volume will require careful logistical planning for energy generation, data processing, and storage. While improvements in computing technology may reduce this burden in the future, meeting these requirements will remain a key operational challenge.

Power Requirements

Following the calculation of total daily data generation from the autonomous asset proxies, the analysis proceeds to estimate the energy required to transmit, process, and store this volume of data. This portion of the study carries the greatest level of uncertainty due to significant variability in published estimates of energy consumption per gigabyte. Specifically, these estimates depend heavily on system architecture, assumed processing depth, network security protocols, and data storage strategies.

Most existing literature addresses energy consumption in the context of commercial internet usage. A widely cited analysis conducted by Carnegie Mellon University reported that the internet consumes approximately 7 kilowatt-hours per gigabyte (kWh/GB) of data, and this value is commonly used in a number of different studies.⁸ In contrast, another study reported significantly lower values for fixed-line networks, with an average of 0.06 kWh/GB of data.⁹ An analysis published by the American Council for an Energy-Efficient Economy proposed a more moderate estimate of 5 kWh/GB.¹⁰

These figures, however, are not directly transferrable to the BCT context. Most studies address energy costs for an open, globally distributed civilian network, whereas the envisioned military system operates on secure, closed networks with higher encryption standards and reduced tolerance for latency or data loss. Additionally, the scope of energy accounting differs across studies, with some including only data transmission, others considering limited storage, and few offering a comprehensive estimate that reflects secure processing and encrypted transmission within tactical edge environments.

For this study, a nominal value of 5 kWh/GB is adopted. This estimate lies at the midpoint of published figures and is adjusted upward to reflect the anticipated energy overhead associated with encryption, secure routing, and hardened storage. While likely conservative compared to projected civilian energy consumption rates in 2040, this value provides a reasonable approximation for a tactical network operating in austere conditions with limited access to cloud-based efficiencies.

It is also important to recognize that not all data will be processed at a centralized data center. A portion may be processed locally on the autonomous platform itself, with only relevant or filtered information transmitted to the central node for further aggregation or long-term storage. However, even when processing occurs at the edge, energy is still consumed by the onboard systems of UAs, UGVs, or UGSs. Because a BCT is responsible for recharging these systems, the associated energy expenditure remains within the unit’s overall power requirement.

Applying this energy factor to the projected data volume of 53,370 GB per day results in an estimated daily energy demand of 266,850 kWh to support data transmission, limited processing, and storage. This energy requirement is substantial. For example, the average household in the United States only uses 29 kWh per day.¹¹ Further, a current BCT’s largest generator is only 60 kilowatts; to produce the required energy for this data center, a BCT would need 185 such generators.

Energy Sources for Future BCT Operations

The total daily energy requirement calculated in this study can be satisfied by both traditional energy sources used by the military and experimental sources that will likely be functional in 2040. These energy sources include new generators powered by diesel fuel and biodiesel, solar panels, and a modular nuclear reactor. The results of this analysis are summarized in figure 3.

Diesel fuel is the most commonly used form of fuel in the Army today and is likely to remain in use for the foreseeable future. For a diesel generator to provide the requisite energy each day, and assuming that data processing is evenly spread across the day, a twelve-megawatt generator would be needed. This generator would produce 288,000 kWh of energy each day, allowing for some surge capacity and downtime. This generator is quite large and may need to be composed of multiple small generators, for example, six two-megawatt generators.

Additionally, the generator would require a steady supply of diesel fuel. The energy content of diesel is approximately 10.8 kilowatt-hours per liter (kWh/L). Modern diesel generators are 35 to 40 percent efficient at the megawatt level.¹² Looking toward 2040, this efficiency will likely increase to 45 percent due to advances in engine material properties and improved combustion processes. As a result, approximately 55,000 liters of diesel fuel would be required each day.

By 2040, biodiesel is predicted to become much more widely available and may become a logistics fuel for the United States.¹³ The generator used for diesel can also be used for biodiesel. Biodiesel’s energy content is slightly less than traditional diesel at 9.9 kWh/L. With a similar 45 percent efficiency, 60,000 liters of biodiesel would be required each day.

For perspective, the standard refueling truck in the U.S. arsenal, the M969A1, has a capacity of 19,000 liters. An armored BCT typically includes fifteen of these tankers, along with an additional forty-eight small tankers. This illustrates that sustained operations would consume a significant amount of fuel.

