69 Autonomous Vehicle Fleet Staging, Charging and PUDO Statistics

How many rides and miles do US robotaxi fleets deliver?

Company snapshots, cumulative exposure and quarterly reporting describe different populations. Keep the geography, date and service program attached to any reuse; a city count does not count depots.

1. AV-001Reported measurement

   Waymo reported more than 500,000 trips per week in September 2026.

   Waymo · 2026-09-14 · Source

   Scope and limit: Dated current rate in a Tokyo expansion announcement; Tokyo launch remains future.

2. AV-002Reported measurement

   Waymo reported commercial fully autonomous trips across 15 major U.S. cities in September 2026.

   Waymo · 2026-09-14 · Source

   Scope and limit: City count is not a depot count or a claim of citywide coverage.

3. AV-003Reported measurement

   Waymo reported a fleet of 3,000 vehicles on February 17, 2026.

   Waymo · 2026-02-17 · Source

   Scope and limit: Dated snapshot; does not establish the fleet count in October 2026.

4. AV-004Reported measurement

   Waymo vehicles drove more than 4 million miles per week in February 2026.

   Waymo · 2026-02 · Source

   Scope and limit: No separate passenger-occupied mileage is supplied in this statement.

5. AV-005Reported measurement

   Waymo delivered 15 million rides in 2025.

   Waymo · 2025 · Source

   Scope and limit: Rides are reported by the operator; no depot utilization denominator accompanies the total.

6. AV-006Reported measurement

   Waymo reported 271.3 million cumulative rider-only miles through June 2026.

   Waymo · Cumulative through 2026-06 · Source

   Scope and limit: Rider-only means no human driver; includes unoccupied miles. This is the dashboard coverage, not every Waymo market. Mileage files unchanged in Oct. 2 correction.

7. AV-007Original calculation

   Waymo completed 4,220,075 California driverless deployment trips in Q2 2026.

   Waymo filed with CPUC · 2026 Q2 (April-June) · Source

   Scope and limit: Sum TotalTrips for Apr-Jun in PSG0038152_2026_08_AV_Month_Part0-Deployment.csv; excludes pilot and non-California operations.

8. AV-008Original calculation

   Waymo’s California driverless deployment recorded 28,143,284.70 vehicle miles in Q2 2026.

   Waymo filed with CPUC · 2026 Q2 (April-June) · Source

   Scope and limit: Sum Period 1+2+3 VMT; equals TotalVMTZEV. Includes empty driving.

How much robotaxi mileage is driven without passengers?

Period 1 covers repositioning and waiting for requests; Period 2 covers driving to confirmed pickups. Intertrip time also includes activities beyond parking, and passengers per trip does not establish unrelated-rider pooling. The periods differ, so this comparison does not isolate an improvement caused by staging.

1. AV-010Original calculation

   41.09% of Waymo’s California Q2 2026 driverless deployment miles were in unoccupied Periods 1 and 2.

   Waymo filed with CPUC · 2026 Q2 (April-June) · Source

   Scope and limit: (P1+P2)/(P1+P2+P3). P1: no passengers and not en route to pickup; P2: en route to pickup. Does not measure parked idle time or uptime.

2. AV-011Reported measurement

   Approximately 53.6% of total fleet VMT had a passenger onboard over the full study period.

   Awad Abdelhalim / Findings (analysis of Waymo reports to CPUC) · 2023-08 through 2025-12 · Source

   Scope and limit: California only; passenger-onboard share is source-reported and approximate. Deadheading is separately defined as repositioning/awaiting requests plus travel to pickup.

3. AV-012Reported measurement

   Empty mileage remained roughly 43-45% of total VMT after passenger-onboard share plateaued at 55-57% from mid-2025.

   Awad Abdelhalim / Findings (analysis of Waymo reports to CPUC) · 2023-08 through 2025-12 · Source

   Scope and limit: Later-period plateau, not the full-period average; geography and service changed during the period.

4. AV-013Reported measurement

   Deadheading per completed trip declined from approximately 5.1 miles in August 2023 to 2.8 miles at the end of 2025.

   Awad Abdelhalim / Findings (analysis of Waymo reports to CPUC) · 2023-08 through 2025-12 · Source

   Scope and limit: Approximate graph-discussion values; declining miles per trip does not establish a staging policy caused the improvement.

