Liquid-Cooled HPC in 2026: Cable Current, Cabinet Cooling and the Uptime Economics Behind 800 A

Quick Answer

Liquid cooling became the default for high-power charging in 2026 because air cannot carry the current. Commercial deployments now run liquid-cooled cables at up to 600 A on all-in-one 400 kW chargers and up to 800 A on modular dispensers, with cabinets that scale from 400 kW to a 1.2 MW corridor hub by adding power modules and dispensers rather than rebuilding infrastructure, as described in the ABB E-mobility and Oreve announcement of 5 November 2026. The engineering consequence is that cooling capacity — not nominal power — determines sustained output on a hot afternoon. Buyers should specify cooling at the cabinet, cable and connector level together, then verify derating curves rather than datasheet peaks.

Key Takeaways

• Cooling sets the real power ceiling. A 400 kW air-cooled charger that derates at 35 °C delivers less annual energy than a 360 kW liquid-cooled unit that holds its rating.
• Cable current is the binding constraint at the vehicle. Liquid-cooled assemblies sustain 600–800 A continuously while staying light enough for drivers to handle, which is why heavy-duty and megawatt-adjacent projects specify them as standard.
• Modular cabinets protect capital. Scaling from 400 kW to 1.2 MW by adding modules and dispensers avoids replacing power infrastructure when utilisation grows.
• Uptime is a specification, not an outcome. Public funding regimes increasingly require 97% uptime; independent module fault isolation is the hardware feature that makes that number achievable.
• Cooling units are a separate procurement line. Dispenser-integrated units in the 3.5–9 kW class and cabinet-level units in the 25–72 kW class serve different thermal loads and should not be conflated.

Why Liquid Cooling Moved From Premium to Baseline

Liquid cooling stopped being a differentiator in 2026 because the alternative stopped meeting the duty cycle. A 500 A continuous session through an air-cooled cable assembly generates heat in the conductor that cannot be removed fast enough at ambient temperatures above roughly 35 °C, so the charger either derates or the connector exceeds its touch-temperature limit. Both outcomes are commercial failures at a public hub where the value proposition is a 20-minute stop.

Thermal limits are also why liquid cooling arrived first on cables rather than cabinets. The conductor cross-section needed to carry 800 A without active cooling would make the cable too heavy for a driver to manipulate against a truck’s inlet, particularly at the low mounting heights used on heavy-duty vehicles. Pumping coolant through a smaller conductor solves the weight problem and the heat problem at the same time.

The market signal in 2026 is that liquid-cooled cables now appear on mainstream 400 kW all-in-one products, not only on flagship megawatt equipment. In the ABB E-mobility and Oreve corridor rollout announced on 5 November 2026, both the A400 all-in-one chargers and the modular M-Series OM dispensers use liquid-cooled cables, sustaining up to 600 A and up to 800 A respectively. That is a mainstream freight deployment in France, not a technology demonstration.

Air Cooling Still Has a Place

Air cooling remains the correct choice below roughly 250 A of continuous output, where the thermal load is manageable and the maintenance burden of a coolant circuit is not justified. Sungrow’s IDC80E destination charger, launched at Power2Drive Europe on 26 June 2026, uses isolated air cooling with IP65, IK10 and C5 ratings, claims 99% uptime and 99.5% one-time charging success, and supports currents up to 250 A with 97% efficiency.

The lesson for buyers is segmentation, not partisanship. A depot doing overnight and opportunity charging at 60–120 kW per vehicle gains nothing from a coolant loop. A highway hub selling minutes gains everything.

The Three Cooling Layers Buyers Must Specify Separately

Liquid cooling is often sold as a single feature and delivered as three subsystems with independent failure modes. Specifying them together is the difference between a charger that holds its rating and one that protects itself into mediocrity.

Layer 1: Connector and Cable

The connector is the thermally hardest point because the contact interface is small and the touch-temperature limit is safety-critical. MIDA’s liquid-cooling system range covers 3.5 kW, 4.5 kW, 6 kW and 9 kW integrated units for dispenser and cable cooling, plus 2.4 kW and 3.5 kW split-type units where the heat exchanger sits away from the dispenser. Cable assemblies for CCS1, CCS2, GBT, NACS and CHAdeMO are manufactured in the same group, which matters because a coolant circuit is only as good as its quick-couplings and hose routing.

