2026-06-18 — views

Physical AI Geographic Expansion — Waymo $10M-$30M Per-City Entry vs Tesla Mapless Nationwide FSD: The Coverage Race

Waymo spends an estimated $10M-$30M and 12-36 months to enter each new city. Tesla Cybercab needs only a driverless permit — no HD maps, no dedicated depot.

Overview

Geographic coverage determines how many potential customers each autonomous vehicle company can serve and how quickly the business scales. Waymo must invest heavily in each new city — HD mapping, depot infrastructure, regulator engagement, fleet logistics — before earning a single dollar there. Tesla’s FSD-based approach could theoretically enable Cybercab to operate anywhere FSD works without city-specific infrastructure investment.

This article benchmarks the two geographic scaling strategies across cost, speed, and risk dimensions. This is article 160 in the Physical AI Benchmark Series.

Section 1 — Waymo’s City-Entry Playbook: What It Costs to Open a New Market

Every Waymo city launch requires a coordinated multi-million-dollar investment that begins months or years before the first commercial ride. The requirements are not optional: regulators demand operating permits, HD maps are required for the sensor stack to function safely, and a local depot is needed to charge, clean, and maintain the fleet.

Entry requirement	Cost / time estimate	Notes
HD map creation	3–6 months; est. $1M–$5M per city (est.)	Every lane, curb, crosswalk, and traffic signal mapped at centimeter accuracy; must be refreshed as roads change
Depot establishment	Est. $2M–$10M per depot (est.); 1–3 depots per city	Vehicle storage, charging, cleaning, maintenance, sensor calibration; lease or build
Regulator engagement	6–24 months; requires permit applications, safety reviews, public comment periods	Each state and city has a different process; CA is the slowest and most rigorous; AZ/TX faster
Remote operations center	Est. $500K–$3M buildout (est.); ongoing staffing	City-specific maps must be loaded; remote operators must be familiar with local scenarios
Fleet logistics setup	3–6 months; vehicle shipping, inspection, sensor install	Gen 6 vehicles arrive from Zeekr (China) with US sensor stack installation
Total per-city entry cost (est.)	$10M–$30M+ per new city (est.)	Does not include vehicle cost; covers infrastructure and operational setup; highly variable by city size
Time to first commercial ride (est.)	12–36 months from decision to launch (est.)	Mapping → permit → depot → fleet → driverless validation → commercial service
Key bottleneck	Regulatory approval timeline is the longest and least controllable step	CA and Northeast US markets likely slowest; Sunbelt/TX-type markets faster

Key insight: HD mapping and depot buildout can proceed in parallel with permit applications, but commercial service cannot begin until the driverless permit is granted. Regulatory approval is the critical path.

Section 2 — Waymo’s Current and Announced City Pipeline

City	Status	Notable facts
Phoenix (AZ)	Commercial driverless since 2020; Waymo’s most mature market	Largest fleet, highest rides per week outside SF; flat terrain, good weather; permissive regulation
San Francisco (CA)	Commercial driverless since 2023 (CPUC permit); most visible market	Complex urban terrain, fog, hills; 24/7 operation; highest media coverage; CPUC oversight most rigorous
Los Angeles (CA)	Commercial driverless expanding; Santa Monica, WeHo, downtown	Second California market; good weather; freeway complexity; expanding ride corridors
Austin (TX)	Commercial driverless launched 2025	Coincides with Tesla’s supervised Robotaxi launch; Texas permissive framework; warm weather
Atlanta (GA)	Announced expansion; timeline est. 2026–2027	5th city; first non-Sunbelt new city; rain, humidity, variable road conditions
Nashville / Miami / others (est.)	Rumored or speculated next markets (est.)	Both have permissive AV frameworks and growing tech populations; not officially announced as of mid-2026
International: Tokyo, Japan	Waymo announced Japan partnership (Nihon Kotsu taxi) 2024; limited scope	Unique road rules, left-hand traffic, narrow streets; significant mapping challenge
International: UK / EU	Not announced; UK AV Act 2024 creates legal pathway	EU AI Act compliance burden; UK faster pathway; neither market has announced Waymo timeline
Coverage ceiling (est.)	At current pace: est. 8–12 US cities by 2028 (est.)	Each city requires 12–36 months; capital-constrained by per-city investment

Section 3 — Tesla’s Nationwide Approach: FSD as the Geographic Unlock

Tesla’s FSD architecture removes the two largest per-city cost drivers — HD map creation and dedicated depot infrastructure — from the geographic expansion equation.