5. AV-014Reported measurement

   Time between one passenger trip ending and the next beginning averaged approximately 18 minutes by end-2025; full-period average was 28.6 minutes.

   Awad Abdelhalim / Findings (analysis of Waymo reports to CPUC) · 2023-08 through 2025-12 · Source

   Scope and limit: Intertrip time includes varied fleet activities; it is not reported parked time or passenger wait time.

6. AV-015Reported measurement

   Passenger occupancy averaged about 1.4 people per completed trip throughout the study period.

   Awad Abdelhalim / Findings (analysis of Waymo reports to CPUC) · 2023-08 through 2025-12 · Source

   Scope and limit: Passengers per trip are not proof that unrelated riders pooled; a single booking may contain multiple passengers.

How do curb-staging rules change fleet travel and waiting?

These are hypothetical operations with synthetic demand, not measured citywide traffic reductions. Detailed dwell and productivity values below come from the authors’ 2025 preprint; the official 2026 abstract verifies the main travel results. Free-NR allows staging on free, non-residential curbs; demand staging keeps vehicles near expected requests. The model does not simulate traffic-flow dynamics.

1. AV-017Simulation

   A curb-staging ban raised modeled daily fleet travel by nearly 60%, or more than 200,000 km, relative to curb-staging scenarios.

   Younghun Bahk, Michael Hyland, Susan Shaheen, Brooke Wolfe, Adam Cohen / UC Irvine, UC Berkeley, Ajou University · 2023 demand inputs; hypothetical 2026-2027 service · Source

   Scope and limit: Scenario-specific; does not prove XoomPark off-street staging reduces congestion or fleet costs by 60%. Autofleet does not simulate traffic flow.

2. AV-018Simulation

   Excluding residential and metered curbs raised empty VKT by 5.4%; corresponding total VKT increase was about 2%.

   Younghun Bahk, Michael Hyland, Susan Shaheen, Brooke Wolfe, Adam Cohen / UC Irvine, UC Berkeley, Ajou University · 2023 demand inputs; hypothetical 2026-2027 service · Source

   Scope and limit: Relative change in empty distance, not percentage-point change in empty-mile share; 2% total-VKT detail is from the preprint.

3. AV-019Simulation

   Targeted curb restrictions increased curb-staging productivity from 7.4 to 8.0 served requests per staging hour.

   Younghun Bahk, Michael Hyland, Susan Shaheen, Brooke Wolfe, Adam Cohen / UC Irvine, UC Berkeley, Ajou University · 2023 demand inputs; hypothetical 2026-2027 service · Source

   Scope and limit: Productivity denominator excludes pickup/drop-off dwell; hypothetical baseline fleet and requests.

4. AV-020Simulation

   Free-NR demand staging produced a 15.6-minute mean event duration and a 4.3-minute median.

   Younghun Bahk, Michael Hyland, Susan Shaheen, Brooke Wolfe, Adam Cohen / UC Irvine, UC Berkeley, Ajou University · 2023 demand inputs; hypothetical 2026-2027 service · Source

   Scope and limit: Event duration differs from total staging minutes per served request; supply shaping pulls idle vehicles to lots.

5. AV-021Simulation

   Only 2.0% of Free-NR demand-staging events exceeded two hours.

   Younghun Bahk, Michael Hyland, Susan Shaheen, Brooke Wolfe, Adam Cohen / UC Irvine, UC Berkeley, Ajou University · 2023 demand inputs; hypothetical 2026-2027 service · Source

   Scope and limit: No enforced maximum dwell; result depends on supply shaping and specific demand distribution.

6. AV-022Simulation

   Prohibiting curb staging overnight increased modeled daily VKT by 2.2%.

   Younghun Bahk, Michael Hyland, Susan Shaheen, Brooke Wolfe, Adam Cohen / UC Irvine, UC Berkeley, Ajou University · 2023 demand inputs; hypothetical 2026-2027 service · Source

   Scope and limit: Comparison permits up to 50% of fleet to stage at curbs during 00:00-06:00; all-day and day/evening Free-NR scenarios.

7. AV-023Simulation

   Mean customer wait was 3.1 minutes for circling-only and 3.4 minutes for Free-NR demand staging.