Layer 2: Power Module and Cabinet

Inside the cabinet, the power modules are the dominant heat source. Liquid-cooled modules allow higher switching density in a smaller enclosure, which is why a 1.2 MW cabinet fits on a footprint that would be impossible with air-cooled modules of the same rating. MIDA’s liquid-cooled power module range spans 40 kW, 60 kW, 75 kW and 125 kW units, and cabinet-level cooling units are offered in the 25–72 kW class for HPC applications.

Layer 3: Site and Ambient

The third layer is the site itself: heat rejection to ambient, acoustic limits, and dust ingress in industrial or desert locations. A cabinet rated for 50 °C ambient with a coolant loop sized for 45 °C can still fail if the heat exchanger is installed in a recess with no airflow. Layout drawings should show the heat-rejection path, not only the equipment pad.

Reading a Datasheet: Peak Power Versus Sustained Power

The single most expensive mistake in HPC procurement is comparing peak ratings. A charger advertised at 480 kW may deliver that figure only at 25 °C ambient with a single outlet active and a battery pack that accepts the full current. Real sites rarely meet all three conditions.

Three numbers matter more than the headline:

1. Derating curve versus ambient temperature. Ask for output power at 25 °C, 35 °C, 45 °C and 50 °C. A flat curve to 45 °C is a strong indicator of adequate cooling margin.
2. Simultaneous-output behaviour. A 400 kW cabinet serving two dispensers may share dynamically, so the second vehicle’s arrival halves the first session’s power. Dynamic sharing is fine for retail forecourts and wrong for hub operators who promise a specific session time.
3. Efficiency at part load. A charger that is 96% efficient at full load and 91% at 30% load wastes more energy across a year than its headline efficiency suggests. Part-load efficiency is where cooling design and module count show up.

A Worked Derating Comparison

The following figures are illustrative modelling assumptions used to show the method; they are not measured test data for any specific product.

The 40 kW nameplate advantage of Unit A disappears at 40 °C and reverses at 45 °C. In markets where summer ambient exceeds 40 °C for several weeks, the lower-rated liquid-cooled unit outperforms on delivered energy — which is the only metric that generates revenue.

Uptime: The Number That Decides Public Funding

Reliability moved from an operations concern to a procurement gate. In North America, the NEVI Formula Program ties public funding to 97% uptime and 150 kW minimum power per port, a requirement widely summarised in the industry outlook published on 16 January 2026. In Europe, AFIR requires open access, ad-hoc card payment and ISO 15118-20 support, all of which depend on a charger that stays online.

Liquid cooling contributes to uptime in two ways that are easy to overlook. First, lower operating temperatures slow the aging of capacitors, fans and semiconductor junctions, which reduces mid-life failure rates. Second, a liquid-cooled cabinet can be sealed more aggressively against dust and humidity, because it does not need the same air volume moving through the enclosure — a genuine advantage in coastal, desert and industrial sites.

The hardware feature that matters most for the 97% target, however, is independent module fault isolation. A cabinet built from 30 kW or 40 kW modules can lose one module and continue serving at reduced power, which keeps the port available and the uptime metric intact. A monolithic power stage fails to zero.

Design for Service, Not Just for Power

Service access determines realised uptime more than mean time between failures does. Two questions decide it: can a module be replaced without a crane or a full cabinet shutdown, and is the coolant circuit serviceable by a trained technician with standard tools on site? MIDA’s split architecture, including liquid-cooled split DC fast chargers in the 800–960 kW class, separates the power cabinet from the dispenser so that heavy maintenance happens away from the customer-facing forecourt.

Modular Scaling: Why 400 kW Now and 1.2 MW Later Beats 1.2 MW Now

Modularity is the mechanism that reconciles uncertain demand with multi-year grid lead times. The ABB M-Series described on 5 November 2026 lets a site start at 400 kW with a handful of charging points and grow to a 1.2 MW corridor hub by adding power modules and up to 24 dispensers without rebuilding the underlying power infrastructure. That is the same architectural principle behind MIDA’s split-type range, which spans 360 kW to 1,680 kW and includes a 720–800 kW liquid-cooling satellite dispenser design for motorway sites.

Staged scaling changes the financial profile in three ways:

• Capital tracks demand. Modules are bought when utilisation justifies them, not on a forecast.
• Connection capacity is negotiated once. A single medium-voltage application covering the ultimate site load avoids a second queue.
• Dispenser placement stays flexible. Satellite dispensers can be added at parking bays without touching the cabinet yard.

The counter-argument is that modular systems cost more per kW at small scale. That is true and usually irrelevant, because the alternative is paying for capacity years before revenue arrives.

