SVZ interface study. Independent concept work, not a Vinci product release.
Peak temperature
83.4°C
Demo valuePrototype values. Read from Vinci’s published 157-second demonstration, with intermediate points modelled to keep the object continuous. Wired to swap for your solver output, no redesign required.
You have the hardest problem in engineering software solved, and a website that
makes it look like a faster version of what everyone else sells. We thought the
fastest way to prove we understood the gap was to close a piece of it.
The problem is not that people doubt the physics
More than half the top twenty semiconductor companies have already benchmarked
Vinci. That audience is convinced. The gap is everyone downstream of them: the
VP who signs, the analyst who writes the category, the automotive or aerospace
engineer who has not yet realised the same model applies to their stack. They
arrive at a page of claims and leave with a number they cannot feel.
A thousand times faster is a spec. It is also the least interesting thing
Hardik says about the product. The interesting thing is that physics stops
being a gate and starts being a constraint the design lives inside. You cannot
argue someone into that. They have to feel a design answer back.
So we rebuilt the object from your demo, and nothing else
We watched the public thermal demonstration once. No codebase, no solver
access, no briefing. From it we reconstructed the eighteen-layer stack, the
field, the threshold and the verdict, and put them in a browser where a
stranger can orbit the package, drop a probe through the stack, and move the
convection axis until the design passes.
The point is not this package. The point is the distance between watching a
recorded demo and holding the thing. That distance is where your pipeline
currently leaks, and it is the distance a real partner closes.
What is real here, and what is not
Saying this before you ask is the whole point. The instrument is a
reconstruction, and we would rather be the agency that draws the line than
the one that gets caught at it.
From your demo
Eighteen layers · peak 83.9 °C at the baseline condition · minimum 25.0 °C ·
95 °C target · top three sources 39% of power · top ten 68% · vertical to
lateral conduction 3:1 · h=125 fails at 97.8 °C · h≈144 passes at 94.5 °C.
Every one of these is read from the published run.
Modelled by us
The values between your measured points, so the object stays continuous when
a visitor drags the axis. Layers the demo never names are left unnamed rather
than invented. The source concentration follows a fitted power law that
reproduces your 39% / 68% to within a tenth of a point.
The fourth axis is convection, which your own roadmap places next and which
Hardik is explicit is a boundary condition today rather than a solved field.
We built it exactly that way. Swapping any of this for solver output is a data
change, not a rebuild.
What this says about working together
This took one pass, from a single public video, before a contract existed. It
is a fair sample of the standard: we build the thing rather than describe it,
we mark the boundary between evidence and estimate without being asked, and we
would rather show you a working object with honest limits than a beautiful
page that promises more than the model does.
The same object drives the homepage, the launch surface, sales collateral,
docs and every digital surface after that. One reconstruction, reused
everywhere, retuned with your team.
Independent SVZ interface study. Not a Vinci product release, and not a claim on
your solver.