   Younghun Bahk, Michael Hyland, Susan Shaheen, Brooke Wolfe, Adam Cohen / UC Irvine, UC Berkeley, Ajou University · 2023 demand inputs; hypothetical 2026-2027 service · Source

   Scope and limit: Wait advantage for circling came with much more VKT; modeled customer acceptance is an assumption.

8. AV-024Announcement/plan

   40 dedicated Caltrain station parking spaces planned for staged Waymo vehicles

   Waymo · 2026-09 announcement · Source

   Scope and limit: Source says 'will lease'; it does not confirm all 40 spaces were active on October 6, 2026.

How do off-street parking strategies affect AV fleets?

These scenarios use an Austin parking inventory from 2021 and include extensive free parking. Outcomes change with fleet size and search strategy; modeled fees do not represent current commercial depot lease prices. Argonne National Laboratory developed POLARIS, which the UT Austin researchers used for the analysis.

1. AV-025Simulation

   For 5,000 SAVs, mandatory nearest-available parking increased mean rider wait from 4.7 to 6.3 minutes.

   Fatemeh Fakhrmoosavi, Krishna M. Gurumurthy, Kara M. Kockelman, Christian B. Hunter, Matthew D. Dean / UT Austin and Argonne National Laboratory · 2021 parking input; simulation published 2024 · Source

   Scope and limit: Compared with idling in place; modeled Austin network; smaller fleet scenario.

2. AV-026Simulation

   For 15,000 SAVs, mandatory parking increased empty VMT share from 11.9% to 14.4%.

   Fatemeh Fakhrmoosavi, Krishna M. Gurumurthy, Kara M. Kockelman, Christian B. Hunter, Matthew D. Dean / UT Austin and Argonne National Laboratory · 2021 parking input; simulation published 2024 · Source

   Scope and limit: Larger fleet result; 5,000-vehicle scenario moves in the opposite direction, so no universal increase.

3. AV-027Simulation

   Parking charges averaged $1.52-$2.28 per SAV/day with least-cost search, versus $2.31-$3.49 with nearest-space search.

   Fatemeh Fakhrmoosavi, Krishna M. Gurumurthy, Kara M. Kockelman, Christian B. Hunter, Matthew D. Dean / UT Austin and Argonne National Laboratory · 2021 parking input; simulation published 2024 · Source

   Scope and limit: Austin supplied extensive free parking; excludes costs beyond modeled parking fees and travel; not market pricing.

4. AV-028Simulation

   SAVs made approximately 15.63-18.86 parking stops per vehicle per modeled day.

   Fatemeh Fakhrmoosavi, Krishna M. Gurumurthy, Kara M. Kockelman, Christian B. Hunter, Matthew D. Dean / UT Austin and Argonne National Laboratory · 2021 parking input; simulation published 2024 · Source

   Scope and limit: Range across fleet sizes and search strategies; parking stops can occur between passenger assignments.

5. AV-029Simulation

   SAVs parked about five hours per day on average, with roughly half-hour parking events.

   Fatemeh Fakhrmoosavi, Krishna M. Gurumurthy, Kara M. Kockelman, Christian B. Hunter, Matthew D. Dean / UT Austin and Argonne National Laboratory · 2021 parking input; simulation published 2024 · Source

   Scope and limit: Narrative result for off-street parking; idle time and time actually parked are distinct metrics.

6. AV-030Simulation

   The paper reports about 22% higher parking fees for private vehicles when SAVs use designated parking.

   Fatemeh Fakhrmoosavi, Krishna M. Gurumurthy, Kara M. Kockelman, Christian B. Hunter, Matthew D. Dean / UT Austin and Argonne National Laboratory · 2021 parking input; simulation published 2024 · Source

   Scope and limit: Percentage summarized in abstract/conclusion; actual table changes vary by fleet size. Abundant free parking keeps absolute fees small.

How does PUDO spacing affect curb-space demand?

The space total sums capacity sized to each location’s own busiest hour; it is not a simultaneous network peak. The calculation assumes 30-second stops and excludes explicit curb queues, irregular arrivals and accessibility dwell. Results use 4,000 SAVs at $1 per mile and older network, population and travel inputs.

1. AV-031Simulation

   Moving PUDOs from every block to every third block reduced estimated total required curb spaces from 263 to 160 (39%).

