60 accounts · two separate questions —
pricing (demand held identical) and distribution
(demand explicitly not) · illustrative price points for a robotics
foundation-model company · synthetic data
Question 1 — Pricing: revenue by month, demand held identical
Same robots, same tasks, same customers — only the pricing
logic differs, so the spread between these lines isolates the pricing
architecture and nothing else. On identical units, subscription earns most
because it charges most per robot; the royalty rate billed on the
same base earns $34.8M against
$49.0M. As a price, royalty is simply
worse.
The four dimensions that actually decide it
Model
Revenue (18mo)
Exit ARR
Gross margin
MoM volatility
Top-3 concentration
subscription
$48.98M
$49.2M
95.7%
0.1%
32%
metered
$28.29M
$28.2M
92.6%
5.3%
29%
royalty
$34.83M
$35.0M
93.5%
0.1%
32%
hybrid
$34.80M
$34.9M
94.0%
3.4%
30%
Question 2 — Distribution: royalty + OEM reach
This is a different question, and it is not a
pricing choice. Embedding in OEM production reaches
2.4× the units you could sell
direct, so demand is deliberately not held constant here. On that
basis royalty earns $83.60M against
$48.98M for subscription — it wins on
reach, not on rate.
And that is where the risk appears. The same royalty rate
carries 32%
concentration when billed direct and
61% when it flows through OEM
partners who own the customer relationship, the hardware roadmap, and
therefore the renewal. Concentration is a consequence of the
distribution decision, not the pricing one — which is
exactly why the two questions have to be answered separately.
Metering aligns price with delivered value
and is the least forecastable — 5.3%
month-over-month volatility against
0.1% for subscription. That alignment
is bought with forecast error, and forecast error at a company still proving
its plan to a board is expensive in ways the revenue line doesn't show.
Royalty is not a pricing choice. It wins revenue
here only because it reaches 2.4× the units — it is a
go-to-market bet, trading the customer relationship for reach. Comparing it
to the direct models on revenue alone is a category error.
Among the direct models, hybrid sits between the
other two on both axes: $34.8M revenue at
3.4% volatility.
What this model cannot answer. Demand here is exogenous and
identical across all four architectures — that is what makes the comparison
clean, and it is also the limitation. It cannot value the two things that
actually push companies toward hybrid: the expansion motion, where revenue
grows with usage without a renegotiation, and the churn avoided by not
overcharging light users. On the axes this model can measure, plain
per-robot subscription wins among the direct options. Anyone reading that as
"therefore price per robot" has taken the model past what it supports —
which is worth saying out loud on the page rather than in a footnote.