2026-06-18 — views
Physical AI Charging Infrastructure 2026 — Waymo Depot-Per-City vs Tesla 60K-Supercharger Network: The EV Fleet Infrastructure Benchmark
Tesla has 60,000-plus Supercharger connectors globally. Waymo must build a depot per city. Charging infrastructure is Physical AI's hidden structural moat.
Article 180 in the Physical AI Benchmark Series — EV Fleet Charging Infrastructure
The charging infrastructure dimension of Physical AI is almost never analyzed in robotaxi coverage. Both Waymo and Tesla’s Cybercab operate electric vehicles — but their charging strategies are structurally opposite. Waymo must build a proprietary depot in every new city it enters. Tesla has already built the world’s largest EV charging network: 60,000-plus Supercharger connectors globally, with over 20,000 in the United States. This article benchmarks charging infrastructure as a critical, underanalyzed Physical AI operational and competitive variable.
Section 1 — The Charging Challenge Unique to Driverless Fleets
Human-driven EVs solve the charging problem simply: the human plugs in at home, at the office, or at a public charger. A commercial driverless EV fleet has no human to plug in. Three operating models exist to address this.
Model 1: Depot charging. Vehicles return to a proprietary maintenance and charging base between shifts. Human technicians connect charging cables. This is the current Waymo model. It is operationally proven but requires a physical depot in every city and does not scale to unattended 24/7 operation without human labor.
Model 2: Autonomous charging. The vehicle docks itself at a charging station without human intervention. This model enables truly unattended driverless fleet operation but is not yet commercially deployed at scale anywhere in the industry.
Model 3: Opportunity charging. Vehicles charge briefly at en-route public stations during low-demand periods. This requires available chargers, short charge times, and software to route vehicles to charging opportunities without unacceptably degrading service availability.
The fleet utilization tension: A vehicle that is charging is not earning revenue. Fleet operators must balance charging downtime against vehicle count and charger speed. Charging speed matters enormously for fleet economics:
| Charging level | Power range | Range added in 30 min (est.) | Fleet economics implication |
|---|---|---|---|
| Level 2 AC | 7–11 kW | est. 20–35 miles | Very slow; best for overnight depot charging; unacceptable for in-service opportunity charging |
| DC Fast Charging | 50–150 kW | est. 70–150 miles | Practical for fleet operations; 45–60 min to full from low battery |
| Ultra-Fast DC (V3/V4) | 250–350 kW | est. 150–200 miles in 15–20 min | Enables high-utilization opportunity charging; game-changing for fleet economics |
An EV with a 300-mile range and Level 2 charging requires 8–12 hours to fully charge — fine for overnight depot operations but incompatible with in-service charging. DC Fast Charging at 150 kW can add 100–150 miles in under 30 minutes. Tesla’s V3 Supercharger at 250 kW peak can add an estimated 200 miles in 15 minutes. Fleet operators must balance charger capital expenditure against vehicle downtime cost — and the math changes dramatically with charging speed.
Section 2 — Waymo’s Charging Infrastructure: Depot-Based Model
| Dimension | Detail |
|---|---|
| Current charging model | Waymo vehicles return to proprietary maintenance and charging depots; human technicians connect charging cables; depot locations are not publicly disclosed |
| Vehicle range (est.) | Jaguar I-PACE Gen 5: est. 234 miles EPA range; Gen 6 Zeekr-based vehicle: range not disclosed; effective commercial range est. 150–200 miles per shift accounting for HVAC, sensor compute load, and urban driving cycles (est.) |
| Charging hardware at depots (est.) | Mix of Level 2 and DC Fast Charging at Waymo depots (est.); exact hardware configuration not disclosed; depot charging is more cost-effective per kWh than public fast charging |
| Autonomous charging status | Not deployed commercially; requires a mechanical docking system or robotic arm; Waymo has not announced a commercial autonomous charging solution; human technicians plug in vehicles at depots |
| Depot real estate requirement | Each Waymo depot requires: covered parking stalls for all fleet vehicles, a charging inlet at each stall, maintenance bay space for repairs and sensor calibration, and space for Remote Operations Center (ROC) and dispatch; commercial real estate in San Francisco, Phoenix, and Los Angeles carries significant cost |
| New city charging challenge | Each new city requires: a new depot facility (find, lease or build, equip with charging infrastructure) plus electrical infrastructure upgrades; estimated 6–18 months to stand up a new depot in a new city (est.); this constraint directly limits new city expansion speed |
| Energy cost per mile (est.) | EV energy cost est. $0.03–$0.05/mile at commercial electricity rates (est. $0.08–$0.12/kWh commercial rate, est. 3–4 miles/kWh EV efficiency) — significantly lower than ICE vehicle fuel costs |
| Charging as operational bottleneck | If depot charging stall count is insufficient, fleet utilization is capped regardless of vehicle count; adding charging capacity in an older commercial building may require expensive electrical panel upgrades |
The depot model’s structural constraint: Waymo’s charging infrastructure cost and build time is not a one-time investment — it repeats with each new city. Every market entry requires finding a suitable commercial facility, negotiating a lease or construction, procuring and installing charging hardware, and arranging electrical infrastructure upgrades. This is a real, significant constraint on expansion velocity.
