👋 Another week, another set of developments that remind us why autonomous vehicles remain one of the most fascinating and complex technology sectors to watch unfold.
This edition covers Waymo putting numbers to its U.S. fleet, Tesla widening its Austin robotaxi footprint and beginning supervised tests underground in Vegas, and fresh data dropped on how AVs are already shifting rideshare economics.
Let’s dive in!
⏱️ ~1600 words, ~7 min read
🚗 The Acceleration Phase Begins
Waymo's co-CEO Tekedra Mawakana made a striking prediction this week:
“You’re going to start seeing our cars in a lot of cities. If you think about our business in terms of scale, we’re currently giving hundreds of thousands of rides every week and, in all likelihood, by the end of next year, we will be offering around one million rides per week.”
If Waymo hits anything close to that run-rate, the U.S. ridehail map changes.
The SF Examiner reports that Waymo now operates over 2,000 commercial vehicles across the U.S., with ~800 deployed in the Bay Area alone, another ~500 in Los Angeles and ~400 in Phoenix.
We already know Washington DC, Miami, and New York City are in the pipeline, with data collection underway in Houston, Orlando, San Antonio, Las Vegas, and San Diego. The company has even extended its reach internationally to Japan, indicating global ambitions that stretch well beyond American cities.
Waymo is the clear leader in autonomy, and continues to lead.
Meanwhile in Texas: Tesla’s supervised service is getting bigger
Tesla's Austin operations continue to expand as well, with the service area now covering 173 square miles — nearly double the previous coverage. The company has also increased its fleet by 50%, though from a relatively small base of ~20 vehicles to what may now be around 30 Model Y robotaxis. The program remains supervised (passenger-seat safety monitor in Austin; safety drivers in the Bay Area).
Source: Statesman
The company’s own account teased that public access could begin in September and Texas’ AV permit framework takes effect Sept. 1 (with enforcement phased later), for which Tesla has already secured a TNC license via “Tesla Robotaxi LLC.”
Tesla and the tunnel wildcard
The Boring Company has begun testing Tesla’s Full Self-Driving (Supervised) inside the LVCC tunnel system with safety drivers onboard — no passengers yet.
Steve Hill, CEO of the Las Vegas Convention and Visitors Authority, confirmed that Tesla has been testing Full Self-Driving (Supervised) in the 3.5-mile tunnel system for several months.
Steve Hill calls fully autonomous operations “a ways off,” noting that safety drivers still need to “periodically intervene” as vehicles encounter “interesting but odd lighting” and semi-smooth rock walls that present challenges for a vision-only system. Hill also emphasized that the tunnel system is the convention center’s highest-rated feature, and that removing the safety driver would follow a formal safety validation process (with potential third-party review).
From a systems perspective, tunnels should be a “friendly” ODD: controlled access, predictable conditions, and no pedestrians, cyclists, or unpredictable human behavior to navigate around. That should make autonomy “easy” — yet Hill noted that safety drivers still need to "periodically" intervene, and the colorful tunnel lighting combined with rock walls creates challenges for Tesla's camera-only approach.
Radar/lidar proponents will point to this as textbook justification for sensor diversity; Tesla argues the opposite, that extra modalities create sensor contention.
Regardless, even before driver-out, the ability to bypass surface traffic via the tunnels is a compelling value proposition (my bet is Tesla eventually gets to driver-out in the tunnels after safety validation). That proposition strengthens especially if/when tunnel coverage reaches the airport and major resorts, where traffic avoidance creates a clear consumer narrative.
Signals point to Las Vegas being strategically important for Tesla’s broader robotaxi business, not just the tunnels. Tesla is actively recruiting Autopilot Supervisors in Las Vegas/Henderson—a classic indicator.
Translation: Las Vegas looks set to become the next battlefield for Tesla’s robotaxi business—with the tunnels serving as a controlled beachhead that can then extend to selected surface routes.
The bigger picture: fleet scaling will accelerate
Zooming out: Waymo is scaling fleets and signaling multi-city launches; Tesla is expanding supervised service with a path to driverless in at least some ODDs; and others (Zoox, May Mobility, Nuro, and more) are lining up market entries.
I’m convinced the robotaxi fleet size will bend upward over the next 12 months. The benefits are real — fewer impaired or distracted drivers, lower crash severity, and potentially more equitable mobility access. But: robotaxis also introduce externalities.
This week, two stood out to me in both reporting and data:
Externality #1 — Curb & neighborhood friction
Residents in Los Angeles and Phoenix report Waymo vehicles repeatedly idling at the same curb spots between trips, sometimes for minutes or hours. The Verge documented cases where cars prefer two specific spaces on a block; if both are occupied, the vehicle slows but won’t choose other open spaces nearby — behavior likely emerging from ML-driven staging logic trained to balance demand, energy conservation, and congestion.