   Christian B. Hunter, Kara M. Kockelman, Shadi Djavadian / UT Austin and Ford · 2015-2021 inputs; first online 2023, journal issue 2024 · Source

   Scope and limit: Each site is sized to its own busiest hour, with assumed 30-second dwell and perfectly spaced arrivals; this is a capacity illustration, not observed demand or a simultaneous network peak.

2. AV-032Simulation

   Three-block spacing required multiple spaces at 75% of PUDO sites, versus about 36% with one-block spacing.

   Christian B. Hunter, Kara M. Kockelman, Shadi Djavadian / UT Austin and Ford · 2015-2021 inputs; first online 2023, journal issue 2024 · Source

   Scope and limit: Fewer sites concentrate demand; based on the same illustrative 30-second dwell capacity assumption.

3. AV-033Simulation

   One-block to three-block PUDO spacing reduced modeled daily SAV trips from 151,590 to 141,384 (about 6.7%).

   Christian B. Hunter, Kara M. Kockelman, Shadi Djavadian / UT Austin and Ford · 2015-2021 inputs; first online 2023, journal issue 2024 · Source

   Scope and limit: 4,000 SAVs at $1/mile; sensitive mode-choice model; authors summarize this as roughly 6% in abstract.

4. AV-034Simulation

   A 30-second assumed passenger stop implies a theoretical capacity of 120 vehicle stops/hour per PUDO space.

   Christian B. Hunter, Kara M. Kockelman, Shadi Djavadian / UT Austin and Ford · 2015-2021 inputs; first online 2023, journal issue 2024 · Source

   Scope and limit: A modeling assumption and idealized throughput calculation; ignores irregular arrivals, queues and accessibility needs.

5. AV-035Simulation

   At 4,000 SAVs and $1/mile, average modeled VMT per SAV fell from 86 to 84 miles/day with one-block to three-block PUDOs.

   Christian B. Hunter, Kara M. Kockelman, Shadi Djavadian / UT Austin and Ford · 2015-2021 inputs; first online 2023, journal issue 2024 · Source

   Scope and limit: Demand also fell; lower VMT does not isolate efficiency improvement. Author manuscript Table 2.

What facility sizes and production capacities support fleet growth?

The records below distinguish service depots from vehicle assembly and integration plants. Manufacturing floor area, full-scale production potential and actual deployed fleet size have different meanings; announced output is not achieved output. Factory-to-first-pickup timing also differs from cleaning, charging or repair turnaround.

1. AV-036Reported measurement

   Flexdrive’s new Nashville AV depot covers 80,000 square feet.

   Lyft / Flexdrive · 2026 · Source

   Scope and limit: Ribbon cut Oct. 2; Waymo vehicles scheduled Oct. 12, after this asset’s Oct. 6 cutoff.

2. AV-037Announcement/plan

   Flexdrive announced more than 70 new full-time Nashville depot positions for 2026.

   Lyft / Flexdrive · 2026 staffing plan · Source

   Scope and limit: Includes technicians, fleet operations and maintenance roles; announcement does not confirm all hires completed.

3. AV-038Reported measurement

   Waymo and Magna’s Mesa integration plant covers 239,000 square feet.

   Waymo · 2025 · Source

   Scope and limit: Manufacturing and integration floor area; not parking capacity.

4. AV-039Announcement/plan

   Waymo planned to add more than 2,000 autonomous Jaguar I-PACEs through 2026.

   Waymo · 2025-2026 plan · Source

   Scope and limit: Future build plan stated in May 2025; not confirmed production or deployed fleet.

5. AV-040Announcement/plan

   Waymo said its Mesa plant could build tens of thousands of autonomous vehicles per year at full capacity.

   Waymo · Announced 2025; future full capacity · Source

   Scope and limit: Full-capacity potential; actual annual output undisclosed.

6. AV-041Reported measurement

   Phoenix-bound Waymo vehicles could serve their first public passenger less than 30 minutes after leaving the Mesa factory.

   Waymo · 2025 · Source

   Scope and limit: Capability after Driver installation and commissioning; not average assembly or depot turnaround time.

How much charging capacity is being built for AV and light-duty fleets?

These supplier disclosures mix AV-exclusive projects with broader fleet facilities. Ports, chargers, stalls and site megawatts are different quantities, and development assets are not all energized. Terawatt’s Hollywood project is scheduled for 2028; this source does not show operating capacity there today.

