**Summary**
The speaker surveys the global robo‑taxi landscape as of early 2026, giving fleet sizes, operating cities, 2026 targets, and structural challenges for each major player:
| Company | Current fleet (≈) | Cities / regions | 2026 target | Key strengths | Main weaknesses |
|---------|-------------------|------------------|------------|---------------|-----------------|
| **Whimo** (leader) | 3,000 vehicles | 10 U.S. cities (Phoenix, SF, LA, Austin, Atlanta, Miami, Dallas, Houston, San Antonio, Orlando) | 3,500 vehicles (year‑end) | Early mover, brand trust, 6th‑gen driver, Magna partnership doubling capacity | Not profitable, high vehicle cost (~$175 k each), slow production (single‑digit/day), LiDAR/HD‑map dependent |
| **Apollo Go** (Baidu) | >1,000 in Wuhan; 26 cities total | China‑centric (largest deployment in Wuhan) | Frozen – only pre‑approved additions can proceed | Largest Chinese fleet, 20 M+ cumulative orders | March 31 cloud‑dispatch failure stalled >100 cars; Chinese regulators froze new permits, test programs, and city expansions |
| **WeRide** | ~1,023 (Jan 2026) | Guangzhou, Beijing, Abu Dhabi, Dubai, Riyadh, Singapore (launching) | 2,200‑2,800 vehicles globally (realistic) | First Chinese‑origin driverless service with city‑level permits outside the U.S.; strong Middle‑East & SE‑Asia diversification; partnerships with Gilly Ferros, Uber, Lenovo | Still LiDAR/HD‑map reliant; hit by China license freeze on domestic side |
| **Pony AI** | >1,400 (late Mar) | Domestic: Beijing, Shanghai, Guangzhou, Shenzhen, Hefei, Changsha; International: Zagreb (Croatia), Doha, Dubai | 2,400‑3,000 vehicles (realistic) | Strong Toyota joint‑venture for mass‑produced Gen‑Z B7 4X robo‑taxi; 160% YoY robo‑taxi revenue growth; break‑even unit economics in two Chinese cities | Expansion in China halted by license freeze; still depends on LiDAR/HD‑map and sensor‑fusion stack |
| **Zoox** (Amazon) | <500 (guess) | Las Vegas (Sept 2025), San Francisco (Nov 2025); expanding to Austin, Miami (employee‑first) | 500‑1,000 vehicles (optimistic) | Purpose‑built bidirectional vehicle; unlimited Amazon runway; beginning to charge riders | Fleet still small; limited by NHTSA exemption of 2,500 vehicles/year |
| **Wave** | 0 paid‑ride vehicles (pre‑revenue) | Planning London & Tokyo trials later 2026 | Few dozen trial vehicles by year‑end | AI‑first end‑to‑end driving stack; $1.5 B Series D (Huizu, Mercedes, Stellantis, Nissan, Uber, Microsoft, Nvidia) | No commercial robo‑taxi service yet; still relies on Nvidia Hyperion + sensor‑fusion; unproven at scale |
| **Cruise** (GM) | Defunct (funding stopped Dec 2024) | — | — | — | $10 B investment written off after 2023 pedestrian incident |
| **Motional** (Hyundai‑Aptiv) | Pilot only, no paid service at scale | — | — | — | Limited to pilot phase; no commercial robo‑taxi revenue |
| **Tesla** | 25 unsupervised vehicles (≈19 Austin, 3 Dallas, 3 Houston) | Texas (Austin, Dallas, Houston) – expanding to 12 states by Dec 2026 (Phoenix, Miami, Orlando, Tampa, Las Vegas, etc.) | **≥10,000** unsupervised robo‑taxis by Dec 2026 (minimum; could reach 15‑20 k) | • CyberCab volume production started Apr 24 2026 (target 15‑25 k units/yr; 30‑50% → 5‑12 k for own fleet) <br>• Existing Model Y fleet (millions) can be converted via AI4 chip + software update <br>• Cost per vehicle <$30 k vs. Whimo’s ~$175 k (≈6× cheaper) <br>• Production rate: thousands/week at Giga Texas vs. Whimo’s few/day <br>• Software lead: 1 M+ FSD supervised users, Cortex with 40 k AI4 chips, native integer training <br>• Regulatory: 4 U.S. states live, dozen by year‑end; EU approval pathway (Netherlands done, others summer 2026) <br>• Operational: 80 k+ Superchargers, 1.3 k+ service locations, live robo‑taxi app taking payments | • Past missed timelines (2020, 2022, 2024) <br>• CyberCab volume still unproven (Q2 earnings late July will be key) <br>• Regulatory delays in any state could shift timeline <br>• Whimo still scaling to 3,500 vehicles and 1 M weekly rides <br>• Potential cloud‑dispatch‑type failure risk (mitigated by local AI4 processing) <br>• Consumer demand at scale still being tested |
**Why Tesla could flip the race in ~8 months**
1. **Cost advantage** – CyberCab target <$30 k per unit vs. Whimo’s ~$175 k (≈6× cheaper).
