Frontier lab for physical AI / worldAI

TAM TOP-DOWN

Four market estimates span $0.89B to $81.4B depending on scope. The $81.4B figure includes all hardware and services; industrial robotics hardware alone accounts for roughly $34B. A frontier lab selling intelligence platforms and foundation models maps to the embodied AI platform layer: $3.8B in 2025 per MarketIntelo, corroborated directionally by the $0.89B software-only figure from MarketsandMarkets at the narrow end. Splitting the two at the platform-plus-model layer yields an honest midpoint near $3.8B to $5.1B. Using $5.1B as the honest upper bound for the intelligence layer TAM in 2025.

MarketIntelo platform TAM $3.8B x 1.34 growth factor (38.5% CAGR, one year) = $5.1B

TAM BOTTOM-UP

IFR reports 4.664M industrial robots in operation globally in 2024, with 542,000 new units installed annually. AI feature add-ons cost $10,000 to $50,000 per robot (evidence midpoint $30,000). Applying $30,000 AI software value to the 542,000 annual install base yields a gross figure. Not all installations adopt frontier AI; current penetration of advanced AI features is assumed at 35% based on the evidence that 45% of advanced labs collaborate with AI startups, discounted for actual purchase conversion.

542,000 installs x 35% AI adoption x $30,000 AI feature value = $5.7B gross; discounted to $4.0B for software-only revenue share (~70% captured by platform vendors)

SAM FILTERS

From $5.1B TAM, three filters applied. Geography: Americas and Europe represent 25% of new deployments (IFR: Asia 74%, Europe 16%, Americas 9%); Western markets are the only near-term addressable geography for a new frontier lab given regulatory familiarity and enterprise procurement cycles, yielding 25%. Segment: frontier model buyers (OEMs, large manufacturers, logistics players) represent roughly 40% of the platform market; the rest is low-cost commodity software. Readiness: only 40% of qualified buyers are at procurement-ready stage in year 1 to 3.

$5.1B x 25% geography x 40% segment x 40% readiness = $204M bottom; top-down: $5.1B x 10% reachable share = $510M; display $510M, range $290M to $810M

SOM BUILD

A competent new entrant in year 1 signs 5 enterprise pilots at $200,000 each plus API licensing to 20 smaller robotics firms at $50,000 annually: $1M to $2M year 1, below 1% of SAM. Year 2 scales to 15 enterprise contracts plus 60 API customers: roughly $8M. Year 3 targets 35 enterprise contracts at $300,000 average plus 120 API customers at $75,000: approximately $31M, representing 6% of the $510M SAM. Competitors with data flywheels constrain share above this level without a proprietary hardware or data partnership.

Year 3: 35 enterprise x $300,000 + 120 API x $75,000 = $10.5M + $9.0M + $11.5M platform fees = $31M