Design and Investment
Streets, Depots, Buildings, Districts
7.1Design decisions are governance decisions
The previous two chapters described what a city needs to see and what instruments it needs to deploy. Every spatial decision a city takes in the next five years is simultaneously a governance decision, an investment decision, and a path dependencyPath dependencyThe property that early decisions constrain later ones, so that what is cheap to choose now becomes expensive or impossible to change afterwards. The paper’s central mechanism: the terms of the first operating licences become the baseline for every subsequent negotiation. (Ch.1.) decision. Where should charging and idling depots be sited? How should curbs be redesigned? What should be required of new parking construction? Design and governance cannot be thought of independently here. Trying to separate them is one reason early deployment elsewhere has already created avoidable lock-in.
This matters because design operates at a fundamentally different timescale from rules and pricing. A licensing condition can be revised annually if the evidence demands it. A congestion charge can be recalibrated quarterly in response to traffic data. But a depot built in the wrong location, a parking structure designed without conversion capability, a street corridor locked into a geometry that cannot accommodate future use changes – these decisions persist for decades. They compound across planning cycles. They become the constraints within which every subsequent governance decision must operate. The spatial layer is where path dependency bites hardest, and correspondingly where optionality has the highest value.
The interventions described in this chapter operate at five scales: streets and rights-of-way, curb and public realm, depots and charging infrastructure, buildings and parcels, and districts and corridors. At each scale, the same design principle applies: create optionality, build in reversibility, and treat spatial decisions as governance instruments.
7.2Streets and rights-of-way
CAMCAMConnected and Autonomous Mobility — the commercial deployment of autonomous vehicles operating connected to data and control networks. The paper uses CAM rather than "AV" to foreground the network and governance dimensions that determine how the technology lands in cities. (Ch.1 §1.1; Ch.2 §2.1.) changes what streets do. Autonomous vehicles require different operational environments than human-driven traffic: different lane widths for mixed-autonomy operation, different intersection geometries for sensor line-of-sight, different signal timing regimes for platooning and coordinated dispatch. The temptation, when facing this reality, is to redesign streets for the autonomous future – to invest in permanent infrastructure that optimizes the right-of-way for the technology as it is currently understood.
The future of CAM deployment is not sufficiently certain to justify permanent street redesign at this point in the technology’s development. What is certain is that flexibility has value, and that governance infrastructure is a precondition for physical infrastructure to deliver its intended outcomes. The UK created clear, governable pilots. Of the Low Traffic NeighborhoodLTN (Low Traffic Neighbourhood)A UK street-design intervention restricting through-traffic on residential streets. The paper uses the LTN record as its clearest evidence on legitimacy: of those implemented 2020–2024, 18 percent were later reversed, almost always where engagement came after the intervention. (Ch.7.) (LTN) programs implemented between 2020 and 2024, 18% were subsequently removed because the governance failed.EV40EV40Street Pilot Failures and ReversalsAV street pilots cancelled or rolled back globally; regulatory gaps, public trust deficits, and incident liability cited as recurring failure driversOpen in the Evidence Base → Wandsworth’s LTN lasted less than one monthEV40EV40Street Pilot Failures and ReversalsAV street pilots cancelled or rolled back globally; regulatory gaps, public trust deficits, and incident liability cited as recurring failure driversOpen in the Evidence Base →: two agencies made changes simultaneously with no coordination. Redbridge scrapped its program after a monthEV40EV40Street Pilot Failures and ReversalsAV street pilots cancelled or rolled back globally; regulatory gaps, public trust deficits, and incident liability cited as recurring failure driversOpen in the Evidence Base → because no community engagement preceded installation. Ealing’s bollards were uprooted and planters daubed with graffitiEV40EV40Street Pilot Failures and ReversalsAV street pilots cancelled or rolled back globally; regulatory gaps, public trust deficits, and incident liability cited as recurring failure driversOpen in the Evidence Base → because the process had no democratic legitimacy. They were governance failures expressed as physical removals.
San Francisco’s JFK Drive car-free closure ran for over two years as an extended pandemic pilot before the Board of Supervisors approved permanent status by a 7–4 voteEV40EV40Street Pilot Failures and ReversalsAV street pilots cancelled or rolled back globally; regulatory gaps, public trust deficits, and incident liability cited as recurring failure driversOpen in the Evidence Base →. Voters confirmed it with 63% approvalEV40EV40Street Pilot Failures and ReversalsAV street pilots cancelled or rolled back globally; regulatory gaps, public trust deficits, and incident liability cited as recurring failure driversOpen in the Evidence Base → in a subsequent referendum. Park visitation increased 36%EV40EV40Street Pilot Failures and ReversalsAV street pilots cancelled or rolled back globally; regulatory gaps, public trust deficits, and incident liability cited as recurring failure driversOpen in the Evidence Base → over the pilot period. The difference was governance design: extended duration that generated evidence, measured outcomes that informed democratic decision-making, multi-level endorsement that conferred legitimacy, and explicit engagement with accessibility concerns that prevented the program from becoming a proxy dispute about other issues.
