Data: Bot Memo synthesis of company IR disclosures, NHTSA filings, California DMV reports, and peer-reviewed research
The global artificial intelligence (AI) in transportation market reached $5.53 billion in 2025 and is projected to hit $34.83 billion by 2034 at a 22.7% CAGR. That headline figure understates the actual money in motion. Once you add the embedded compute layer (ADAS), the freight layer (autonomous trucking), and the orchestration layer (AI in supply chain at $14.49 billion), AI in transportation statistics for 2026 describe a market where four revenue stacks now run in parallel. This piece pulls every number from primary sources: earnings calls, 10-Ks, NHTSA filings, the California DMV, and peer-reviewed safety research. Click through any figure to see the source.
On this page
- Key Takeaways
- AI in Transportation Market Size 2026
- Robotaxi Commercialization: Waymo, Tesla, and Apollo Go
- Pony.AI, WeRide, and Zoox: The Second Wave
- Autonomous Trucking Crosses Into Commercial Revenue
- ADAS at Scale: Mobileye and the Embedded AI Layer
- AI in Logistics, Supply Chain, and Aviation
- AV Safety Data: California Test Miles and Peer-Reviewed Research
- Regulation 2025-2026: NHTSA SGO, EU AI Act, and China L3 Approvals
- Investment and Outlook: What the Funding Data Shows
- Bottom Line for Operators and Investors
- Frequently Asked Questions
- Methodology
Key Takeaways
- Waymo scaled from 250,000 weekly paid rides in April 2025 to 450,000+ by December 2025, with 170.7 million cumulative rider-only miles by March 2026.
- Tesla’s FSD (Supervised) fleet crossed 9 billion cumulative miles by April 2026, adding about 1 billion miles a month at its current run-rate.
- Baidu’s Apollo Go logged 3.4 million driverless rides in Q4 2025 and 20 million+ cumulative public rides by February 2026.
- The global ADAS market hit $72.1 billion in 2025 per Future Market Insights, with Mobileye shipping 35.7 million EyeQ and SuperVision systems that year. (Precedence Research’s 2025 ADAS estimate is more conservative at $38.5 billion.)
- A peer-reviewed study of 56.7 million Waymo rider-only miles found 79% fewer any-injury crashes and 85% fewer suspected serious-injury crashes versus human drivers in the same cities.
AI in Transportation Market Size 2026
Precedence Research pegs the global AI in transportation market at $5.53 billion in 2025, growing to $34.83 billion by 2034 at a 22.7% CAGR. That number captures the AI software and services layer. It excludes the hardware sitting underneath, where ADAS sensors and silicon already produce vastly more revenue.
The four revenue layers running in parallel:
| Layer | 2025 Market Size | Source |
|---|---|---|
| AI in transportation (software/services) | $5.53B | Precedence Research |
| ADAS (hardware + software) | $72.1B | Future Market Insights (Precedence: $38.5B) |
| AI in supply chain | $14.49B | MarketsandMarkets |
| Robotaxi rides + autonomous freight | Pre-revenue/early commercial | Company IR disclosures |
Source: Bot Memo synthesis of analyst firm reports and company disclosures, 2025 figures
That $5.53 billion software-and-services layer is itself a bundle of use cases: AI traffic management systems, congestion and traffic flow optimization, predictive analytics for public transit scheduling, real-time routing, and computer-vision incident detection. Those are the applications most “AI in transportation” market reports actually count when they size the category.
So when commentators say “AI is coming to cars,” the right question is which layer. ADAS is already 13x bigger than the dedicated AI-in-transportation software market, and it’s growing too: Future Market Insights projects ADAS reaches $228.2 billion by 2035 at 12.2% CAGR. Each layer has its own buyer, regulator, and revenue model. The rest of this article works through them.
Robotaxi Commercialization: Waymo, Tesla, and Apollo Go
Three operators dominate the robotaxi conversation in 2026, and they’re running different playbooks.
