The Pack That Is Its Own Chassis

A useful way to evaluate a pack architecture is to ask how much of it is battery. The industry metric is the gravimetric cell-to-pack ratio: total cell mass divided by total pack mass. Published teardown data and pack databases put conventional designs commonly in the range where cells account for roughly 60 percent of pack mass, with the remaining 40 percent consumed by enclosure, thermal management, battery management electronics, and interconnection. Aggregated pack data show the ratio of cell-level to pack-level energy density clustering near 0.62, which says the same thing from the energy side: integration overhead consumes something like a third of the energy density the cells offer.

The industry's response has been consolidation. Cell-to-pack and cell-to-chassis designs eliminate intermediate module structure and push cell mass fractions toward 85 to 90 percent in the best production examples. The direction of travel is clear: every kilogram that does two jobs is a kilogram that does not need to exist twice.

Most of that consolidation work has happened in high-volume automotive programs, where the engineering investment amortizes over hundreds of thousands of units. Low-volume and development programs have largely been left with the conventional split: a structural frame designed by one discipline, a cooling system designed by another, and an integration phase where the two negotiate for the same volume.

The MonoLith architecture applies the consolidation logic at a scale where it has historically been unaffordable, by making one decision early: the cold plate is the structure.

Each battery sub-unit is built around an extruded aluminum thermal plate, I-shaped in cross section, with internal coolant passages running its full length. The plate presents thermally conductive faces on both sides. Battery modules mount in rows on both faces, attached through lengthwise T-slot channels that hold each module in compressive thermal contact with the plate surface. The compressive attachment matters: the same mechanical joint that carries structural load is what guarantees the thermal interface, so structural integrity and thermal contact are not separate quality outcomes to verify. They are one outcome.

The I-section is not incidental. An I-beam concentrates material where bending stiffness demands it, and the web between the flanges is where the coolant passages live. The passages include internal surface finishing to increase coolant contact area, so the volume that provides structural depth is simultaneously the volume that provides heat exchange area. Nothing about the cross section is dedicated to only one function.

Mounting two rows of modules back to back on opposing faces has a second-order benefit worth naming: the plate is loaded symmetrically. Thermal expansion and mechanical loads from the two module rows oppose each other across the web rather than bending the plate in one direction.

The consolidation would be a curiosity if it did not change how the pack scales. It does, because of what an extrusion is.

An extruded profile has a fixed cross section and an arbitrary length. Once the die exists, a plate of any length costs the same per meter, carries the same coolant passages, and presents the same mounting channels. Because modules attach along continuous lengthwise channels rather than to discrete mounting features, module count per plate is set by plate length, not by tooling. A program that needs more energy specifies a longer plate and more modules. The cross section, the module design, the thermal interface, the manifolds, and the attachment method are all unchanged.

This is the mechanical fact underneath the platform's configuration space. Energy scales along the plate axis. Voltage scales through module series configuration. Capacity scales through parallelization and sub-unit count. None of these axes requires new structural engineering, which is why a large variant catalog can exist without a correspondingly large tooling catalog.

Scaling continues at the next level of assembly. Sub-units mount side by side along rails, and their coolant connections are deliberately linear: end cap manifolds at the plate ends align so that supply and return runs are straight tubing sections that extend as sub-units are added. The plumbing scales the same way the structure does, by getting longer rather than by getting redesigned.

Structural panels then close the assembly. Panels attach top and bottom into the same T-slot channel system, through connectors in the thermal plates themselves, stiffening the assembled group of sub-units and enclosing them against contamination and contact. End caps complete the enclosure. The panels are not an afterthought; they are the members that tie the parallel plates into a single torsionally stiff assembly, a point revisited in the trade-offs section.

Because modules attach to open channels rather than being captured inside a welded or bonded structure, individual modules and entire sub-units can be removed and replaced without disassembling the pack. The mounting rail arrangement preserves access, and the linear coolant runs mean a sub-unit can be disconnected without disturbing its neighbors' plumbing.

For a production automotive pack, field-level module replacement is a minor consideration; packs are replaced as units and economics are driven by volume. For the programs this platform serves, the calculus is different in two specific ways.

The conventional cost of a ground-up pack structure for a low-volume program is spread across mechanical design of the frame and enclosure, thermal design of the cooling system, the analysis loop where the two are reconciled, prototype tooling, and validation of the integrated result. Much of that work is not reusable: it is specific to one energy target, one envelope, one program.

The consolidated architecture moves most of that engineering into the platform, where it is done once. What remains program-specific is configuration: plate length, module count and series arrangement, enclosure selection, and interfaces. This is the difference between designing a structure and specifying one, and it is why the platform's deployment timeline is measured in weeks where ground-up development is measured in quarters.

A structural point sits underneath the economic one. Because the load path and the thermal path run through the same validated part, the platform's structural validation and thermal validation travel with it from program to program. A ground-up design revalidates both, every time.

The architecture fits programs where engineering time is the scarce resource and requirements are still in motion: prototype and development vehicles, low-volume specialty platforms, and fleets that value commonality across mixed vehicle types. It fits applications that can accept quantized energy steps in exchange for configuration speed, and envelopes compatible with a fixed cross section scaled in length.

It is not a cell-to-chassis production architecture, and does not claim to be. High-volume programs with frozen requirements will always extract more from a bespoke integrated design. The argument here is narrower and, for its intended user, more useful: the consolidation benefits that high-volume programs obtain through program-specific engineering can be obtained at development scale through architecture instead.