The lesson for street interventions focused on CAM is to invest in governance infrastructure before committing to physical infrastructure. That might look like instrumented pilots with defined measurement frameworks, planned durations long enough to generate robust evidence, explicit coordination agreements across the agencies whose decisions affect the same corridor, and democratic endorsement processes that are initiated before installation rather than imposed after opposition emerges. The best way to eliminate uncertainty about which spatial configurations will serve city goals across the full range of CAM deployment scenarios is a structured, reversible pilot.
7.3Curb and public realm
Autonomous fleets enter the curb – the contested strip described in Chapter 3 – as a new and demanding claimant. Fleet pickup and drop-off operations require reliable curb access at predictable locations; goods delivery requires dwell time for vehicle loading and unloading; charging infrastructure requires curb or near-curb access to electrical supply. Without explicit governance of curb allocation, fleet operators will capture the most convenient locations, displacing existing uses and creating the allocation failure mode described in Chapter 4.
Micro-terminals represent the most effective physical intervention for managing fleet curb demand in high-density locations. A micro-terminal is a small, purpose-designed pickup and drop-off facility – two to four vehicle bays – that consolidates fleet activity into defined, managed locations rather than allowing it to distribute opportunistically across the curb. The design requirements are passenger shelter against weather, real-time information on vehicle arrival and waiting times, accessibility compliance for wheelchair users and those with visual impairments, integration with cycling infrastructure to avoid conflict, and sight-line design that allows both vehicle operators and pedestrians to see approaching traffic clearly. The footprint can be as small as one parking bay repurposed, but the governance investment – in location selection, stakeholder agreement, and operational management – is substantial.
Loading consolidation applies the same logic to commercial freight. Consolidated loading points serving multiple addresses from a single defined location reduce the total dwell time and double-parking that autonomous goods delivery would otherwise generate on residential and commercial streets. The evidence from existing freight consolidation programs in London, Utrecht, and Bath and Bristol shows consistent reductions in curb conflict when consolidation is formally required as a licensing condition rather than offered as a voluntary alternative.EV36EV36Freight Access Rule Outcomes: Consolidation Zones, Delivery Restrictions, Time WindowsConsolidation zones, delivery time windows, and access restrictions reshape last-mile freight; compliance burden falls heaviest on SME operatorsOpen in the Evidence Base → City-scale freight consolidation – where operator routing is constrained to use defined loading points in sensitive corridors – requires the operating picture to be in place first.
Accessibility is the central design requirement. Every curb intervention must be fully operable by wheelchair users, those with visual impairments, and people with mobility limitations. Beyond the legal obligation under the UK Equality Act and the ADAADA (Americans with Disabilities Act)US federal civil-rights legislation setting accessibility obligations. Relevant because a fleet with no driver cannot rely on driver assistance to meet them, which makes accessibility a vehicle-design and service-design condition rather than a staffing one. (Ch.7.), it is the condition for public acceptance. The failure to address accessibility concerns was a contributing factor in multiple LTN reversals. JFK Drive’s success, by contrast, included the specific accommodation of mobility-impaired users through designated vehicle access points that preserved the car-free character of the corridor while meeting accessibility requirements.
7.4Depots and charging infrastructure
Fleet depots are the new industrial infrastructure of autonomous mobility. They are large, energy-intensive, operationally demanding facilities whose location has consequences that reach far beyond transport planning into energy infrastructure, land use, neighborhood character, and urban equity. They are also, in most jurisdictions, being sited by private operators without systematic public oversight – creating the spatial and energy lock-in that this chapter argues cities must prevent.