Waymo disclosed 250,000 weekly paid rides on Alphabet’s Q1 2025 earnings call (April 24, 2025), then scaled to 450,000+ weekly rides by December 2025. Cumulative rider-only autonomous miles reached 127 million by mid-December 2025 per the company’s Safety Hub update, and crossed 170.7 million by March 2026. The fleet stood at roughly 2,500 vehicles in commercial service by late 2025, including Phoenix, San Francisco, Los Angeles, Austin, Miami, Dallas, Houston, San Antonio, and Orlando, with Atlanta and additional metros in earlier-stage rollout.
Tesla took a different route. Its Robotaxi service launched in Austin on June 22, 2025 with about 10 Model Y vehicles and human safety monitors riding in the passenger seat. By April 2026, the Tesla FSD (Supervised) fleet crossed 9 billion cumulative miles, adding about 1 billion miles a month at its current run-rate. Annual FSD miles in 2025 hit 4.25 billion, up from 2.25 billion in 2024. The mileage split runs 64% highway and 36% city streets.
Baidu‘s Apollo Go matched Waymo’s spring 2025 weekly volume by Q4 2025. The company reported 3.4 million fully driverless rides in Q4 2025 with weekly volume peaking above 300,000, and cumulative 20 million+ public rides by February 2026. Apollo Go now runs in 22-26 cities including Dubai and Abu Dhabi.
The denominators are not comparable. Waymo reports rider-only miles. Tesla reports supervised miles. Apollo Go reports rides. Each metric implies a different safety standard, so cross-platform numerical comparisons mostly mislead.
| Operator | Weekly Paid Rides | Cumulative Miles or Rides | Fleet Size |
|---|---|---|---|
| Waymo | 450,000+ (Dec 2025) | 170.7M rider-only miles (Mar 2026) | ~2,500 vehicles |
| Tesla | Not disclosed publicly | 9B supervised miles (Apr 2026) | Robotaxi: ~10 launch vehicles + FSD fleet |
| Apollo Go | 300,000+ peak | 20M+ cumulative public rides (Feb 2026) | 1,000+ vehicles |
Source: Bot Memo compilation of company IR disclosures, April 2025 to April 2026
Pony.AI, WeRide, and Zoox: The Second Wave
Beneath the top three, a second tier of operators is putting up real numbers, often with cleaner unit economics.
Pony.AI reported 961 robotaxis as of November 23, 2025, including 667 Gen-7 vehicles. The target is 1,000 by year-end 2025 and 3,000 by end of 2026. The 2025 results matter more than the fleet count: Pony.AI’s robotaxi services revenue grew 128.6% YoY to $16.6 million, with fare-charging revenue up about 400%. The company says it achieved Gen-7 city-wide unit-economics breakeven in Guangzhou at 232 daily orders per vehicle. That’s a single-city claim, but it’s the first time a Chinese L4 operator has put breakeven math on the record.
WeRide reported 1,023 vehicles in its global robotaxi fleet as of January 2026, with full-year 2025 revenue up 90% YoY. The expansion story is global: fully driverless service in Guangzhou, Beijing, and Abu Dhabi, plus public passenger service in Dubai, Riyadh, and Singapore by April 1, 2026. WeRide and Uber agreed to deploy at least 1,200 robotaxis across the Middle East by 2027.
Zoox, Amazon’s robotaxi subsidiary, moved from R&D to consumer rides in 2025. Free public service launched in Las Vegas on September 10, 2025, and a free pilot opened in San Francisco’s South of Market, Mission District, and Design District on November 18, 2025 under the Zoox Explorers program. Paid commercial service is pending regulatory approval.
The economic reality: even with Pony.AI’s $16.6 million in 2025 robotaxi revenue or Zoox’s free pilot, this tier’s fleet sizes are 30-40x smaller than Waymo or Apollo Go. The path to scale runs through manufacturing (WeRide’s 2,000-unit GXR deal with Geely Farizon) and unit economics (Pony.AI’s 232 orders/vehicle threshold), not press releases.
Autonomous Trucking Crosses Into Commercial Revenue
Trucking moved from pilot to revenue in 2025, and it did so faster than passenger robotaxis.
