Capacity Is Not the Same Thing as Capability

If you read only the headlines this month, you would conclude that the industry is in excellent shape. Battery manufacturers have returned to profitability. A major automaker has committed a gigafactory to grid storage. A large Texas utility has published an economic case for tens of billions in annual benefit from new transmission. A national trade association has quantified demand growth through 2050. A university laboratory has demonstrated a transformer technology that has been theoretical for two decades.

Read them as an engineer and a different picture emerges. Every one of these items describes capacity —manufacturing capacity, transmission capacity, generating capacity, conversion capacity. None of them describes capability: the ability to place that capacity into a specific network at a specific point of interconnection, with validated models, coordinated protection, a defensible facility rating, and a schedule that survives contact with a cluster study.

That gap is where projects die. It is not usually a technology failure. It is a study that was never run, a model that never matched the as‑built equipment, a fault‑current assumption that stopped being true when the surrounding generation mix changed, or a protection scheme that was coordinated against a synchronous machine and now sits next to a converter that will not produce the current it was expecting. The equipment in this issue is genuinely impressive. The engineering discipline required to install it safely has not become any easier.

The through‑line in this issue: Cell capacity, conductor rating, converter topology and transformer technology are all upstream design inputs. Each one changes the study set required before energization. When the hardware changes faster than the study practice around it, the risk does not disappear — it moves to the commissioning window, where it is most expensive to find.

Ford has entered grid‑scale storage through a new energy division, converting Kentucky cell manufacturing capacity from electric‑vehicle LFP production toward stationary storage. The flagship product is a liquid‑cooled 20‑foot DC block built on 512 Ah LFP cells, rated at 5.45 MWh, offered in two‑hour and four‑hour configurations, with a 1040–1500 V DC operating window, IP55 enclosure rating, roughly 43.5 tonnes container weight and operation to 4,000 m altitude without derating. The company has indicated 20 GWh of annual manufacturing capacity, with first customer deliveries expected in late 2027. Because the units are assembled domestically, they are positioned as compliant with foreign entity of concern and prohibited foreign entity restrictions and as eligible for domestic content adders.

Why the cell capacity number matters at the site, not just on the datasheet

Most fielded storage still uses 280 Ah or 314 Ah cells. The industry has been migrating toward a 587–628 Ah class, which is now approaching volume production at several manufacturers. Ford is entering above the old standard but below the emerging one, and at the container level that difference is material: competing blocks in the 587–588 Ah class are rated near 6.25 MWh against Ford’s 5.45 MWh.

For a 400 MWh project that is roughly a dozen additional containers. Additional containers mean additional foundations or piers, additional DC collection runs and combiner terminations, more medium‑voltage transformer positions, a larger fenced footprint, longer AC collection, and a proportional increase in the number of thermal management and fire suppression systems to commission and maintain. Cell capacity is not a procurement detail. It propagates into civil layout, collection design, land acquisition and the commissioning schedule.

Comparative positioning based on publicly announced manufacturer specifications. Figures are as published by the respective manufacturers and have not been independently verified by Keentel Engineering.

Three items an owner’s engineer should press on

• The thermal ceiling. An upper operating limit of +55 °C is below much of the competing field. In West Texas, Arizona, the Central Valley or the Gulf Coast, the governing number is not shade ambient — it is enclosure surface temperature under solar gain with the liquid cooling system already working. Ask for the derate curve, not the operating range.
• Undisclosed round‑trip efficiency and cycle life. These are the two inputs a storage revenue model is most sensitive to. Where they are not published, they should be contractual: guaranteed RTE at defined conditions, and a capacity maintenance schedule with defined augmentation obligations.
• No fielded operating history. A first‑deliveries date of late 2027 means no MOD‑026 or MOD‑027 validation record, no established generic model stack in the transmission planning libraries, and no EMT model with a track record of matching field response. For a project inside a cluster study, that is a schedule risk before it is a performance risk. Learn more about NERC compliance services

Interim results across the listed Chinese battery manufacturers show the broadest profitability recovery the storage supply chain has seen in several years. After the 2024–2025 price war drove utility‑scale cell pricing to roughly RMB 0.3/Wh, prices recovered quarter over quarter through the first half of 2026. Leading manufacturers used their storage segments to lift gross margins; second‑tier suppliers returned to positive gross margin across the board, helped by a sharp rebound in capacity utilization.

