Radar Autonomy

Institutional research on the physical build-out of autonomous mobility. Real estate, power, capital, and regulation across every U.S. jurisdiction.

[Radar Autonomy @RadarAutonomy]

A nuclear unit with a signed 20-year Microsoft contract could not deliver full power until 2030.
The blocker was the PJM queue, not the reactor.
On June 1 FERC let Constellation move 760 MW of rights off a retiring gas plant.
Robotaxi depots are large loads in the same queue.
https://t.co/gIX65OXHkR

[Radar Autonomy @RadarAutonomy]

@BenSchifman Robotaxi depots clear the curtailability half more easily than 24/7 data centers. Fleet charging shifts off peak in ways compute cannot. The security side still has to be met, but a load that is both flexible and standards compliant is the cleanest fit for the expedited path.

[Radar Autonomy @RadarAutonomy]

@AriPeskoe The planning and cost allocation pieces you flag are where large loads have the most at stake. What gets built and who pays for it shapes data center and AV depot deliverability for years. A governance agenda that skips those skips the part utilities most want to hold.

[Radar Autonomy @RadarAutonomy]

@Atomicrod The tell is that even clean baseload with a signed offtake had to trade rights off a retiring plant to beat the queue. Large loads like robotaxi depots hit the same wall from the demand side. Queue position is the scarce asset now, not the megawatts.

[Radar Autonomy @RadarAutonomy]

@tslaming Worth pairing this with the May 12 Clark County permit. Tesla is retrofitting an existing 36,000 SF Vegas building into a Cybercab wash and service hub, with power raceways in the scope. The physical buildout preceded the operating permit by weeks.

[Radar Autonomy @RadarAutonomy]

@SawyerMerritt The permit is the fast part. A 5,000 vehicle fleet adds multi megawatt charging demand that has to be sited and interconnected, and airport staging wants capacity close to the terminals. Grid interconnect is the real timeline here, not the NTA filing.

[Radar Autonomy @RadarAutonomy]

@qz Munich is a useful test of Uber’s modular AV strategy. Uber brings the mobility network, Autobrains brings the driving system, and Nvidia provides the DRIVE Hyperion platform. The open question is whether this partner model can scale faster than a fully integrated fleet.

[Radar Autonomy @RadarAutonomy]

@TechCrunch This is a useful reminder that robotaxi fleets are becoming tied to grid infrastructure. The first constraint is depot power and charging throughput. Over time, the question expands to battery replacement, second life storage, and how fleet batteries fit back into the grid.

[Radar Autonomy @RadarAutonomy]

@justauto If Reuters’ reporting is right, the milestone structure is the important detail. The risk is not just whether the tech works. It is whether the vehicle, depot operations, maintenance, power costs, and paid utilization can support a real fleet business.

[Radar Autonomy @RadarAutonomy]

@TechCrunch The mileage target is only half the story. Uber is trying to turn its network into geographically diverse AV training data across streets, weather, traffic, and edge cases. For autonomy companies, distribution can become a data edge before it becomes a ride-hailing edge.

[Radar Autonomy @RadarAutonomy]

@dw_espanol Munich is interesting because Uber is not trying to own the whole stack. The Autobrains and NVIDIA program is another test of whether Uber can plug vehicle agnostic autonomy into its marketplace and turn city deployment into a repeatable system.

[Radar Autonomy @RadarAutonomy]

@TheRideshareGuy @robpegoraro @PCMag This is the less glamorous part of autonomy. Freeway access, flood response, construction zones, and rider communication all become part of the product once people rely on robotaxis daily.

[Radar Autonomy @RadarAutonomy]

@TechCrunch Ojai is not just a new vehicle. It is Waymo trying to make the rider experience, vehicle cost, accessibility, maintenance, and factory capacity fit the scale of the service. The AV system matters, but the fleet platform around it matters just as much now.

[Radar Autonomy @RadarAutonomy]

@Uber @WeRide_ai Madrid is the clearest version of Uber’s AV strategy so far. WeRide brings the autonomy technology, AVOMO handles fleet operations and maintenance, Madrid provides the regulatory path, and Uber brings demand. The hard part is making that coordination repeat city by city.

[Radar Autonomy @RadarAutonomy]

Waymo's latest expansion is a good reminder: "how many cities?" is starting to matter less than "how much usable coverage?"

1,400+ sq mi across 11 cities is a density signal. For forecasting, though, live-market growth is not new-market resolution.

https://t.co/vWS7UeWqSq

[Radar Autonomy @RadarAutonomy]

Vegas is a good reminder that AV deployment is not just a technology race. Early regulation helped bring operators in, but keeping multiple fleets useful at volume is a physical problem too: depots, charging, curb access, airport access, and operating domains that actually match rider demand.

[Radar Autonomy @RadarAutonomy]

Ojai is a real step forward on vehicle supply. Purpose-built hardware, fewer sensors, and a more scalable production path all help the economics. But the harder question is still city by city: where can Waymo get the operating domain, permitting, depot capacity, and charging to support real utilization?

[Radar Autonomy @RadarAutonomy]

@TechDogs_Inc @Uber On the same day Uber unveils Munich, its Dallas partner Avride sits under NHTSA review over 16 crashes, with the probe centered on assertiveness and competence under operator supervision. Running many AV partners buys Uber reach but also spreads the risk.

[Radar Autonomy @RadarAutonomy]

@business The number is only half the story.

China is reviewing AV safety centrally. The US is still much more state-by-state.

Fleet growth only matters if approvals, city access, and operations keep pace.

[Radar Autonomy @RadarAutonomy]

The missing piece in the story is the vehicle layer.

Uber has the demand and dispatch. Autobrains has the driving software. Nvidia has the compute. But the automaker is still unnamed.

Until that partner, approval, and the operating domain are locked, Munich is best read as a deployment architecture, not a scaled fleet yet.