Aurora Innovation launched the first commercial driverless heavy-duty trucking service on US public roads on May 1, 2025, running the Dallas-Houston I-45 corridor with launch customers Uber Freight and Hirschbach Motor Lines. Aurora later expanded to a 600-mile Fort Worth-El Paso lane and crossed 100,000 driverless miles in Q3 2025.
Kodiak AI went public via SPAC at a $2.5 billion valuation on September 25, 2025. As of September 30, 2025, the company disclosed 10 driverless trucks, 3 million+ cumulative autonomous miles, and 5,200+ paid driverless service hours, projecting roughly 400% revenue growth into 2026.
Why is trucking ahead of passenger robotaxis on per-vehicle utilization? Highway-dominant routes mean fewer geofencing constraints than dense urban driving, and long-haul operators have a payer (the freight customer) plus a labor-cost wedge (the driver). The revenue pool is still small; Aurora and Kodiak combined disclosed numbers measured in low millions of dollars in 2025. But the unit story (paid driverless hours, paid loads, named carrier customers) is now in SEC filings, not pitch decks. For more on how funding is flowing into this space, see our analysis of how AI startups vs Big Tech are splitting innovation in transportation and beyond.
ADAS at Scale: Mobileye and the Embedded AI Layer
The volume play in AI-driven transportation isn’t robotaxis. It’s the chip in your next car.
Mobileye shipped 35.7 million EyeQ SoC and SuperVision systems in 2025, up from 29 million in 2024, with cumulative deployment around 230 million EyeQ chips in vehicles globally. Mobileye’s full-year 2025 revenue was $1.894 billion (up 15% YoY), and the company’s 8-year future revenue pipeline grew to $24.5 billion per its Q4 2025 disclosure.
That 35.7 million annual unit volume is the right way to size AI in transportation by reach. Every mainstream new vehicle now ships with computer vision-based driver assistance. The global ADAS market reached $72.1 billion in 2025 and is projected to reach $228.2 billion by 2035 at 12.2% CAGR per Future Market Insights.
A definitional point worth getting right: ADAS sits at SAE Level 1-2 (driver-assist, eyes still on the road), while autonomous driving means Level 3 and above (the system is responsible for the dynamic driving task within its operational design domain). Both run on AI. The regulatory and revenue models are different. ADAS sells per vehicle to OEMs; L4 robotaxi sells per ride to consumers.
AI in Logistics, Supply Chain, and Aviation
The non-vehicle layer of AI in transportation has its own market sizing. MarketsandMarkets puts the AI in supply chain market at $14.49 billion in 2025, growing to $50.01 billion by 2031 at 22.9% CAGR.
The use cases here run across route optimization for fleets, machine learning demand forecasting, predictive maintenance for trucks, warehouse robotics integration, and last-mile dispatch. Most of it runs on the same predictive analytics stack that ingests real-time telematics from connected fleets. Some of this overlaps with autonomous trucking (Aurora’s freight customers buy route optimization too) and some with ADAS (telematics platforms ingest sensor data from ADAS-equipped fleets). Analyst firms still report it as a separate market because the buyers are different: supply-chain VPs and operations leaders, not auto OEMs. For deeper coverage of where investor money is going in this segment, see our list of top logistics and supply-chain AI startups.
Aviation AI sits outside the four-layer framing in this article. The detailed aviation AI statistics aren’t covered in our current dataset; we’d point readers back to the broader $5.53 billion AI-in-transportation umbrella figure for now.
AV Safety Data: California Test Miles and Peer-Reviewed Research
The AV safety conversation in 2026 anchors on a peer-reviewed Waymo study, the California DMV’s annual mileage release, and NHTSA’s Standing General Order data.
First, the California DMV reported AV permit holders logged more than 9 million test miles between December 1, 2024 and November 30, 2025, a record annual high. That’s the regulatory denominator: official, audited, geographically bounded.
Second, the peer-reviewed safety case got stronger. A May 2025 Traffic Injury Prevention study analyzed 56.7 million Waymo rider-only miles and found 79% fewer any-injury crashes (CI: 71%-85%) and 85% fewer suspected serious-injury-or-worse crashes than human-driver benchmarks in the same operating cities. Waymo’s own (non-peer-reviewed) March 2026 update on 170.7 million miles reported 92% fewer crashes causing serious or fatal injuries and 82% fewer crashes involving any injuries.