A 100-vehicle autonomous fleet depot requires 45,000 to 50,000 square feet of operational space and 2 megawatts of managed chargingManaged chargingShaping when a fleet charges so that demand falls outside network peaks, in exchange for lower cost or faster connection. The mechanism behind the paper’s "fleet as a battery" framing: a controllable load is an asset to the grid rather than a threat. (Ch.2.) capacity. A 500-vehicle deployment needs 175,000 to 250,000 square feet across potentially multiple sites, with 8 to 10 megawatts of connected charging load.EV96EV96Robotaxi Depot and Grid Infrastructure Constraint Becomes Industry-Acknowledged — 4–12 MW Per Depot, 12–36-Month Grid-Upgrade Lead Times — 29 April 2026Axios establishes 4–12 MW per depot, 12–36-month grid-upgrade lead times as binding pacing constraintOpen in the Evidence Base → In dense urban areas, this space competes directly with housing, logistics, manufacturing, and affordable commercial workspace – all of which have their own political constituencies and planning frameworks. Siting is a land use decision, an energy infrastructure decision, and a neighborhood planning decision at once, and it requires the cross-agency coordination described in Chapter 6 to resolve without creating new conflicts.
The 2-megawatt figure for a 100-vehicle fleet represents peak connected load – the maximum instantaneous demand when all vehicles are charging simultaneously. In practice, average demand across the charging cycle will be lower, shaped by vehicle arrival patterns, battery state-of-charge distribution, and the charging schedule applied. Peak load is the critical number for grid connection sizing and for triggering distribution network upgrade assessments. Average load drives the operating energy cost calculation and the economics of time-of-use tariff management. Both numbers matter, but they serve different planning purposes and confusing them – using average demand to size a grid connection, or peak demand to model energy costs – produces errors that are expensive to discover at the operational stage. Any city or investor underwriting a depot business case should require explicit declaration of both figures from the operator, with supporting modeling that shows the charging load profile across the 24-hour operational cycle. For a city, this belongs in the consent rather than the underwriting: a declared 24-hour load profile, on both peak and average, should be a condition of depot approval, not an analyst’s request.
The design response to the energy failure mode is managed charging by default. Every new depot development should be required to design electrical systems for time-of-use load shaping, with the ability to shift charging demand away from grid peak periods. Vehicle-to-grid readiness – designing vehicle connections to allow bidirectional power flow – keeps open the option of depot fleets contributing to grid balancing services rather than simply drawing from them. No one should underwrite a depot on vehicle-to-grid revenue today. The case for building it in anyway is narrower and firmer: cheap at design time, expensive to add later. The 35 to 45 percent peak reductions reported in Chapter 3 mean that depot electrical design decides whether a fleet worsens or relieves the grid stress it creates.
Depot siting should be driven by three data layers in combination: grid hosting capacityHosting capacityHow much additional load or generation a given point on the electricity network can accept without reinforcement. Publishing it is what turns depot siting from a series of private negotiations into a planning question a city can participate in. (Ch.2.) from the energy domain of the operating picture, land availability and zoning compatibility from the spatial domain, and proximity to service areas from the mobility domain. No single agency currently holds all three, which is why rational depot planning cannot happen without the cross-agency governance wrapper. Cities that establish that wrapper before operators select depot sites will be able to guide siting towards locations where infrastructure costs are lowest, grid impact is manageable, and service area coverage is equitable. Cities that allow operators to select sites unilaterally will then face the far harder task of negotiating after the fact.
7.5Buildings and parcels
Connected autonomous mobility changes the economics of parking – and therefore the economics of every building that contains it. If fleet vehicle services replace private car ownership at scale, the demand for parking collapses. A parking garage in a dense urban center becomes a stranded assetStranded assetAn asset that loses its economic purpose before the end of its physical life. In the paper this is chiefly structured parking: if car ownership falls at scale, a multi-story garage in a dense centre generates no revenue and cannot easily become anything else. (Ch.4.), generating no revenue while its replacement cost sits on the balance sheet. Parking assets are already becoming stranded in US downtowns where office vacancy is highest, and the trajectory will accelerate with autonomous deployment.EV19EV19Parking Structure Utilisation Rates, Current and ProjectedParking utilisation declining as AV adoption rises; peak-hour demand projected to fall 40–60% by 2040, stranding urban garage assetsOpen in the Evidence Base → The question for building owners and investors is not whether parking economics will change but how quickly, and whether their assets can adapt.
Conversion-ready design has a specific technical specification. Designing a parking structure for future conversion to residential, office, or logistics use requires floor-to-floor heights of 10 feet or more (standard parking garages use 7 to 8 feet), column spacing of at least 27 feet to allow flexible floor plan layouts, and load capacity of 65 pounds per square foot rather than the standard 40 (residential and office uses are heavier than parked vehicles). These specifications add approximately 40 percent to the structure’s initial construction cost.EV20EV20Parking Structure Conversion Costs and FeasibilityParking-to-mobility-hub conversions cost $25K–$50K/space; flat floors, structural load, and ventilation are key feasibility gatesOpen in the Evidence Base → But they preserve the option to convert – and in a market where parking demand is structurally declining, that option has asymmetric value. At a 9% discount rate, the premium breaks even only where conversion value is realized within roughly ten years.
