DESIGN INTELLIGENCEJune 27, 2026·Mary · DEPIX Design Intelligence

Clay no longer decides the car. The headset does.

For sixty years the most important room in any car company was the one with the clay plate. A full-size model, sculpted by hand over a steel armature, kept warm and worked by people who had spent decades learning to read a highlight running down a fender. The clay was where a car stopped being a drawing and started being an object. It was also where the decisions got made: which surface lives, which dies, which proportion the chief executive signs off when he walks around it under studio lights. Everything upstream was a proposal. The clay was the verdict.

That order is now inverting, and a small British studio just said the quiet part out loud. Avant Design built the Longbow electric sports car, shown at CES in January 2026, almost entirely in virtual reality. Its director, Chris Gould, is blunt about why. "Fundamentally, to produce a car you need a clay model that then gets scanned into CAD," he told Develop3D in March. "And this is why we use virtual reality, because it essentially just eliminates that step." Avant has offered physical clay modelling since 2018. In all that time, Gould says, not a single client has asked for it.

The claim that should make every studio director sit up is not about cost. It is about where the verdict happens. Standing inside a full-scale VR suite in a Varjo headset, a team can walk around a one-to-one model, reshape a surface with their hands, switch the paint, the light and the location instantly, and line up as many as fifteen variants in the same scene. Gould reckons this delivers roughly eighty per cent of the feedback a physical clay model gives. The missing twenty per cent is real—you cannot run your palm down a virtual flank, you cannot bump into it—but eighty per cent is more than enough to kill or keep a surface. The decision has moved off the clay plate and into the headset.

Clay is not gone. It has been demoted. Look at how sculptors like Nick Graveley now work: surfaces are resolved in Gravity Sketch, the mesh goes straight to the mill, and the physical model exists only to take the last one or two millimetres of tactile refinement before it is scanned back into the digital file. Graveley says VR has replaced about ninety per cent of his form-finding clay work, and that a milled model out of VR is "astonishingly close—it looks exactly like the model you have been looking at." That is the whole point. When the milled object merely confirms what the headset already showed, clay has stopped being the verdict and become the finishing pass.

This is uncomfortable because clay carried more than form. It carried authority. The clay review was a ritual—the one moment when the most senior person in the building had to physically commit. Move that commitment into VR and you change who decides and how fast. Fifteen variants in one scene means a chief designer can reject thirteen of them before lunch, with no modeller waiting three weeks to rebuild the loser. The friction that used to protect a decision—the sheer cost of being wrong in clay—is gone. That is liberating and dangerous in equal measure. Speed is only an asset if the judgement keeping pace with it is good.

Which is the real lesson here, and it has nothing to do with headsets. The bottleneck was never the clay. It was the quality of the decision made in front of it. Tooling that compresses weeks into an afternoon does not make those decisions better; it just makes them faster and more numerous. A studio that pushes evaluation as far upstream as possible—judging proportion, stance and surface intent while it is still cheap to be wrong—wins. A studio that simply generates fifteen options and picks the prettiest thumbnail has automated its own indecision.

This is exactly the seam DEPIX works in: not making the model, but sharpening the call before anyone commits material or money to it. The headset moved the verdict upstream. The job now is to make sure there is real design intelligence waiting for it when it arrives—because a faster way to be wrong is not progress, and the studios that understand the difference will own the next decade of how cars get made.