2. **Production speed** – Giga Texas can output thousands/week; Whimo’s Magna upfit shop builds only a few per day.
3. **Software & data lead** – Massive fleet of supervised FSD users provides continuous intervention data; Cortex training loop with 40 k AI4 chips is unmatched.
4. **Regulatory momentum** – Already approved in four U.S. states; on track for a dozen by year‑end; EU approvals progressing (Netherlands done, others summer 2026).
5. **Operational ecosystem** – Own Supercharger network, service centers, and payment‑enabled robo‑taxi app give end‑to‑end control no competitor possesses.
Even with only 25 unsupervised vehicles today, Tesla’s underlying cost, production, software, and operational advantages mean its fleet growth trajectory is far steeper than rivals’. If CyberCad production meets the 30‑50 % allocation assumption, Tesla could field 5‑12 k vehicles by year‑end; combined with Model Y conversions, a conservative forecast is **≥10,000** unsupervised robo‑taxis by December 2026—roughly three times Whimo’s confirmed 3,500‑vehicle target and far ahead of all other players.
**What to watch**
- **Tesla unsupervised count** (weekly): look for 100 → 500 → 1 k → …
- **Q2 2026 earnings (late July)** – first real CyberCab production number; determines if the 10 k minimum is conservative or aggressive.
- **State rollout** – each new state (Phoenix, Miami, etc.) is a concrete signal.
- **Whimo’s growth** – whether it exceeds its 3,500 target and closes the gap.
- **China license freeze duration** – longer freeze benefits Tesla by keeping Apollo Go, WeRide, and Pony AI from expanding.
- **European approvals** – next country after the Netherlands (Germany/France/Italy) triggers EU‑wide rollout.
In short, while Tesla currently trails in raw vehicle numbers, its cost structure, production capacity, software/data advantage, regulatory progress, and operational stack give it a decisive growth edge that could make it the global robo‑taxi leader by the end of 2026.
1. CyberCap volume production officially started on April 24th.
2. The first CyberCap rolled off the line on February 17th.
3. As of March, Whimo operates approximately 3,000 robo taxis.
4. Whimo operates in 10 US cities: Phoenix, San Francisco, Los Angeles, Austin, Atlanta, Miami, Dallas, Houston, San Antonio, and Orlando.
5. Whimo does about 500,000 paid rides per week as of March.
6. Whimo logs 4 million rider‑only miles per week as of March.
7. Whimo crossed 20 million cumulative paid rides as of December.
8. Whimo’s year‑end fleet target is 3,500 vehicles.
9. Whimo announced adding 2,000 vehicles to its existing fleet this year, starting from about 1,500 in early 2025.
10. Whimo targets 1 million weekly rides by the end of 2026.
11. Whimo is expanding to 11+ new cities including Denver, Detroit, Las Vegas, Nashville, San Diego, and Washington DC.
12. London is Whimo’s first international city; Tokyo testing is continuing.
13. Whimo’s sixth‑generation driver is in production and the Magna Manufacturing Partnership is doubling capacity.
14. Each Jaguar I‑Pace used by Whimo costs about $75,000 plus another $100,000 in autonomous hardware, roughly $175,000 per vehicle.
15. The Magna Upfit shop builds at a single‑digit number of vehicles per day.
16. Whimo remains LiDAR dependent and HD‑map bound.
17. Apollo Go (Baidu’s robo‑taxi service) operates in 26 cities globally.
18. Wuhan alone has more than 1,000 Apollo Go robo taxis, their largest single deployment.
19. Apollo Go has crossed 20 million cumulative orders.
20. On the night of March 31st, more than 100 Apollo Go robo taxis stalled due to a cloud‑based dispatch system failure.
21. Chinese regulators stopped issuing new robo‑taxi permits and suspended expansion for driverless fleets after the incident.
22. We Ride’s global fleet as of January 2026 was 1,023 vehicles.
23. We Ride plans to surpass 2,600 vehicles by the end of 2026.
24. We Ride operates commercial driverless in Guangzhou, Beijing, Abu Dhabi, Dubai, Riyadh, and Singapore (launching this month).
25. We Ride is the first Chinese robo‑taxi company with city‑level fully driverless permits outside the US.
26. We Ride has a partnership with Gilly Ferros to produce 2,000 purpose‑built robo‑taxi GXR vehicles by year‑end, production starting Q3.
27. We Ride has a partnership with Uber in the Middle East for 1,200 robo‑taxis planned across Abu Dhabi, Dubai, and Riyadh by 2027.
