Autonomous vehicles & the robotaxi transition (new) Disruption slower and less total than priced ◆

2026-AUG-15 — Joseph Carlson (Qualtrim deep dive, "Why Uber Will Win the AV War"): the density argument — AV operators must build per-city density from scratch (~$100k lidar vehicles, 3-D mapping, permitting, parking, maintenance) while Uber can plug AVs into an already-dense network at 100% utilization from day one. The structural constraint is peak demand: Friday spikes ~6× base load, so an AV-only fleet must either hold a fleet idle ~90% of the day or concede the peak, whereas humans supply the surge free because they park their own cars. Hence a hybrid human+AV network beats a 100% AV network even at the end state, and "Uber can be late and still be first" (window 5–7 years; only a decade-long delay lets Waymo take critical share). Waymo's San Francisco proof point is an outlier: ≈2× LA's population density, 4–5× the US top-50 average, ~2× as affluent, ~half the trip distance. Tesla's robotaxi dismissed as "somewhat fanciful" so far (no lidar, small areas). Ackman's Q2 letter (relayed 2026-AUG-13) makes the same call: Uber at ~19× earnings, EPS on pace for ~35% growth, "valuation increasingly disconnected from the fundamentals." Pieter Slegers (Compounding Quality), 2026-FEB-26 sizes the robotaxi option against the price being paid for it: the global autonomous-vehicle market is projected at "over $214 billion by 2030 growing at a 19.9% CAGR" — a fraction of Tesla's $1.3trn market capitalisation — while the incumbent is already commercial: "Waymo (Google): The leader in Robotaxis, with driverless cars already picking up passengers in cities like Phoenix and San Francisco." McKinsey is cited for autonomous-vehicle costs falling ~80% over the next decade. His conclusion on the equity expressing the theme is a published rejection: at a 201.2x forward PE (101.6x even on 2028 EPS) the reverse DCF requires 54.5% annual free-cash-flow growth for ten years — "We're not buying Tesla… There are way better companies available on the stock market today." CNBC Halftime — Kevin Simpson, a Tesla holder (2026-SEP-04): the Cyber Cab event “turned out to be a big fizzle” and the stock fell 6½% with NHTSA evaluating the rollout. His reading is that the demonstration is the fundamental for a promise-valued business — “you’re betting on robotics, you’re betting on autonomous driving… all this does is delay the reality of that coming to fruition, if it ever does” — plus two constraints: “the technology that they use is different than what we see with Waymo,” so a rival’s progress does not validate it, and the settling test is commercial — “at some point you’re going to want to see these things out there with paying customers.” 2026-SEP-07 (RiskReversal — Nathan & Adami): Tesla's invite-only Austin cyber-cab (two-passenger robotaxi) event produced the classic fade — the gap from ~370 to 300 filled into it, stock −6% on the day. Nathan on the other leg: “Optimus is just dumb because humanoid robots are just not going to be a thing… they're going to be in factories… in health tech,” and take the over on Musk's timeline. Adami declines a view but notes a series of lower highs and lower lows since the ~$500 December all-time high, on a good tape. 2026-SEP-08 (Joseph Carlson, After Hours): the post-Cybertruck "Uber is cooked" consensus (Ross Gerber: "on a whole, I think Uber's cooked") gets a cost-structure rebuttal rather than a valuation one — highly variable demand. A fleet owner must "either decide between handling the normalized demand… and then they don't have enough vehicles for the peak demand, or they have to overinvest and buy too many vehicles just to handle the times of peak demand," idle the rest of the day; Uber's drivers supply the rush-hour spike from cars already in their garages, at no capital cost. Hence the hybrid human+AV network is the endgame, not a transitional stage, and Uber already runs both halves (driverless vehicles on the network today, "more and more of them") against Tesla's 40 on the road in one Texas area. He rates the launch's boringness as the real achievement — "drive a car like a normal person, but it was a robot." Steve Eisman (2026-SEP-11): Tesla's robotaxi event showed "not much new" (stock -6%); at 220x 2026 EPS vs GM's 6.5x, "the future of this company depends on the success of the robotaxi," from a CEO with "a history of over-promising and under-delivering" - "count me a skeptic." 2026-SEP-07 (Ryan Tunis, Cantor P&C analyst, Eisman Playbook): ADAS is already cutting auto-insurance exposure before full autonomy — driver-assist is on ~40% of new vehicle sales (15% three or four years ago), driving “pretty sharp declines” in collision frequency and shrinking the ~$350B personal-auto premium pool: “it's here today.” Auto insurers are over-earning into a once-in-a-generation soft market, likely through end-2027. Joseph Carlson (Sep-5 / Sep-15, 2026): Uber completes Waymo's weekly paid trips in ~13 minutes and the gap is widening; Uber aggregates autonomy (Wayve robotaxis, Zipline drones) while DoorDash builds in-house. Waymo will "eat away at the edges"; Tesla's certification "will take years and years." George Noble on The Real Eisman Playbook Ep 76, 2026-SEP-21: scaling Tesla's geofenced Austin robotaxi rides per million people to 350M Americans (Gordon Johnson's data) values the business at "800 million not 800 billion," though most of Tesla's valuation rests on it. Optimus is "a complete freaking joke."