Compiled from company interim reports and performance forecasts; shipment shares reflect third‑party market research estimates. Several entries are forecasts rather than filed results and should be confirmed against final filings before use in a commercial decision.

What this changes in a project financial model

The deflation assumption is dead for now. Models built on continued cell price decline through 2027 should be revisited. Quarter‑over‑quarter price recovery with capacity effectively sold out is the opposite condition.

Gross margin is a bankability input. The spread between a manufacturer earning roughly 24% on storage and one earning roughly 12.5% is the difference between a supplier funding twenty years of warranty obligation out of margin and one relying on future volume to cover it. Independent engineers should be reading supplier financials alongside datasheets.

Lead times and configuration flexibility tighten together. When lines run at or near full capacity, the ability to negotiate non‑standard container configurations, custom protection interfaces or accelerated delivery erodes. Procurement decisions need to be sequenced earlier relative to the interconnection study schedule, not later.

The compliance and cost frontiers are diverging. The manufacturers with the deepest cost position and the longest operating record are largely not the ones that resolve prohibited foreign entity exposure. Procurement now has to solve for both constraints at once, and the engineering consequence is that the preferred technical solution and the preferred tax‑credit solution may specify different equipment.

A grid reliability study published this month by the National Electrical Manufacturers Association projects United States electricity demand rising roughly 55% by 2050, driven by artificial intelligence workloads and hyperscale data centers, industrial growth, transportation electrification and shifting consumer patterns. The more useful part of the analysis is the framing: the central challenge is not producing more energy, it is moving existing generation to where it is needed. That reframing turns a generation problem into a transmission, protection and interconnection problem — which is to say, an engineering problem rather than a capital problem.

Grid‑enhancing technologies are study deliverables, not products

The discussion of solutions covers grid‑enhancing technologies, advanced transmission and reconductoring, dynamic line ratings, demand response, storage, vehicle‑to‑grid, data center interconnection, supply chains, permitting reform, and a widening workforce gap. It is worth being precise about what these actually require, because they are frequently described as though they were equipment purchases. See our power system studies capabilities.

Keentel Engineering assessment of the study and commissioning work implied by each technology class.

CenterPoint Energy has released a third‑party economic study of proposed resiliency and reliability transmission projects in Southeast Texas built around 765 kV development. The study estimates more than $18 billion annually in statewide economic benefit, over $120 million per year in local property taxes, and roughly 41,000 construction jobs, with joint efforts alongside other ERCOT utilities projected as high as $30 billion in annual economic value and $300 million per year in new tax revenue.

These are claimed benefits from a utility‑commissioned study supporting a capital program that requires regulatory approval. They should be read as advocacy, and cited as such. The engineering content underneath them, however, is real and consequential — and largely absent from the coverage.

An EHV overlay is a different design practice, not a bigger version of the same one

ERCOT’s transmission backbone is built at 345 kV. Introducing a 765 kV layer is a step change in design practice, and the differences are not incremental. Explore our transmission line design services.

Keentel Engineering assessment of the design and study implications of an EHV overlay in a predominantly 345 kV network.

Resiliency and reliability are not synonyms. The coverage uses the two terms interchangeably; the engineering does not. Reliability is steady‑state adequacy — enough capacity and enough network to serve load under credible contingencies. Resiliency is the ability to withstand and recover from low‑probability, high‑consequence events, and it is delivered through different measures: structure class upgrades, storm loading criteria, selective undergrounding, sectionalizing and restoration strategy. A 765 kV overlay is primarily a reliability and deliverability investment. Undergrounding and hardening programs are resiliency investments. Both may be justified; they answer different questions, and the studies that support them are different studies.

North Carolina State University, working with the New York Power Authority and EPRI, has designed and tested a high‑efficiency solid‑state transformer rated to 1 MW under realistic operating conditions, demonstrated at EPRI’s laboratory facility in Lenox, Massachusetts. The technology is presented as improving efficiency at high‑demand installations including data centers, reducing conversion heat loss, and offering a compact footprint suited to space‑constrained urban sites.

What the technology actually is

A solid‑state transformer replaces the 60 Hz magnetic core with a power electronic conversion chain: medium‑voltage AC is rectified, converted through a high‑frequency isolated DC‑to‑DC stage built around a medium‑frequency transformer, and inverted back to AC or delivered directly as DC. Because transformer core volume scales inversely with operating frequency, moving isolation from 60 Hz to the kilohertz range collapses the magnetics. Megawatt‑class ratings at medium voltage have become practical largely because of silicon carbide devices with the blocking voltage to handle MV directly.