One near-term pathway is particularly attractive: demand for last-mile logistics space is growing, and a conversion-ready parking structure in a dense employment center can be adapted to logistics use far more quickly and cheaply than a standard structure. Camden Property’s Lincoln Station in Denver, Hudson Pacific’s EPIC in Los Angeles, and Google’s Alta structure in California all show the commercial viability of conversion-ready specifications in major markets. The governance intervention is a design standard: cities should require conversion-capable specifications for new parking construction in CAM-sensitive locations, creating institutional demand for what is already a commercially sound design approach.
7.6Districts and corridors: where governance becomes investable
The interventions described at building, curb, and depot scale compound at district and corridor scale – and that compounding is where the governed dividend of Chapter 4 becomes a concrete investment proposition rather than a policy objective. A CAM corridor is a street or route where autonomous fleet operations are concentrated enough to generate simultaneous, material effects across all four domains of the Governability StackGovernability StackThe four-layer model the paper uses to organise what cities must govern: Flows (movement), Space (kerb, road, depot), Power (grid and energy), Data (sensing and authority). Each layer has its own institutional surface and timescales. (Ch.3.). Curb demand, grid load, property values, fee revenue, and the distribution of benefits and burdens across adjacent neighborhoods all move together. The corridor is the unit at which these effects can be coordinated, and at which public authority creates shared value instead of only preventing private harm.
The precedent is the Elizabeth Line: 6 percent above-trend property growth within 500 meters of stations, and up to 100 percent uplift over a decade around east London stationsEV22EV22Ch.4 Evidence Notes — The StakesAutonomous mobility governance chapter 4; stakes framing — safety, liability, equity, economic disruption; foundational 'why it matters' lensOpen in the Evidence Base → in regeneration corridors – the direct consequence of better access, shorter journeys, and the investment signal a coordinated public commitment sends. Singapore’s Circle Line created $0.9 billion (SGD 1.23 billion) in housing wealthEV22EV22Ch.4 Evidence Notes — The StakesAutonomous mobility governance chapter 4; stakes framing — safety, liability, equity, economic disruption; foundational 'why it matters' lensOpen in the Evidence Base → for proximate households. A meta-analysis of fifteen studies found a consistent 2.4% property value increase for every 250 meters closerEV22EV22Ch.4 Evidence Notes — The StakesAutonomous mobility governance chapter 4; stakes framing — safety, liability, equity, economic disruption; foundational 'why it matters' lensOpen in the Evidence Base → to a rapid transit station. The mechanism is the same for CAM: coordinated mobility infrastructure generates spatial value premiums. Whether public authority is positioned to capture a share of them for reinvestment is the open question.
Three elements distinguish a governed CAM corridor from an ungoverned one, and it is the combination of all three that generates the full governed dividend. First, value capture mechanisms – tax increment financingTIF (Tax Increment Financing)A value-capture mechanism that borrows against the future property-tax uplift a public investment is expected to create. One of the fiscal instruments available to recycle the land-value effects of CAM infrastructure. (Ch.6.), special assessment districts, developer contribution frameworks, or CIL in the UK context – that recycle property uplift in the corridor back into public infrastructure investment. Without value capture, the fiscal sustainability of corridor governance is entirely dependent on fleet charging revenues and licensing fees, which are more volatile and more politically contested. Value capture provides the stable, long-term revenue base that enables the public commitment that makes the corridor attractive in the first place. Second, equity planning that ensures the corridor serves all residents and businesses, not only those commercially attractive to operators. This requires explicit service obligation conditions in fleet licenses, combined with the operating picture data that makes those conditions auditable. Third, grid coordination that sizes energy infrastructure to the corridor’s full buildout, not merely its first phase. A corridor whose grid infrastructure is designed for 100 vehicles but will eventually host 500 creates exactly the energy failure mode that managed planning is designed to prevent.
A district where autonomous fleet operations, grid infrastructure upgrade, and land use transition are coordinated by public authority creates value across transport, energy, and real estate simultaneously. A district where the same elements proceed through uncoordinated private optimization creates value destruction across all five domains instead. The difference is institutional design – how the interaction between fleets, grid, land, and public realm is governed. For investors, that is the difference between a value creation thesis and a value extraction thesis, and it is within the control of the cities reading this paper.
The design pattern library – reversible street templates, micro-terminal specifications, depot overlay standards, conversion-ready parking specifications, and district corridor coordination frameworks – is available to Field Discovery members.