1. AV-045Supplier report

   60 charging ports in a parking garage exclusively serving an autonomous vehicle fleet

   Terawatt Infrastructure · Not disclosed; accessed 2026-10-06 · Source

   Scope and limit: Customer, location, energy throughput, utilization and vehicle-to-port ratio are not disclosed. This is a single-site count, not a fleet-wide average.

2. AV-046Announcement/plan

   Voltera’s April 2025 announcement covered 294 stalls and about 28 MW across opened and upcoming light-duty fleet sites

   Voltera · 2025 announcement · Source

   Scope and limit: Announcement combines operational and upcoming sites; do not describe all stalls as live, or all as AV-only. Text gives 28 MW and a bullet says over 28 MW.

3. AV-047Announcement/plan

   79 DC fast-charging stalls planned for Voltera's Miami light-duty depot

   Voltera · 2025 announced plan · Source

   Scope and limit: Upcoming at the release date; this source does not confirm commissioning or name an AV customer.

4. AV-048Supplier disclosure

   20 U.S. properties with total power capacity exceeding 115 MW in Voltera's portfolio

   Voltera · 2025 portfolio snapshot · Source

   Scope and limit: Properties include development assets and several fleet types. This is not an energized AV-only capacity total.

5. AV-049Supplier specification

   180 kW DC fast charging offered at Voltera urban light-duty fleet sites

   Voltera · 2025 offering · Source

   Scope and limit: Advertised charger capability; actual vehicle power depends on acceptance rate, temperature, state of charge and shared site constraints.

6. AV-050Announcement/plan

   More than 1,000 operational and development stalls across 11 U.S. metros expected from the Voltera-Revel combination

   Voltera and Revel · 2026 announced transaction and combined pipeline · Source

   Scope and limit: Contingent on transaction closing in this release. Includes under-development assets and mixed customer uses; not 1,000 live AV depot stalls.

7. AV-051Supplier report

   29 DC fast chargers at Terawatt's Inglewood fleet site, three miles from LAX

   Terawatt Infrastructure · 2024 site opening · Source

   Scope and limit: Opening release does not identify an AV customer or measured utilization. Count is chargers, not necessarily the number of simultaneously chargeable ports.

8. AV-052Announcement/plan

   Planned Hollywood AV charging site: 0.8 acres and 6 MW, with commercial operation scheduled for 2028

   Terawatt Infrastructure · Not disclosed; accessed 2026-10-06; planned operation 2028 · Source

   Scope and limit: A future project, not operating capacity. Power and land figures do not establish charger count, realized charging power or vehicle throughput.

What do charging uptime and reliability statistics actually measure?

Supplier uptime claims require a denominator and observation window before comparisons are meaningful. California’s summarized 97% requirement is a standard for specified public-funded chargers and explicitly excludes exclusively fleet-serving sites. AFDC’s recorded private-fleet inventory is incomplete and is not an AV census.

1. AV-053Supplier report

   Electric Era reported 98.42% per-port uptime for January-November 2024 and a 92.6% 30-day-average successful charge-attempt rate across a 60-port network

   Electric Era · January-November 2024 uptime; reported 30-day SCAR average · Source

   Scope and limit: Uptime is time-based; session figure is a 30-day SCAR average. Source counts payment declines and improper cable insertion. These are separate metrics, and this is not evidence of robotaxi depot performance.

2. AV-054Supplier report

   99%+ uptime claimed for the full charging infrastructure stack

   Terawatt Infrastructure · Not disclosed; accessed 2026-10-06 · Source

   Scope and limit: No denominator, observation window, exclusions or AV-only breakdown is published here. Uptime does not measure charging-session success or AV availability.

3. AV-055Standard

   97% uptime standard for specified publicly funded California DC fast chargers

   California Energy Commission · Current standard summary as of 2026-10-06; applies to covered installations on/after 2024-01-01 · Source

   Scope and limit: CEC's summary specifically excludes exclusively fleet-serving chargers from this uptime category and notes exclusions. Do not present 97% as measured robotaxi depot reliability or interpret legal applicability beyond this summary.