The efficiency claim, stated honestly: A well‑designed conventional medium‑voltage transformer operates above 99% efficiency. A solid‑state transformer is a multi‑stage converter, and stage‑by‑stage it will not beat that. The credible efficiency argument is at the system level: if a data center campus can accept medium‑voltage‑to‑DC conversion directly, the solid‑state transformer displaces a transformer plus a rectifier stage plus part of the uninterruptible power supply chain. Compared against that stack, the case is strong. Compared against a transformer alone, it is not. Any specification that claims otherwise is comparing the wrong boundary.

What you give up, and what you gain

Keentel Engineering comparison of design and operating considerations.

Iberdrola has agreed to acquire 80% of the equity in Caruna, Finland’s largest electricity distribution operator, for approximately $2.3 billion, in a transaction valuing the company including debt at roughly $5.7 billion. Swedish and Finnish pension insurers retain minority positions of 12.5% and 7.5%. Caruna serves more than 1.5 million people — over a fifth of the Finnish population — across roughly 89,000 kilometres of network operating under two regional distribution concessions. Notably, 67% of that network is underground, concentrated around Helsinki and the Joensuu region. Industrial activity and demand growth tied to new data centers and residential development in western and north‑eastern Finland were cited as drivers.

Data center demand is now global

The data center demand story is now global and it is showing up in asset pricing. The same driver behind United States interconnection queue congestion is being underwritten into acquisition multiples in Northern Europe.

Two‑thirds underground at distribution scale is the number engineers should notice. That ratio changes the engineering profile of a network: cable ampacity and thermal environment become the governing rating constraint, fault location practice shifts from visual patrol to injection and travelling‑wave methods, and restoration times move from hours to days when a failure does occur — in exchange for far fewer failures.

It is a useful comparison to the Texas hardening programs. Finland arrived at heavy undergrounding largely through outage‑duration regulation; Houston is arriving at it through storm experience. Same engineering solution, two different regulatory routes to it — and a reminder that undergrounding decisions are ultimately driven by the cost of interruption, not by the cost of cable.

Pension capital in the cap table signals the asset class. Regulated distribution is being priced as long‑duration, inflation‑linked infrastructure. That raises the standard for technical due diligence, condition assessment and asset‑life modelling on both sides of a transaction.

The U.S. Trade and Development Agency is hosting a delegation of roughly fifteen government and private‑sector power sector decision‑makers from Angola, Botswana, Mozambique and Zambia, travelling to Denver and Austin from 29 August to 5 September. The programme focuses on securing and modernizing transmission and distribution systems, with demonstrations of network technologies and AI‑enabled grid control, and meetings with utilities and regulators including the Colorado Public Utilities Commission and ERCOT.

Why the engineering problem there is not the engineering problem here

It is easy to read “grid modernization” as the same work described in section 03. It is not. These four countries sit in and around the Southern African Power Pool, a network characterized by very long high‑voltage interties, large hydroelectric resources on the Zambezi system, and mining and industrial loads whose profiles look nothing like a North American distribution feeder. The binding problems are voltage stability over long lines, system integrity protection schemes, inter‑area oscillation, metering and revenue protection, and telemetry coverage — not congestion driven by rapid data center load growth.

The same caution applies to AI‑enabled control as applies domestically, only more sharply: an optimization layer is only as good as the measurement layer beneath it. Where SCADA coverage, time synchronisation and data quality are the constraint, the highest‑value early work is instrumentation, communications architecture and model validation — the unglamorous foundation that makes advanced control meaningful rather than decorative.

Keentel note

Keentel maintains an Austin office and works on transmission and substation design, power system studies and SCADA architecture across all of these problem classes. USTDA reverse trade missions, feasibility studies and technical assistance grants are a recognised route by which US engineering firms enter these markets. Firms interested in Southern African T&D work should be tracking USTDA programme announcements directly.

Seven items, one recurring pattern: the hardware is changing faster than the study practice around it. Below is what we would put in front of an owner, developer or EPC on the strength of this month’s developments.

Keentel Engineering recommendations. Project‑specific advice requires project‑specific study.

How Keentel supports this work

Keentel Engineering service offerings. For specific engagements, contact our team.