Why GM will give you Gemini – but not CarPlay (www.theverge.com)

GM announced a major reorientation of its product and software strategy: a Google Gemini–powered in-car AI assistant arriving in 2026, a ground-up hardware and software platform debuting on the Escalade IQ in 2028 that will enable an upgraded Super Cruise with true Level 3 “eyes-off” and expanded hands‑off driving, plus new robotics and home battery businesses. The moves come alongside the hire of Sterling Anderson (ex‑Tesla/Aurora) to run product, and a $1.6 billion EV-related writedown as demand, tariffs, and the expiry of U.S. EV tax credits reshape the market — prompting capacity shifts like repurposing a Lake Orion plant for high-demand full‑size vehicles. GM also continues its decision to sideline Apple CarPlay/Android Auto in favor of its own integrated software stack. For the AI/ML community this matters because it signals large‑scale deployment of multimodal LLMs inside the vehicle domain and an industry push toward vertically integrated, software‑defined cars. Integrating Gemini implies heavy cloud connectivity, on‑vehicle inference/latency tradeoffs, and product‑specific fine‑tuning for safety, voice, and sensor fusion. Choosing a proprietary UI over CarPlay highlights tradeoffs between openness and tighter control needed for autonomy and certification. The announcements create opportunities (data for model improvement, energy-aware control across EVs and home batteries, robotics convergence) as well as challenges around data governance, regulatory approval for Level 3 systems, and the engineering work to make LLMs reliable and safe in mission‑critical driving contexts.