Section 3 — Tesla Supercharger: Structural Charging Advantage for Cybercab
| Dimension | Detail |
|---|---|
| Supercharger network scale | 60,000-plus Supercharger connectors globally (est. mid-2026); 20,000-plus in the United States; Tesla has been expanding at est. 30–40% per year |
| Supercharger speed | V3 Supercharger: up to 250 kW peak charging speed; adds est. 200 miles range in 15 minutes; V4 Supercharger: up to 350 kW — the fastest in the Tesla network |
| Coverage density | Superchargers placed along major US highways and in urban centers; designed initially for inter-city travel but increasingly deployed in dense urban areas where Cybercab would operate |
| Cybercab charging model (est.) | Cybercab could use the existing Supercharger network for opportunity charging during low-demand periods; no depot required if Supercharger density is sufficient in a given market; Tesla could also build dedicated Cybercab charging facilities co-located with Supercharger stations (est.) |
| Autonomous Supercharger docking (est.) | Tesla demonstrated a robotic charging arm (Snake Charger prototype, first shown 2015); not commercially deployed; if Cybercab docks autonomously at Superchargers, it eliminates the need for a human to plug in — enabling truly unattended Cybercab operation (est.) |
| No depot real estate required (est.) | If Cybercab uses the distributed Supercharger network instead of proprietary depots, Tesla avoids the per-city depot real estate and electrical infrastructure cost that constrains Waymo’s expansion velocity (est.) |
| Energy cost for Cybercab (est.) | Supercharger commercial rate varies by market; at est. $0.25–$0.35/kWh Supercharger rate and est. 4 miles/kWh Cybercab efficiency: est. $0.06–$0.09/mile energy cost (est.); higher than depot commercial electricity rate but avoids depot capital expenditure |
| V2G potential (vehicle-to-grid) | Tesla has V2G capability in some markets; a Cybercab fleet could potentially sell energy back to the grid during peak demand and buy during off-peak — a fleet energy arbitrage opportunity not available to competitors without Tesla’s grid integration |
| Supercharger as competitive moat | Tesla’s Supercharger network cost est. $5B-plus to build (est.); competitors (Waymo, Zoox, Cruise) would need to build their own depot charging or contract with third-party fast charging networks (Electrify America, EVgo, ChargePoint) — none of which match Supercharger on speed, density, or software integration |
The moat dimension: Building the Supercharger network took Tesla more than a decade and est. $5B-plus in capital expenditure. A competitor entering the robotaxi market today cannot replicate this asset on any reasonable timeline. Waymo’s alternative — contracting with Electrify America or EVgo for third-party fast charging — provides public DC Fast Charging access but without the software integration, coverage density, or reliability of Supercharger.
Section 4 — Autonomous Charging: The Unsolved Driverless Fleet Problem
Autonomous charging — a vehicle docking itself to a charger without human assistance — is the enabling technology for truly unattended 24/7 driverless fleet operation. Without it, both Waymo and Cybercab require human labor at some point in the charging loop.
