Hey —
Uber published the operating specifications for its Houston robotaxi depot this week. Four megawatts of power. Forty fast chargers. Fifteen maintenance bays. Fifty thousand square feet. The structural detail buried in the announcement matters more than the geographic expansion the headlines covered.
Three signals worth thinking about.
1. Uber's Houston specs quantify what the operations layer actually costs
On June 17, Uber, Nuro, and Lucid announced Houston as the second planned market for their robotaxi program, launching mid-2027 after the San Francisco Bay Area launch later this year. The headline was the geographic expansion. The structural details deeper in the announcement quantify the operations layer that has been visible thematically but not specifically until now.
The Houston operational specifications: a 50,000-square-foot depot facility plus a dedicated charging pitstop in downtown Houston. The depot will draw more than 4 megawatts of power, house 40 fast chargers, and include 15 maintenance bays. Groundbreaking is planned for early 2027.
The 4-megawatt figure is the data point worth sitting with. For comparison, the Ojai 800V transition piece analyzed that a 36-charger 800V depot operating at peak design load would draw 12.6 megawatts. Uber's Houston depot at 40 chargers and 4 megawatts of contracted capacity suggests a design assumption closer to 100 kW average per charger with significant load management software, rather than the peak 350 kW per charger that 800V architecture supports. That makes sense at scale. Uber is sizing for operating average plus headroom, not peak design ceiling. The math implies disciplined load management as a core operating capability, not an afterthought.
The strategic detail worth naming separately: Uber is building operations capability in-house through Uber Autonomous Solutions, the fleet-management unit launched earlier in 2026. Per Yahoo Finance coverage of the Houston announcement: "Under Uber Autonomous Solutions, the fleet-management unit it unveiled in February, the company has pulled infrastructure and logistics in-house. It is a part of the business that Alphabet's Waymo and Tesla also have to solve."
That is the opposite strategic answer to Waymo's choice. Waymo signed Element Fleet Management for end-to-end operations in San Diego. Uber is building Uber Autonomous Solutions to do operations in-house for its multi-AV partner portfolio (Waymo, WeRide, Wayve, Nuro, Motional, Avride, Rivian, and others). Two strategic answers, both responding to the same structural recognition that operations is its own category.
The question worth tracking over the next 24 months: which strategy wins at scale. Outsourcing to institutional specialists like Element gets validated if institutional players deliver superior operational performance with lower cost of capital. Building in-house gets validated if companies like Uber Autonomous Solutions capture operating margin that would otherwise flow to third-party operators. The Houston numbers tell us Uber is making real capital commitments to the in-house thesis: groundbreaking in less than a year, infrastructure that requires utility interconnection well before that, and operational staffing that needs to be hired and trained before vehicles arrive.
2. Tesla's Texas registrations jumped 64% in a month
The Texas DMV's automated vehicle tracker published updated data on June 21. Compared to the May 28 disclosure that anchored Issue #9's analysis, the registered fleet counts have moved meaningfully.
Waymo grew from 577 to 620 vehicles registered in Texas, a 7.5% increase. Tesla grew from 42 to 69 registered vehicles, a 64% increase in less than a month. Zoox grew from 35 to 43, a 23% increase. Avride, Nuro, and Volkswagen MOIA held steady at 317, 47, and 12 respectively.
The Tesla growth rate is the data point that requires updating the narrative-vs-fleet-data thread Issue #9 established. Tesla's registration scale is now growing rapidly, with the registered fleet nearly doubling in a month. Commercial operational scale appears to remain significantly smaller (Electrek tracker analyses earlier this month suggested Tesla's total active commercial fleet at roughly 34 vehicles across all markets including supervised Bay Area FSD), but the regulatory registration pipeline is filling rapidly enough that operational scale could follow within the next 6-12 months.
For operations infrastructure planning, the registration data implies real depot and charging buildout requirements ahead. 620 Waymo vehicles in Texas need depot capacity, charging, and workforce scaled to support that fleet. 69 Tesla vehicles registered means Tesla is also committing to Texas operations infrastructure, not just speculation. The Avride 317-vehicle figure (unchanged since May 28 but already substantial) represents Texas-based operations infrastructure that exists today, supporting the Dallas robotaxi service.
For comparable scale context, Avomo manages approximately 400 AVs across Austin and Atlanta with a team of more than 200 specialists, per the June 2 Madrid partnership announcement. That's the operational footprint required to keep registered vehicles productive — depot infrastructure, charging, maintenance, technician workforce, dispatch coordination. Each new registered AV represents commensurate operations infrastructure scaling behind it.
The September 1 Texas autonomous vehicle safety law that requires commercial vehicles to get DMV authorization before operating on public streets without a human driver will provide the next data point. The structural question worth tracking: whether the gap between regulatory registrations and operating fleet narrows or persists as that authorization requirement takes effect.
3. Waymo pulled back from highway service across four markets
On June 17, Waymo filed a voluntary software recall with NHTSA covering 3,871 Jaguar I-PACE vehicles equipped with its fifth-generation automated driving system. The recall covers thirteen incidents during April and May (six in Phoenix, seven in the San Francisco Bay Area) where vehicles entered active freeway construction zones. NHTSA published the filing June 18. The software remedy remained "under development" as of the June 13 filing.
The autonomy and safety story has been well covered by trade and mainstream press this week. The operations-side implications matter for this newsletter's beat.
Waymo paused all freeway service across San Francisco, Los Angeles, Phoenix, and Miami pending the software fix. Surface-street routing only. That means measurably longer trip times in markets where Waymo competes most directly against conventional ride-hailing on point-to-point speed. It also means reduced vehicles-per-hour throughput on certain routes, which affects asset utilization across the affected fleet during the pause period.
The expansion timing matters more. Waymo had planned 2026 expansion to more than 20 new cities including London and Tokyo. Highway routing was the capability that made many of those expansions economically viable on commute-distance trips. Until the fix deploys, the realistic ramp slows. For operators planning to support Waymo expansion (Avomo, Moove.io, Avis, Element, others), the recall translates into delayed depot buildout timing and pushed-back hiring schedules in upcoming markets.
This is also Waymo's second recall in roughly thirty days, following the May 2026 recall related to vehicles entering flooded roads in San Antonio. Two capability-focused recalls in a month is a different pattern signal than isolated incidents. For AV companies planning institutional capital raises against operational projections, regulatory drag at production scale is now a real input.
Looking ahead
This week's content lineup:
"How AV Fleet Operators Win" framework piece, the prescriptive analysis of the structural moves that define category leaders. Building on the four-category framework and the operations-as-separate-category thesis from this month's analysis.
Continued analytical work tracking the institutional capital convergence pattern as additional partnerships emerge.
Until next Tuesday,
AVFleetTech