Uber will operate its own robotaxis again—this time with Rivian’s not‑yet‑built EVs

Uber agreed to purchase 10,000 fully autonomous Rivian R2-based robotaxis with an option to scale to 50,000 and is investing $300M upfront plus up to $950M more contingent on milestones. Deployment is planned for San Francisco and Miami in 2028 with a goal of 25 cities by 2031. The deal shifts Uber away from an asset-light model, exposing it to asset and operational risk while effectively subsidizing Rivian’s autonomy R&D as Rivian delays EBITDA profitability expectations beyond 2027. Key risks: Rivian has not completed R2 development or production, regulatory/safety exposure given Uber’s 2018 fatality history, and potential strain on Uber’s existing AV partnerships.

Analysis

Uber’s move from pure marketplace to owning robots is a levered bet on capture of unit economics rather than on control of software. If 10k robotaxis operate 40–60k miles/year and reduce per‑mile costs by $0.40–$0.80 vs human drivers, that implies $160–$480m of incremental gross margin annually at scale — enough to swing segment-level ROIC even after depreciation, but only if utilization and software uptime hit targets. The milestone‑based purchase structure outsources product execution risk to OEMs while concentrating utilization, insurance and residual‑value risk on Uber; that mismatch makes short‑term KPI tracking (fleet utilization, downtime, incident rate) more valuable than headline deployment promises.

Strategically, OEMs who secure large order books (or equity commitments) get implicit non‑dilutive financing and demand visibility; expect more vehicle OEMs to chase captive fleet orders as a financing channel. Conversely, independent autonomy software providers face a tougher sales motion: partners integrated into Uber’s network could be sidelined in core urban corridors, compressing pricing power for pure‑play stacks and advantaging horizontally‑sold hardware/software platforms (benefit: standardized compute suppliers and sensor makers). Downstream, commercial insurers, city permitting authorities, and service/charging networks become marginal-cost drivers for unit economics — a regulatory incident or a localized permit reversal would cascade through utilization and residual values.

Key catalysts and tail risks live on a 6–36 month horizon: OEM development milestones, Rivian/Lucid cash needs and dilution events, first large‑scale incident, and early city utilization data (2028 pilots are critical). A positive sequence (on‑time tech, improving utilization, no major incidents) should re‑rate marketplace multiples as recurring robotic revenue replaces a portion of human driver volume; the reverse (development delays, accidents, utilization <50%) risks materially impairing assets and forcing write‑downs. Monitor OEM capex announcements, insurance pricing, and municipal pilot restrictions as high‑signal datapoints.

Key Decisions for Investors

• Long UBER (12‑18 month tenor): buy UBER call spreads to limit premium outlay (target 2:1 reward:risk). Rationale: optionality on successful rollouts and take‑rate capture; size as 2–3% of equity sleeve. Trim if utilization <50% in first quarter post‑pilot or after any Tier‑1 incident.
• Long LCID (6–12 month): accumulate on weakness given OEMs with fleet orders convert order books into working capital and higher near‑term revenue visibility. Risk: 30–40% downside if EV demand weakens or build costs rise; target asymmetric payoff by using LEAPs or funded collars.
• Long NVDA (6–12 month calls, modest size): play standardized compute demand as multiple OEMs and stacks consolidate on similar silicon/hardware. Reward: secular multiple expansion if robotaxi deployments accelerate; risk: macro slowdown in IT spend.
• Relative trade — short WRD / AUROW vs long LCID or UBER (12 months): expect disintermediation and pricing pressure on independent autonomy stacks as fleet‑owners internalize or vertically preferentially allocate. Keep position small and hedge for binary regulatory/incident jumps which can widen correlations.