Geographic dimension	Tesla’s approach	Advantage	Risk
FSD geography today (supervised)	Supervised FSD available in all 50 US states, Canada, and early EU markets; no geographic restriction once software is released	No city-specific infrastructure needed for supervised FSD; nationwide from day one	Supervised is not driverless; commercial robotaxi requires driverless permits which are state/city-specific
Cybercab driverless geography	Requires driverless permit in each state/city — same city-by-city regulatory process as Waymo; but Tesla does not need HD maps or per-city depot for the AV capability	If driverless permits are granted, Tesla can deploy Cybercab in a new city faster than Waymo (no mapping/depot buildout required)	Driverless permits are still city-by-city; geographic advantage only materializes if permits come through
No HD map dependency	Tesla FSD navigates from real-time camera perception; no pre-built HD map required; can drive anywhere cameras can see	Enters new geographic areas without map creation cost or time	Real-time perception is harder than map-assisted navigation; edge cases in unmapped areas are more dangerous
No per-city depot required (for AV software)	Tesla’s AV software is OTA-delivered; no city-specific software infrastructure; Cybercab can be serviced at existing Tesla Service Centers	Tesla has 1,000+ Service Centers globally; potential Cybercab maintenance network already exists	Cybercab fleet management (dispatch, cleaning, charging) still requires local infrastructure; not zero per-city cost
Charging network geographic match	Supercharger network is in every major US city already; Cybercab can use existing Superchargers	No charging infrastructure investment required in new cities	Superchargers designed for consumer vehicles; fleet duty cycle may require dedicated stalls or priority access
Geographic scaling speed (est.)	If driverless permits are granted in a new city, Tesla could deploy Cybercab within weeks (vehicle logistics + permit) vs Waymo’s 12–36 months	Tesla’s geographic scaling is permit-limited, not infrastructure-limited	The “if permits” assumption is the entire caveat; without driverless permits, speed advantage is theoretical

Section 4 — City-by-City Cost and Speed Comparison

Dimension	Waymo new city entry	Tesla Cybercab new city entry	Edge
HD mapping cost	$1M–$5M (est.) + ongoing refresh	$0 (no HD maps)	Tesla
Depot infrastructure	$2M–$10M (est.) per depot	Partial — fleet management hub needed but can use Tesla Service Centers	Tesla (lower per-city cost)
Remote ops center	$500K–$3M buildout (est.)	Lower — if FSD requires minimal remote ops; still needs some oversight	Tesla (if FSD self-sufficiency holds)
Regulatory timeline	12–36 months (est.)	12–36 months for driverless permits; same city-by-city regulatory process	Even
Total per-city infrastructure cost (est.)	$10M–$30M+ (est.)	$1M–$5M (est., primarily fleet logistics + local ops hub)	Tesla decisive
Time to first driverless ride (est.)	12–36 months (est.)	3–12 months (est.) if permit granted; faster because no mapping/depot buildout	Tesla (if permit granted)
Cities achievable by 2028 (est.)	8–12 US cities (est.) at current pace	Nationwide if driverless permits granted; 5–15 cities if selective strategy (est.)	Tesla (potential scale)
Maximum addressable geography	Limited to HD-mapped and permitted cities; each new city is a new investment	FSD-capable cities = all cities; driverless-permitted cities = subset growing over time	Tesla (structural scalability)

Section 5 — Geographic Expansion Benchmark Scorecard

Dimension	Waymo	Tesla Cybercab	Edge	2028 outlook
Cities with active commercial service today	4 driverless commercial cities	0 driverless commercial; 1 supervised (Austin)	Waymo decisive	Waymo adds 1–2 cities per year; Tesla builds permit portfolio
Per-city entry cost	$10M–$30M+ (est.)	$1M–$5M (est.) if no HD map/depot	Tesla	Tesla’s mapless approach is a permanent per-city cost advantage
Per-city entry time	12–36 months (est.)	3–12 months (est.) once permit granted	Tesla	Tesla’s speed advantage is real if regulatory is the only bottleneck
Geographic ceiling (US)	Approx. 20–30 cities plausible by 2030 (est.)	Nationwide if permits scale	Tesla (structural ceiling is higher)
International expansion	Japan partnership announced; UK/EU possible 2028–2030 (est.)	FSD in US/Canada/early EU; Cybercab international requires local permits and RHD variants	Even (both face international regulatory complexity)

Overall verdict: Waymo’s geographic expansion is capital-intensive, time-intensive, and methodical — each new city requires an estimated $10M–$30M investment and 12–36 months. This limits Waymo to tens of cities by 2030. Tesla’s mapless FSD approach, if driverless permits scale, could enable Cybercab deployment in hundreds of cities for a fraction of Waymo’s per-city cost. The geographic race will be decided by one question: how fast can Tesla accumulate driverless commercial permits? If permits scale linearly with driverless safety miles accumulated — as Waymo’s experience suggests — Tesla’s first driverless permit in each city is still 12–36 months away regardless of FSD software quality. If a policy shift toward lighter-touch driverless permitting occurs nationally (e.g., the AV START Act passes), Tesla’s geographic advantage could materialize rapidly.

All figures labeled (est.) are derived from public company disclosures, analyst estimates, and industry benchmarks. Neither Waymo nor Tesla has published an official per-city expansion cost breakdown. This article is part of the Physical AI Benchmark Series — article 160.