Total Cost of Ownership for a Liquid-Cooled Hub

TCO for HPC is dominated by four lines, only one of which is the charger. A defensible model includes:

A common error is treating equipment CAPEX as the whole comparison. On a 1.2 MW hub, the grid and civil line typically rivals or exceeds the charger line, and the downtime line can erase a year of margin if a single-vendor proprietary module is three weeks away by sea. Buyers should require a spares list, a module swap procedure and a documented MTTR target in the contract, not just a warranty period.

Where Liquid Cooling Does Not Pay

Liquid cooling is the wrong purchase when sessions are long and predictable. Overnight depot charging at 60–120 kW per vehicle, workplace charging, and residential use all fall below the thermal threshold where coolant loops earn their cost. In those cases a wall-mounted DC platform such as MIDA’s WY Series 40–80 kW range delivers the required energy with simpler maintenance. Matching cooling class to duty cycle is the single largest saving available in a mixed portfolio.

Specification Checklist

Use the following checklist in a tender to make cooling and uptime comparable across bidders:

• [ ] Output power stated at 25 °C, 35 °C, 40 °C, 45 °C and 50 °C ambient
• [ ] Continuous current rating of cable and connector, with coolant flow and pressure
• [ ] Cooling unit type, rated heat rejection and service interval
• [ ] Simultaneous-output policy: dynamic sharing or dedicated power per dispenser
• [ ] Part-load efficiency curve at 25%, 50%, 75% and 100% load
• [ ] Module granularity and independent fault isolation behaviour
• [ ] IP and corrosion ratings for the installed environment, including C4 or C5 for coastal sites
• [ ] Communication stack: OCPP 1.6J and 2.0.1, ISO 15118-20 readiness
• [ ] Spare module availability, MTTR target and remote diagnostic capability
• [ ] Coolant specification, top-up procedure and technician training requirement

Frequently Asked Questions

1. What does liquid cooling actually cool in an EV charger? Liquid cooling removes heat from three points: the cable conductor and connector contacts, the power modules inside the cabinet, and in some designs the DC busbars. Coolant is pumped through channels in the cable assembly and a cold plate on the modules, then rejected through a heat exchanger.

2. How much current can a liquid-cooled cable carry? Current commercial heavy-duty products sustain up to 600 A on all-in-one 400 kW chargers and up to 800 A on modular dispensers, as described in the ABB E-mobility and Oreve announcement of 5 November 2026. Higher currents in the 1,000 A class exist for megawatt-adjacent equipment.

3. Is a liquid-cooled charger more expensive to maintain? It has a different maintenance profile rather than a uniformly higher one. Coolant level and quality checks, hose and quick-coupling inspection, and pump service replace air filter and fan replacement. Sealed liquid-cooled cabinets often need fewer filter changes in dusty environments.

4. Do I need liquid cooling for a 120 kW depot charger? No. Below roughly 250 A of continuous output, air cooling with module-level airflow is sufficient and simpler to service. Liquid cooling earns its cost at 400 A and above, or where ambient temperatures routinely exceed 40 °C.

5. What is the difference between split-type and all-in-one HPC? A split-type system separates the power cabinet from the dispensers, so one cabinet serves multiple parking bays and dispensers can be added without new power infrastructure. An all-in-one charger integrates power and dispenser in one enclosure, which is simpler to install but harder to scale.

6. How does liquid cooling affect uptime targets like 97%? Lower operating temperatures reduce component aging, and sealed cabinets resist dust and humidity. The decisive uptime feature, however, is independent module fault isolation, which keeps a port available at reduced power instead of failing to zero when one module faults.

7. Can a liquid-cooled platform be upgraded to megawatt charging later? Often yes, if the cabinet’s DC bus, metering, protection and cooling capacity were sized for the higher load and the dispensers are on a split architecture. Retrofitting megawatt capability onto an undersized all-in-one unit is usually a replacement rather than an upgrade.

Conclusion

Liquid cooling in 2026 is an economic decision disguised as a technical one. It costs more per kW at purchase, returns more per session in hot climates and high-current duty cycles, and protects uptime metrics that now gate public funding. The specification work that matters is unglamorous: derating curves by ambient temperature, simultaneous-output policy, module granularity, and a service plan that a local technician can execute. Operators who buy on sustained output rather than nameplate power will find that the cheaper-looking charger is the more expensive asset.

Sources referenced: ABB E-mobility / Oreve corridor charging announcement, 5 November 2026; Sungrow Charging launch at Power2Drive Europe 2026, 26 June 2026; HoLa public MCS commissioning, 29 September 2025; NEVI uptime and power requirements as summarised in the EV charging outlook of 16 January 2026; MIDA product catalogue data.