The Self-Driving Startup Nobody Saw Coming | E2289

What It Covers

Two self-driving startup CEOs — Wave's Alex Kendall and Wabi's Raquel Ratzen — detail how end-to-end AI and world models are moving autonomous vehicles from science projects to mass-market products. Wave targets 2.5 million Nissan vehicles annually by 2027, while Wabi pursues a minimum 25,000-robotaxi Uber partnership with Volvo as OEM partner.

Key Questions Answered

• •World Model Architecture: World models serve dual functions in autonomous driving: they create rich representations of what physically matters on the road (lane markings, traffic signals, intersecting objects) while simultaneously generating photorealistic simulation environments. Training on hundreds of petabytes of data across dash cams, internet video, and OEM fleets, these models now incorporate radar and lidar alongside cameras, enabling adversarial stress-testing without real-world safety consequences.
• •Licensing vs. Fleet vs. OEM Strategy: Three distinct business models exist in autonomous driving. Building proprietary vehicles limits scale to one brand. City-by-city fleet deployment requires high capital expenditure. Licensing AI to existing manufacturers and fleet operators — Wave's and Wabi's shared approach — captures the largest addressable market by leveraging partners' existing manufacturing scale, distribution, and demand aggregation without owning physical assets.
• •L2 to L4 Gap Is Engineering, Not Science: The transition from hands-off highway driving to fully driverless operation is no longer a scientific unknown — it is an engineering execution problem. Required steps include integrating validated hardware into OEM platforms, scaling training data and compute along a predictable curve similar to LLM scaling, and completing safety validation across diverse global domains before regulatory submission.
• •Nissan Partnership Scale: Nissan's commitment to deploy Wave technology across 90% of its vehicle lineup represents approximately 2.7 million units annually by financial year 2027 — roughly double Tesla's total annual production volume. This single partnership illustrates the leverage of the OEM licensing model: one contract generates more deployment volume than an entire competitor's manufacturing capacity.
• •Subscription Pricing Trajectory: Consumer autonomous driving features are moving toward recurring subscription models, mirroring Tesla's $100 monthly FSD charge. OEMs are testing bundled inclusion, one-time fees, and free trials before converting to subscriptions. The recurring model aligns incentives because ongoing software updates, safety improvements, and insurance cost coverage require continuous revenue rather than a single upfront hardware sale.
• •Wabi's Per-Mile Revenue Model: Wabi charges carriers and operators on a per-mile basis rather than upfront licensing fees, creating direct alignment between customer value and revenue. Volvo, as Wabi's primary OEM partner, plans hundreds of commercially deployed trucks by 2027 through its Volvo Autonomous Solutions unit. The Uber Freight partnership covers billions of miles of deployment across North America's top carriers.