28. We Ride has a deal with Lenovo for 200,000 autonomous vehicles globally over 5 years.
29. Pony AI’s fleet as of late March is over 1,400 vehicles.
30. Pony AI’s target by year‑end is 3,000+ vehicles across 20+ cities, with nearly half overseas.
31. Pony AI operates domestically in Beijing, Shanghai, Guangzhou, Shenzhen, plus Hefei and Changsha, launched in March.
32. Internationally, Pony AI is in Zagreb, Croatia (first commercial robo‑taxi service in Europe), Doha, Qatar, and Dubai (driverless launched in March).
33. Pony AI’s biggest partner is Toyota; a joint program aims to mass‑produce Gen Z B7 4X robo taxis, targeting thousand‑plus units in 2026.
34. Pony AI’s Q4 2025 financials showed robo‑taxi revenue up 160% year‑over‑year and charging revenue up 500%.
35. Unit economics break even in Guangzhou and Changsha within four months of the Gen Z Gen 7 launch.
36. Zuks (Amazon‑backed) operates in Las Vegas (since September) and San Francisco (since November).
37. Since launching less than a year ago, Zuks has logged nearly 2 million autonomous miles and transported over 350,000 passengers.
38. The Zuks vehicle is purpose‑built, bidirectional, with no front/back carriage.
39. For 2026, Zuks plans to quadruple the San Francisco service area, double Las Vegas locations (including Sphere and T‑Mobile Arena), and launch Austin and Miami for employees first.
40. Zuks started charging for rides in early 2026.
41. Zuks’ fleet is estimated to be under 500 vehicles total.
42. Zuks is limited by an exemption allowing only 2,500 total vehicles per year.
43. Wave’s Series D in February totaled $1.5 billion; investors include Huizu, Mercedes‑Benz, Stellantis, Nissan, Uber, Microsoft, and Nvidia.
44. Wave’s valuation was $8.6 billion.
45. Wave plans a first commercial robo‑taxi trial in London later this year with Uber, a Tokyo pilot late this year with Nissan and Uber, and Consumer L2 Plus hands‑off from 2027 in passenger vehicles.
46. Wave currently has zero active commercial robo‑taxi service and zero fleet running paid rides.
47. GM Cruise is officially dead; GM stopped funding Cruise as a robo‑taxi company in December 2024.
48. After the October 2023 incident where a pedestrian was dragged, GM restarted autonomous work only for personal vehicles, not robo‑taxis.
49. Motional (Hyundai joint venture) is still active but limited; it targets commercial service this year after a pilot phase, yet there is no commercial paid robo‑taxi service at scale today.
50. Auto X in China has accumulated about 2 million autonomous kilometers, not comparable to Pony or We Ride.
51. As of today, Tesla has 25 unsupervised robo‑taxi vehicles (some say 26): 19 in Austin, three in Dallas, three in Houston.
52. Tesla announced a dozen state rollout; currently one state (Texas) runs unsupervised, with plans to reach 12 states by December.
53. Confirmed states for the first half of this year are Phoenix, Miami, Orlando, Tampa, and Las Vegas.
54. Tesla can convert existing Model Y vehicles to unsupervised via a software update plus regulatory signoff; the AI4 chip is in every Tesla built since mid‑2023.
55. A documented Tesla robo‑taxi ride in Dallas cost $6.15, while the same trip in Whimo cost $13.93 (56% cheaper).
56. The Netherlands approved Tesla FSD supervised on April 10th; Europe‑wide approval is targeted for summer of this year.
57. Tesla’s Cyber Cap production target is under $30,000 per vehicle.
58. Whimo’s Jaguar I‑Pace costs roughly $175,000 each.
59. Tesla Giga Texas at full Cyber Cab ramp can produce thousands per week; Whimo’s Magna Upfit shop builds a few per day.
60. Tesla has 1 million+ FSD supervised users feeding interventions per day; Cortex uses 40,000 AI4 chips in a training loop.
61. Tesla owns 80,000+ Superchargers globally and 1,300+ service locations.
62. Tesla’s robo‑taxi app is live and accepting real payments.
63. Tesla missed every prior robo‑taxi timeline (2020, 2022, 2024).
64. Cyber Cap volume production is unproven; the first public production number will appear in Q2 earnings (late July).
65. The 12‑state rollout depends on regulators; any single‑state delay shifts the timeline.
66. Whimo is likely to hit its 3,500‑vehicle target and scale rides to 1 million weekly.
67. The Wuhan incident could affect any robo‑taxi service; Tesla’s mitigation relies on local AI4 chip processing, reducing cloud dependence.
68. Consumer demand at scale remains unproven; early pricing data shows Tesla undercuts heavily, but behavior at scale is still being tested.