Third, the NHTSA Standing General Order captures incidents. As of late 2025, 1,429 SGO crash reports involving Waymo had been filed since July 2021. “Involving” is doing real work in that sentence: the SGO records any crash where the AV was present, including rear-end collisions caused by human drivers behind the AV. Tesla’s FSD numbers are supervised, not driverless, which means a human is the legal driver of record.
The honest framing: per-mile safety for L4 robotaxis is improving. Cumulative AV exposure (170.7 million Waymo miles, plus other operators’ contributions) is still small next to 3.279 trillion human vehicle-miles in the US in 2024. A million Waymo miles is roughly 0.00003% of US annual driving. The safety case is real and the scale is tiny. Both are true.
Regulation 2025-2026: NHTSA SGO, EU AI Act, and China L3 Approvals
2026 is the first year the US Standing General Order, the EU AI Act, and China’s L3 framework all have binding pieces in effect simultaneously, each with its own scope and reporting cadence.
United States: DOT Secretary Sean Duffy unveiled an April 24, 2025 framework expanding NHTSA’s AV Exemption Program to domestically built vehicles and amending the Standing General Order on crash reporting. The SGO amendment took effect June 16, 2025.
European Union: The EU AI Act came into force August 1, 2024. Prohibited-practices and AI-literacy rules became binding February 2, 2025, and high-risk vehicle AI obligations now bind August 2, 2028 after the May 2026 Digital Omnibus extension, layered onto existing UNECE/TAFR type-approval. Penalty exposure for prohibited practices runs up to EUR 35 million or 7% of global turnover, whichever is higher (Article 99). For automotive AI, the high-risk classification is the one to watch.
China: MIIT granted the first L3 conditional approvals in December 2025 to the Changan Deepal SL03 and BAIC Arcfox for geofenced expressway sections in Chongqing (50 km/h max) and Beijing (80 km/h max). Two vehicle approvals on two roads at low speeds doesn’t sound dramatic, but it’s the first time non-pilot passenger cars are allowed L3 on Chinese public roads.
The takeaway for global operators: cross-border product compliance is a cost line in 2026, not a future planning item. Any company shipping AI vehicle systems into the US, EU, and China is now navigating three different binding frameworks at once.
Investment and Outlook: What the Funding Data Shows
Bot Memo tracks AI funding across transportation-adjacent rounds: robotaxi operators, autonomous trucking, ADAS sensors, and AI logistics platforms. The pattern in 2025-2026 capital flows is operators with disclosed unit economics getting funded, not pre-revenue concept companies.
Two data points illustrate. Kodiak AI went public via SPAC at $2.5 billion in September 2025 with audited mileage and revenue numbers. Pony.AI’s US-listed equity story now reports 128.6% YoY robotaxi revenue growth. Capital is following operators that publish per-vehicle and per-route metrics.
Adjacent funding flows include ADAS sensor companies (lidar, vision systems), AI logistics platforms (route optimization, demand forecasting), and freight orchestration. We don’t make aggregate funding claims here without source-level data; for more on how funding rounds are reported and tracked, see our methodology on AI startup funding statistics.
The question for 2027-2028 is which players have the data, the regulatory permissions, and the disclosed unit economics to be standing. Today’s leaders by those three measures: Waymo (data and regulatory), Apollo Go (scale and regulatory), Pony.AI (disclosed unit economics in one city), Aurora (commercial trucking revenue), and Mobileye (revenue at OEM volume).
Bottom Line for Operators and Investors
The market splits into two distinct buyer profiles. Robotaxi and trucking operators chase per-mile unit economics on small commercial fleets. ADAS buyers, mostly OEMs, ship driver-assist hardware to every mainstream new vehicle. Mobileye’s 35.7M annual unit volume is what makes ADAS a $72.1B layer today; robotaxi and trucking revenue stay small until fleet sizes climb past five-figure levels.