Another potential source to meet this energy requirement is a solar farm. Unlike diesel generators, solar power does not require a steady stream of liquid fuel; instead, it relies on incident sunlight. The average capacity factor for a solar farm is 22.8 percent. That means a one-megawatt solar farm can produce 5.4 MWh of energy per day under clear skies, which drops with clouds or latitude.¹⁴ To account for these conditions, a 48.5 megawatt solar farm would be needed.

The amount of solar irradiation on the earth is approximately 1,000 watts per square meter.¹⁵ While modern solar panels are 20 to 23 percent efficient at converting solar energy, advances in multijunction cells, tandem perovskite silicon designs, and improved manufacturing techniques could raise their efficiency to 35 percent by 2040. At that point, the solar panel array would need to be approximately 140,000 square meters, which is roughly the area of twenty-eight football fields.

Furthermore, a significant amount of battery storage would be needed to enable the system to provide energy through the night. In this case, the battery bank would need to hold approximately 133,000 kWh of energy. Continuing current battery trends, the battery energy density could reach 500 watt-hour per kilogram. This would require 266,000 kilograms of batteries to provide energy after sundown.

Another potential energy source for a future BCT’s data processing is a mobile, modular nuclear reactor, such as the system under development in Project Pele. This reactor is designed to deliver between one and five megawatts of electrical power continuously for at least three years.¹⁶ The 1.5-megawatt version of this reactor is being designed to fit into four twenty-foot shipping containers.¹⁷ Because of the high energy needs of future BCTs, it will likely be designed closer to the higher end of its range. This analysis assumes it would produce five megawatts continuously, which equates to 120,000 kWh per day. As a result, three of these systems would be needed to meet the data processing needs of a future BCT.

Sensitivity Analysis

Similar to any study about future technology, this analysis had to make a substantial number of assumptions, including the makeup of the future fighting force, their usage on the battlefield, and the power requirements associated with data processing. As such, a sensitivity analysis was performed to better understand the impact of these assumptions.

The analysis assumes autonomous systems are integrated at all levels across a BCT. However, the number of systems was somewhat conservative. There is potential for even more autonomous systems to be present on the battlefield, collecting additional data that will need processing. The power and energy requirements for processing this data scale with the number of autonomous systems. For example, if the number of autonomous systems increases by 20 percent, both the power and daily energy requirements would increase by 20 percent. For diesel options, the generator would need to be 20 percent larger and 20 percent more fuel would be required. For the solar farm, the number of solar panels and battery backups would need to increase by 20 percent. For the portable nuclear reactor, which typically comes in five-megawatt modules, this 20 percent increase can be covered by adding additional reactors.

The systems considered in this analysis are not equal. The UGS, for instance, operates for a longer duty cycle than the UGV or UA and handles significantly more data while in operation. As a result, adding additional UGSs would have a greater impact on power and energy demands than adding additional UGVs or UAs.

Similar to the number of autonomous platforms on the battlefield, the amount of data would increase if the usage of the systems also increases. For example, while UGSs are already used 100 percent of the day, if UAs and UGVs are also used at 100 percent, the amount of energy would increase to 394,200 kWh. This 47 percent increase in energy demand would result in a similar increase in fuel requirements and solar panels. It would also increase the number of five-megawatt modular nuclear generators by one.

This analysis used a standard approximation of 5 kWh/GB of data processed. However, there is uncertainty about this number, with some claiming it to be closer to 9 kWh, and others stating it may be as low as 0.06 kWh. The energy requirement is in large part based on what is being done with the data, with 5 kWh including light processing and storage. Decrypting the data, processing it, and then re-encrypting it, which would be the likely military case, may approach 9 kWh. If the energy required per gigabyte of data processed increases by 80 percent to 9 kWh, the power and energy requirements will likewise increase by 80 percent.

Conclusions

While a common adage is that “data will be the bullets of the future battlefield,” an equally important adage is that “amateurs think of strategy, professionals think of logistics.” Data will be inherently used on the future battlefield, especially as the number of robots increases, effectively turning it transparent. However, this also imposes a significant logistical burden on units as they attempt to process all that data.

