Insight · Data centers & BESS · 16 October 2026

Power sooner. Grow in modules. Support the grid.

What BCG and Deloitte/Agora tell investors about flexible data centers - and how we translate that into a project decision.

The investment question has changed.

A cheaper electricity tariff is little help while a data center waits for power. The relevant comparison includes when customers can start paying, what flexibility costs and who bears the risk when the grid or the customer cannot deliver.

Our view: batteries and distributed compute deserve a joint assessment where they improve the route to paid operation. The investment case must show which benefits the project actually keeps.

What the two papers actually say

Different questions, complementary findings.

BCG · 18 August 2026

The value of being ready earlier

BCG examines Northwest Europe over 2030-2035. It combines four choices: where to locate, whether to accept a connection that can be curtailed, which workloads can move, and how much storage or generation to install behind the meter. Energy efficiency remains a starting point.

The key commercial finding is that earlier energisation can outweigh the added cost of onsite energy resources. Lower electricity bills alone often do not justify the change. A grid-positive design aims to limit additional system investment while making better use of existing infrastructure.

BCG also identifies the trade-offs: moving away from demand can add latency; curtailment needs a credible operating response; onsite resources need space, capital and permissions. Its findings support comparing whole delivery and operating models, rather than choosing a site on its electricity price alone. These are modelled scenarios, not an achieved return for a small modular project. [1]

Deloitte / Agora · 20 April 2026

Flexibility has value at system level

The European study compares minimum flexibility obligations with broader demand optimisation. In its mandated scenario, adjustments during 1.5% of annual hours affect 0.6% of data-center electricity demand, avoiding 4 GW of thermal capacity and €500 million in annual system costs by 2035.

Wider optimisation brings further benefits, but needs suitable incentives and predictable connection agreements. These Europe-wide results do not establish a particular site's connection date, operating income or capacity to shift workloads. [2]

Our interpretation: system value becomes investable only when connection terms or contracts allow the project to capture part of it.

Our logic: price the route from power to paying customers.

Start with three realistic alternatives: wait for a firm connection; accept earlier flexible access and add the necessary assets; or use smaller sites in a coordinated network. Each needs the same customer demand, service requirements and investment horizon. Otherwise, a shorter timetable or a higher revenue forecast can disguise a weaker underlying project.

1. Value earlier operation after all delivery costs.

Bring the expected customer cash flows forward only when the connection, permits, cooling, communications and customer acceptance can genuinely be ready. Include the earlier capital payments, interest during construction and cost of idle capacity. A module delivered quickly is valuable only if it can enter paid service.

2. Agree what can flex before selling the benefit.

Separate workloads that can be deferred from services that must remain continuously available. Specify response time, duration, notice, recovery and compensation. Storage may bridge an interruption, but its usable energy and reserve policy determine for how long. A battery's MW rating alone does not answer that question.

3. Protect the battery's own investment case.

Energy reserved for backup cannot simultaneously earn every merchant-market opportunity. Charging the battery and running the facility must fit the connection limit. Include displaced trading margin, degradation, controls and reserve requirements before adding grid-service revenue. Resilience can have commercial value, but the customer must know which service is being purchased.

4. Make the advantage visible in contracts.

Earlier access, capacity fees, energy pass-through and flexibility payments have different payers and risks. Compare their net cash flows under slower occupancy, lower battery revenue and delayed power. The resulting decision should state which option to progress, what would change the recommendation and which evidence is still missing.

The direction is visible beyond the papers.

Architecture, recruitment and demonstrations provide different kinds of evidence.

NVIDIA

NVIDIA and Emerald AI describe coordinated workloads, generation, batteries and grid signals, with priority workloads protected. This makes energy-aware compute an explicit architecture direction; it is not evidence of a particular project's return. [3]

Tesla

Tesla's Megapod infrastructure recruitment is a signal of interest in modular compute. A job listing does not establish a deployed product, its operating performance or a Supercharger rollout. [4]

China

Original reporting describes a modular compute pod tested in Guiyang. China's energy regulator separately reports examining mobile compute and coordinated electricity-compute dispatch. These are relevant demonstrations and policy signals, not proof of European deployment economics. [5] [6]

Where our design fits

Why Kinetic Shield makes sense to investigate.

The modular, containerized reference design combines 9.9 MW BESS and a 2 MW data center per site. Multiple smaller sites form a coordinated network, linked through private/industrial 5G and/or fiber. Quantum Key Distribution (QKD) can be added as an extra security layer on suitable links.

This offers a way to evaluate phased investment, proximity to available power and users, and coordinated storage and compute. Connectivity, suitable workloads and local power access determine whether distribution helps. The design is offered through a fabless design/IP licence and qualified implementation partners.

Explore the modular design

Three investor takeaways

Compare time to revenue, not electricity price alone.
Contract and price flexibility before counting it.
Scale modules around verified power and customer demand.

QKD supplements the security architecture; it does not replace endpoint protection or guarantee uninterrupted service.

The next commercial step

A scoped pilot can compare site and connection options, quantify BESS and flexibility economics, and identify the energy, water and resilience data needed for reporting. The output should be an investment choice with a delivery plan and open conditions - including a clear reason to stop if the numbers do not work.

The cited organisations are independent sources. Their work does not endorse or validate Kinetic Shield. BESS energy capacity, facility load boundaries and grid import/export limits require separate design.

Request more information Explore a data-center pilot