4. AV-056Recorded inventory

   5,900 private light-duty fleet Level 2 charging ports recorded in AFDC's Q2 2024 inventory

   NREL / U.S. Department of Energy Alternative Fuels Data Center · Q2 2024 · Source

   Scope and limit: Figure 17, printed page 25. Private fleet data collection is still expanding; this is a recorded inventory, not a complete national census or an AV-only count.

5. AV-057Recorded inventory

   182 private light-duty fleet DC charging ports recorded in AFDC's Q2 2024 inventory

   NREL / U.S. Department of Energy Alternative Fuels Data Center · Q2 2024 · Source

   Scope and limit: Figure 17, printed page 25. Private collection is incomplete. Do not infer nationwide AV capacity, shortage, or typical charging mix by comparing this count with public-network totals.

6. AV-058Supplier estimate

   1.5+ years to put a charging hub together

   Terawatt Infrastructure · Not disclosed; accessed 2026-10-06 · Source

   Scope and limit: An operator's estimate, without a project sample, median, range or stage-by-stage definition; not a measured industry construction average.

How do local operations and depot location affect fleet economics?

McKinsey’s projections are model outputs, and Terawatt’s savings figures are supplier scenarios. The revenue illustration assumes 30 minutes of travel each way, a $30 ride and one daily charge per vehicle; it is not audited loss or net profit. Remote-assistance staffing excludes a full depot labor accounting. No comparable fleet-wide uptime or maintenance-cost denominator was disclosed in the selected sources.

1. AV-059Supplier/model scenario

   $8.20 modeled total cost per vehicle-mile for a typical US city with 1,000 operating vehicles

   McKinsey Center for Future Mobility · 2025 model baseline (input years not disclosed) · Source

   Scope and limit: Broad modeled AV economics; charging/staging are included but no standalone depot cost or observed fleet uptime is disclosed.

2. AV-060Supplier/model scenario

   About $4.14 local-operations cost per vehicle-mile, roughly 50% of modeled total cost

   McKinsey Center for Future Mobility · 2025 model baseline (input years not disclosed) · Source

   Scope and limit: Broad modeled AV economics; charging/staging are included but no standalone depot cost or observed fleet uptime is disclosed.

3. AV-061Supplier/model scenario

   $1.64 vehicle cost per vehicle-mile, roughly 20% of modeled total cost

   McKinsey Center for Future Mobility · 2025 model baseline (input years not disclosed) · Source

   Scope and limit: Broad modeled AV economics; charging/staging are included but no standalone depot cost or observed fleet uptime is disclosed.

4. AV-062Supplier/model scenario

   $2.40 global-deployment cost per vehicle-mile, roughly 30% of modeled total cost

   McKinsey Center for Future Mobility · 2025 model baseline (input years not disclosed) · Source

   Scope and limit: Broad modeled AV economics; charging/staging are included but no standalone depot cost or observed fleet uptime is disclosed.

5. AV-063Forecast

   About $1.30 potential total cost per vehicle-mile by 2035 under large-scale operations

   McKinsey Center for Future Mobility · 2035 forecast (model published 2025) · Source

   Scope and limit: Broad modeled AV economics; charging/staging are included but no standalone depot cost or observed fleet uptime is disclosed.

6. AV-064Forecast

   About 70% potential reduction in local operational costs by 2035

   McKinsey Center for Future Mobility · 2035 forecast (model published 2025) · Source

   Scope and limit: Broad modeled AV economics; charging/staging are included but no standalone depot cost or observed fleet uptime is disclosed.

7. AV-065Supplier/model scenario

   Better depot location modeled to reduce AV cost per mile by up to $0.10, approximately 6%

   Terawatt Infrastructure · Not disclosed; accessed 2026-10-06 · Source

   Scope and limit: The full model method and validation data are not published. An upper-bound supplier scenario, not an observed savings average or a guarantee for XoomPark sites.

8. AV-066Supplier/model scenario

   Illustrative depot travel scenario: over $20,000 annual revenue loss per vehicle, over $10 million for 500 vehicles

   Terawatt Infrastructure · Not disclosed; accessed 2026-10-06 · Source

   Scope and limit: The supplied assumptions are 30 minutes each way, $30 per ride and one daily charge per vehicle. This is forgone hypothetical revenue, not measured loss, net profit or an audited fleet result.

9. AV-067Reported measurement

   Approximately 70 Waymo Remote Assistance agents were on duty worldwide at any given time in February 2026.