This analysis examined the potential number of robotic systems that a BCT might employ in 2040, determined the associated data processing requirements, and estimated the power and energy needed to support that processing. The analysis found that data processing requires a substantial amount of power. To meet this demand, a twelve-megawatt diesel generator could be used, requiring 55,000 liters of diesel or 60,000 liters of biodiesel each day. Alternatively, a fifty-megawatt solar farm covering approximately 140,000 square meters or three modular nuclear reactors rated at five megawatts each could provide the necessary power. This logistical burden is significant and should be considered in any analysis of the future of data on the battlefield.

 

---

Notes

1. T. S. Allen, “Finding the Enemy on the Data-Swept Battlefield of 2035,” Military Review 100, no. 6 (2020): 28–37, https://www.armyupress.army.mil/Journals/Military-Review/English-Edition-Archives/November-December-2020/Allen-Data-Swept-2035/.
2. Allen, “Finding the Enemy.”
3. Training and Doctrine Command (TRADOC) Pamphlet 525-92, The Operational Environment and the Changing Character of Warfare (TRADOC, 2019).
4. Vikram Mittal et al., “System Design of the Rifleman of the Future,” Proceedings of the IEEE International Systems Conference (SysCon) (2021): 1–8, https://doi.org/10.1109/SysCon48628.2021.9447053.
5. “Technical Specifications: ANAFI,” Parrot, accessed 17 November 2025, https://www.parrot.com/en/drones/anafi/technical-specifications.
6. Mark Mazzara, “Army Ground Robotics Overview: OSD Joint Technology Exchange Group,” 24 April 2018, https://jteg.ncms.org/wp-content/uploads/2018/04/02-PM-FP-Robotics-Overview-JTEG.pdf.
7. “Technical Specifications: OS2, Long-Range High-Resolution Imaging Lidar,” Ouster, updated 11 February 2021, https://data.ouster.io/downloads/datasheets/datasheet-revd-v2p0-os2.pdf.
8. Christopher L. Weber et al., The Energy and Climate Change Impacts of Different Music Delivery Methods (Microsoft Corporation; Intel Corporation, August 2009), 12, https://www.researchgate.net/publication/252659417_The_Energy_and_Climate_Change_Impacts_Of_Different_Music_Delivery_Methods.
9. Thomas W. Jackson and Ian R. Hodgkinson, “Debate: The Data Threat to 2050 Net Zero—Public Administrations’ Responsibility for the ‘Data-Scape,’” Public Money & Management 44, no. 3 (2024): 182, https://doi.org/10.1080/09540962.2023.2279812.
10. David Costenaro and Anthony Duer, “The Megawatts Behind Your Megabytes: Going from Data-Center to Desktop,” Proceedings of the 2012 ACEEE Summer Study on Energy Efficiency in Buildings (2012): 13-65, https://www.aceee.org/files/proceedings/2012/data/papers/0193-000409.pdf.
11. “Use of Energy Explained: Electricity Use in Homes,” U.S. Energy Information Administration, updated 18 December 2023, https://www.eia.gov/energyexplained/use-of-energy/electricity-use-in-homes.php.
12. John B. Heywood, Internal Combustion Engine Fundamentals, 2nd ed. (McGraw Hill, 2018).
13. Vahid Pirouzfar et al., “Power Generation Using Produced Biodiesel from Palm Oil with GTG, STG and Combined Cycles; Process Simulation with Economic Consideration,” Fuel 314 (2022): 123084, https://doi.org/10.1016/j.fuel.2021.123084.
14. Center for Sustainable Systems, “Photovoltaic Energy Factsheet,” Pub. No. CSS07-08 (University of Michigan, September 2025), https://css.umich.edu/publications/factsheets/energy/solar-pv-energy-factsheet.
15. Raymond A. Serway and John W. Jewett, Physics for Scientists and Engineers, 9th ed. (Cengage Learning, 2014).
16. U.S. Department of War, “DOD to Build Project Pele Mobile Microreactor and Perform Demonstration at Idaho National Laboratory,” press release, 13 April 2022, https://www.defense.gov/News/Releases/Release/Article/2998460/dod-to-build-project-pele-mobile-microreactor-and-perform-demonstration-at-idah/.
17. “Project Pele Begins Taking Shape with Start of Core Manufacturing,” BWXT, 24 July 2025, https://www.bwxt.com/project-pele-begins-taking-shape-with-start-of-core-manufacturing/.

 

 

 

 

Back to Top

January-February 2026