Complex Dynamics of Intelligence-Level Adjustment in Intelligent and Connected Vehicle Manufacturers

Against the backdrop of software monetization and carbon constraints reshaping competition, it is necessary to identify the stability boundary for intelligence adjustment. We build a heterogeneous duopoly of fuel-vehicle (FV) and new-energy-vehicle (NEV) firms that jointly choose output and intelligence each period, where intelligence shifts effective willingness to pay through an own-uplift effect and a rival-squeeze effect. Under a bounded-rational updating rule, we characterize global dynamics using the largest Lyapunov exponent, single-parameter bifurcations, basins of attraction, and a three-dimensional feasible region. Results show a clear L-shaped stability corridor and a joint threshold in the adjustment-speed space. Moderate adjustment converges, whereas faster NEV adjustment tends to reach period doubling and chaos earlier. Higher per-vehicle software revenue raises NEV intelligence and narrows the corridor. Stronger substitutability initially lifts both sides, but as rivalry intensifies, it weakens FV incentives. Higher carbon costs depress the FV side and further elevate the NEV side, with multistability and attractor switching near the boundary. Overall, the study reveals the mechanisms that govern stable and unstable intelligence adjustment under heterogeneous monetization and carbon constraints, and it provides theoretical support for cadence governance at the firm level as well as for standards coordination and paced monetization at the policy level, helping the industry advance toward orderly, resilient, and sustainable development.