Architecture is the most underrated meta-question of the decade. We study how reasoning about it matures across industries.

A cross-industry survey tracing how electrical/electronic architecture converges at the infrastructure layer while control semantics and burden of proof diverge — and what that means for automotive, robotics, and embodied AI. The paper introduces two original frameworks: AR0–AR5 (Architecture Readiness) and AI²-ML (Architecture Intelligence Maturity Levels).

An empirical assessment of E/E architecture maturity across 22 major global automakers, scoring each on five dimensions and the AR0–AR5 ladder in both its deployed state and its confirmed near-term roadmap. The central finding is structural: leadership on architecture maturity has migrated away from the industry’s traditional centers — only Tesla and Huawei HIMA reach AR4, while Chinese new entrants dominate the AR3 cluster and the European and Japanese/Korean majors trail. A methodological throughline runs across the report: functional-safety certification is not a proxy for architecture maturity.

The concluding paper of the series takes the Tesla FSD–Optimus unified stack as the empirical anchor closing the loop between D1’s AR0–AR5 framework and D2’s 22-OEM assessment. Through a clinical dissection from silicon to behavior, it distils Tesla’s eleven officially disclosed shared layers into three reuse mechanisms — physical, intelligence, and an implicit organizational twelfth layer — and characterizes the precise boundary of cross-morphology reuse: accumulated capability is reusable, morphology-dependent high-frequency motion control is not. The central thesis: Tesla functions as an AR4 reference frame, not a template for replication.

A maturity benchmark of 12 architecture-design toolchain vendors on the AI²-ML ladder. Only SCADE reaches L3; the category hugs the ground while its summit — the empty frontier — stands vacant. The heat of AI has decoupled from the capacity for reasoning.