To be clear: In Los Angeles (and similarly elsewhere), what the Waymo vehicles are doing is fully legal: commercial passenger vehicles under 22 feet follow the same parking rules as private vehicles, with a three-hour limit per spot. That means Waymos may park on-street between trips as long as they respect local restrictions
Waymo also emphasizes a good-neighbor posture. When residents raise concerns, Waymo can mark specific curbs as “no-parking” for its own fleet, so vehicles actively avoid those locations going forward. I genuinely like how cooperative Waymo is here — and it’s a reminder that, in an early scaling phase, there are side effects you don’t anticipate at first glance. The important part is adapting quickly and finding solutions that work for residents, cities, and operators — which is exactly what Waymo is doing.
But if we assume many more robotaxis over the next 12 months — especially in SF/LA/Phoenix/Austin — cities and operators will need a clearer playbook for the curb. It’s important that on-street parking stays legal and acceptable, neighborhoods feel respected, and fleet scaling can continue without unnecessary friction.
Externality #2 — Human-driver ride-hail economics
Fresh Gridwise data offers a clear year-over-year lens on how AV entry is already tugging at the rideshare labor market in San Francisco, Los Angeles, Austin, and Phoenix.
Hourly earnings are slipping precisely where autonomous fleets operate: down 6.9% in San Francisco, 5.3% in Austin, 4.7% in Los Angeles, and 3.8% in Phoenix from July 2024 to July 2025—even as the national average inched up 1.0%.
Part of the story is incentives. Platforms pulled back hard on per-trip promos:
−47.1% nationwide—with Los Angeles and Phoenix seeing the steepest cuts (both >=64%), San Francisco roughly matching the national drawdown, and Austin showing a milder ~33% reduction. When you increase effective supply (humans plus AVs) in a metro, you don’t need to pay as much to keep cars on the road.
Utilization reinforces the local nature of the impact. Time-with-passengers fell 8.6% in LA and 3.1% in SF, a touch worse than the −2.9% national drift, while Austin (+1.3%) and Phoenix (+1.4%) actually ticked up.
The takeaway isn’t that AVs “break” the labor market; it’s that their effect is gradual, localized, but very real. As noted above, fleets are poised to scale in the coming months. That expansion could widen gaps in pay and utilization. It will be interesting to closely monitor these effects as deployments deepen.
I’m excited for the months ahead—fleet expansions, new city launches, and richer on-the-ground evidence of how autonomy is reshaping mobility.
🔗 SF Examiner / Road to Autonomy / Gridwise / Statesman / TechCrunch / Fortune / The Verge (1) / The Verge (2) / Driverless Digest
💡 Quick Takes
🇰🇷 Seoul Launches Self-Driving Shuttle Service
South Korea's capital will begin operating its first autonomous shuttle service in September, featuring two 11-seater vehicles on a 4.8km route through central Seoul. The buses have no steering wheel or driver's seat but include a safety monitor — representing another step toward mainstream AV adoption in Asia.
🪖 Scout AI Wins $150M Army Contract
The startup secured a major U.S. Army contract for its Fury autonomy platform, which uses camera-only systems on Infantry Squad Vehicles. The 16-month initial contract could expand to $150 million in additional procurement, highlighting military interest in vision-based autonomy systems.
🇯🇵 Isuzu Building Autonomous Truck Test Facility
The Japanese truck manufacturer will establish a dedicated autonomous driving test course in Hokkaido, opening in September 2027. The facility will focus on Level 4 autonomous trucks and buses, with plans to welcome external companies including startups and suppliers.
🔗 Isuzu
📶 Hesai Group Planning Hong Kong IPO
The world's largest lidar manufacturer is preparing to list shares in Hong Kong as soon as next month, targeting $300 million in funding. The move comes amid renewed U.S. delisting risks for Chinese companies and could signal broader capital market shifts for AV suppliers.
🚗 Stellantis shelves L3 ‘AutoDrive’
Stellantis has put its in-house Level 3 ADAS on ice—despite calling it “ready” in February—citing high costs, tech hurdles, and weak consumer demand; the company maintains the tech is “available” but hasn’t launched.
🔗 Reuters
📚 Worth Reading/Listening:
Harry Campbell & Chris Paxton: Breaking Down the Cost of a Waymo Zeekr
📊 Weekly Performance
Note: Stock performance data as of August 30, 2025. Past performance does not indicate future returns.
Thanks for reading!
If you found value in this newsletter, please consider sharing it with a friend or colleague who might benefit from these insights.
And if you haven't already, subscribe to stay updated on the latest developments in the autonomous driving industry
waymo's scaling is impressive, shows the momentum in the industry. the data on human driver pay is a real-world consequence to watch, highlights the need for a careful transition as these technologies become more widespread.
This is where autonomy transitions from technical achievement to operational reality. Waymo's scaling and Tesla's tunnel struggles reveal a key divergence: robust multi-sensor systems are accelerating while vision-only approaches face even basic environment limitations.