   Waymo · 2026-02-17 · Source

   Scope and limit: Advisory and event-response personnel; excludes a full accounting of depot, cleaning and maintenance labor.

What market estimates and capital commitments support AV infrastructure?

Global Market Insights includes passenger, freight, delivery and industrial operations technology. Fortune Business Insights’ support market includes L2/2+ driver assistance, software/OTA and cybersecurity. Their estimates cannot be added or treated as XoomPark’s addressable property market. Investment commitments and optional debt capacity also differ from cash already spent.

1. AV-068Announcement/plan

   More than $100 million in announced Uber investment in high-capacity AV charging hubs

   Uber Technologies · 2026 announced investment · Source

   Scope and limit: An announced investment program rather than audited capital already spent. This report does not provide site counts or a U.S.-only dollar breakdown.

2. AV-069Announcement/plan

   Terawatt secured a five-year debt facility with $150 million committed plus an optional $150 million increment

   Terawatt Infrastructure, company-issued release distributed by PR Newswire · 2026 financing announcement · Source

   Scope and limit: The $300 million headline is a facility ceiling including an option. Committed funding is $150 million; it is not necessarily debt drawn or infrastructure spending completed.

3. AV-070Supplier disclosure

   More than $150 million invested in Voltera's infrastructure portfolio by April 2025

   Voltera · Cumulative through 2025-04-25 · Source

   Scope and limit: Cumulative provider investment, not per-site cost, an AV-only sum, or a customer lease price.

4. AV-071Market estimate

   $535.8 million estimated global AV fleet-operations market

   Global Market Insights · 2024 · Source

   Scope and limit: Publisher-reported estimate/forecast, not independently reproduced. Scope is broader than US robotaxi staging, charging and PUDO property; these markets cannot be added or used as XoomPark's TAM.

5. AV-072Forecast

   $12.8 billion projected global AV fleet-operations market by 2034; 36.8% forecast CAGR for 2025-2034

   Global Market Insights · 2034 forecast, published 2025 · Source

   Scope and limit: Publisher-reported estimate/forecast, not independently reproduced. Scope is broader than US robotaxi staging, charging and PUDO property; these markets cannot be added or used as XoomPark's TAM.

6. AV-073Market estimate

   $3.33 billion estimated global AV support and maintenance market

   Fortune Business Insights · 2025 · Source

   Scope and limit: Publisher-reported estimate/forecast, not independently reproduced. Scope is broader than US robotaxi staging, charging and PUDO property; these markets cannot be added or used as XoomPark's TAM.

7. AV-074Forecast

   $25.87 billion projected global AV support and maintenance market by 2034; 24.7% forecast CAGR for 2026-2034

   Fortune Business Insights · 2034 forecast, published 2026 · Source

   Scope and limit: Publisher-reported estimate/forecast, not independently reproduced. Scope is broader than US robotaxi staging, charging and PUDO property; these markets cannot be added or used as XoomPark's TAM.

How were these statistics compiled and verified?

The ledger contains 69 cited records across 10 statistical categories, each with a metric, source entity, source URL and data year. Additional fields preserve publication date, evidence type, geography, assumptions and limits so a current article cannot silently turn older inputs or future plans into observed 2026 results.

Original calculations use the monthly California Public Utilities Commission files for April, May and June 2026: sum TotalTrips for activity and the three VMT periods for total distance. The empty-mile share is 100 × (Period 1 + Period 2) ÷ (Period 1 + Period 2 + Period 3), rounded to two decimal places. Pilot and deployment programs remain separate. Empty miles are a distance measure; fleet uptime is a time measure. Neither cumulative rider-only exposure nor intertrip elapsed time is a substitute for occupied mileage or vehicle availability.

Which conclusions would these figures not support?

The nearly 60% VKT increase is a modeled consequence of banning curb staging, not a measured XoomPark benefit. Likewise, the 97% charging requirement and the supplier’s 99%+ infrastructure claim do not establish a robotaxi fleet uptime benchmark.

Use this page as a source-finding and comparison resource. Site sizing, accessibility requirements, power procurement and a commercial return-on-investment model need local demand and operating data. In particular, do not use the idealized PUDO stops-per-hour assumption as a practical service guarantee. For XoomPark’s service context, see the fleet network and Better Cities Alliance. The studies cited here do not evaluate those products.