| Dimension | Waymo approach | Tesla approach | Industry status |
|---|---|---|---|
| Current solution | Human technicians plug in at depots | Human plugs in at Supercharger (for Model 3/Y); Cybercab autonomous charging solution not disclosed | No commercial autonomous charging deployed at fleet scale anywhere in the industry |
| Autonomous docking requirement | For 24/7 unattended driverless fleet operation, vehicles must charge without human intervention; depot technicians represent labor cost that scales linearly with fleet size | Tesla Snake Charger (robotic arm prototype, 2015); not commercially deployed; if deployed for Cybercab, enables unattended autonomous charging | CCS and NACS inlets are not designed for robotic docking; enabling robotic charging requires either hardware modification or a new connector standard |
| Waymo’s autonomous charging path (est.) | Waymo could add autonomous docking hardware to Gen 6 vehicles; would require depot infrastructure modification; not publicly announced; human-technician model is manageable at current fleet size but becomes expensive at 10,000-plus vehicle scale (est.) | N/A | Each autonomous charge event that eliminates a human technician action = direct labor cost reduction at scale; the economics improve non-linearly as fleet size grows |
| Wireless charging as alternative | Wireless/inductive charging pads (vehicle parks over pad, charges without physical connection); several companies developing this approach (WiTricity, Momentum Dynamics); not yet deployed at commercial AV fleet scale | Not deployed commercially by Tesla or Waymo | Wireless charging currently est. 85–92% efficient versus wired (est. 95%-plus efficient); efficiency loss at fleet scale represents a meaningful energy cost increase |
| Fleet charging as expansion constraint | Every new Waymo city requires depot charging infrastructure; Cybercab could potentially deploy in any city with adequate Supercharger coverage without a new depot | Tesla’s Supercharger network already covers most US metropolitan areas | This asymmetry could allow Cybercab to expand to new cities substantially faster than Waymo — a city entry speed advantage derived from charging infrastructure |
The labor cost implication: At a fleet of 1,000 vehicles requiring charging across two shifts, depot technicians performing manual plug-in/out operations represent tens of thousands of technician-hours annually. At 10,000 vehicles, this becomes a material operational labor cost. Autonomous charging solves this — but neither Waymo nor Tesla has commercially deployed it.
Section 5 — Charging Infrastructure Benchmark Scorecard
| Dimension | Waymo | Tesla Cybercab (est.) | Edge |
|---|---|---|---|
| Charging network | Proprietary depots; must build per city | 60,000-plus Supercharger connectors globally | Tesla (massive existing infrastructure) |
| Depot capex per new city (est.) | est. $5M–$20M depot build-out per new city (est.) | est. $0 if using existing Supercharger network (est.) | Tesla |
| Autonomous charging | Not deployed; human technicians at depots | Not deployed commercially; Snake Charger prototype exists | Roughly equal — neither has a commercial solution |
| Charging speed | Mix of Level 2 and DC Fast at depots (est.) | V3/V4 Supercharger up to 350 kW | Tesla (faster charging = higher fleet utilization potential) |
| Energy cost per mile (est.) | est. $0.03–$0.05/mile (commercial depot electricity rate) | est. $0.06–$0.09/mile (Supercharger commercial rate) | Waymo (lower per-mile energy cost at depot) |
| New city expansion speed | Slow: est. 6–18 months depot setup per city (est.) | Fast: immediate if Supercharger coverage exists (est.) | Tesla |
| V2G / grid integration | None | Available in select markets | Tesla |
| Long-run charging moat | Depot model scales linearly with fleet size; limited network effect | Supercharger network has strong network effect; each new charger benefits all Tesla and Cybercab vehicles | Tesla |
Overall verdict: Tesla’s Supercharger network is a Physical AI infrastructure asset that cost an estimated $5B-plus to build and is nearly impossible for competitors to replicate on any near-term timeline. If Cybercab can use the existing Supercharger network — especially with autonomous docking — it avoids the per-city depot capital expenditure that constrains Waymo’s expansion. This could enable Cybercab to enter new cities in weeks rather than the estimated 6–18 months Waymo needs to establish depot infrastructure.
Waymo’s lower per-mile energy cost at commercial depot rates is a real but narrow advantage — the difference between $0.03–$0.05/mile and $0.06–$0.09/mile is modest when set against the broader per-trip economics of a robotaxi service. The charging infrastructure dimension is solidly in Tesla’s structural favor as a Physical AI competitive moat.
About This Series
This is article 180 in the Physical AI Benchmark Series. The series covers the companies, technologies, capital, regulation, and infrastructure shaping the deployment of AI-driven physical systems — autonomous vehicles, humanoid robots, and the compute and energy infrastructure that supports them.
The central finding of this article: EV charging infrastructure is an underanalyzed Physical AI structural variable. The asymmetry between Waymo’s depot-per-city requirement and Tesla’s existing 60,000-plus Supercharger network represents a multi-billion-dollar infrastructure moat. Neither company has solved autonomous charging at scale — but Tesla’s distributed charging asset eliminates the per-city depot capex constraint that limits Waymo’s expansion velocity.
Sources
- Tesla Supercharger network — Tesla ↗
- Waymo fleet operations — Waymo Safety Report ↗
- EV charging infrastructure — US DOE Alternative Fuels Data Center ↗
- Tesla Snake Charger prototype — Tesla blog ↗