Operators should track per-mile safety disclosure standards quarter over quarter, watch regional regulatory deltas (US SGO, EU AI Act, China L3 expansion), and discount any competitor that reports headline ride counts without unit economics. Investors should pay for fleet, regulatory footprint, and disclosed margin per ride. Model claims without those three are vapor.
Every figure in this article is sourced and dated. We’ll refresh as new earnings calls and regulatory updates land in 2026. Our coverage of monthly AI funding trends tracks adjacent capital movements as they happen.
Frequently Asked Questions
How big is the AI in transportation market in 2026?
Precedence Research sized the global AI in transportation market at $5.53 billion in 2025, projecting growth to $34.83 billion by 2034 at 22.7% CAGR. That figure covers the AI software and services layer only; ADAS adds another $72.1 billion and AI in supply chain adds $14.49 billion when measured as adjacent markets.
How many robotaxi rides does Waymo do per week?
Waymo crossed 450,000 weekly paid rides by December 2025, up from 250,000 weekly rides reported on Alphabet’s Q1 2025 earnings call (April 24, 2025). Cumulative rider-only autonomous miles reached 170.7 million by March 2026.
How many cumulative miles has Tesla FSD driven?
Tesla’s FSD (Supervised) fleet crossed 9 billion cumulative miles by April 2026, adding about 1 billion miles a month at its current run-rate. Annual FSD miles in 2025 totaled 4.25 billion (up from 2.25 billion in 2024), split 64% highway and 36% city streets. These are supervised miles with a human driver legally responsible, not driverless miles.
How safe are Waymo robotaxis compared to human drivers?
A peer-reviewed Traffic Injury Prevention study of 56.7 million Waymo rider-only miles found 79% fewer any-injury crashes and 85% fewer suspected serious-injury-or-worse crashes than human-driver benchmarks in the same cities. Waymo’s March 2026 update on 170.7 million miles reported 92% fewer crashes causing serious or fatal injuries (Waymo internal data, not peer-reviewed).
How many autonomous vehicles does Apollo Go operate?
Baidu’s Apollo Go operates 1,000+ vehicles across 22-26 cities including Dubai and Abu Dhabi, with 3.4 million driverless rides in Q4 2025 and 20 million+ cumulative public rides by February 2026. Weekly ride volume peaked above 300,000.
Has China approved Level 3 autonomous driving?
Yes. MIIT granted China’s first L3 conditional permits in December 2025 to the Changan Deepal SL03 and BAIC Arcfox, valid on geofenced expressway sections in Chongqing (50 km/h max) and Beijing (80 km/h max). These are the first L3 approvals for non-pilot passenger cars on Chinese public roads.
How many AV test miles were logged in California in 2025?
The California DMV reported AV permit holders logged more than 9 million test miles between December 1, 2024 and November 30, 2025, a record annual high. The figure covers all permit holders, not just one company.
Methodology
This analysis pulls from primary sources covering AI in transportation activity through April 2026.
Data sources: Company investor relations releases, SEC filings, NHTSA Standing General Order data, California DMV annual disengagement reporting, peer-reviewed Traffic Injury Prevention research, MIIT regulatory announcements, EU AI Act implementing guidance, and analyst firm market sizing from Precedence Research and MarketsandMarkets.
Time period: Statistics span April 2025 through April 2026, with each figure dated to its primary source.
Scope filters: We focused on disclosed, sourced numbers. Aggregate market estimates without primary methodology are flagged as analyst-firm projections. Robotaxi operators are reported using their own disclosure conventions (rides, miles, fleet size); these aren’t directly comparable across operators.
Currency: All amounts in USD. Pony.AI’s $16.6 million 2025 robotaxi revenue and Mobileye’s $1.894 billion 2025 revenue are reported in USD per their SEC filings.
Limitations: Tesla FSD miles are supervised, not driverless, and aren’t directly comparable to Waymo’s rider-only mileage. NHTSA SGO crash reports record incidents involving an AV regardless of fault. Per-mile safety statistics are most meaningful when matched to the same operating cities, time periods, and weather conditions. Aviation AI statistics are outside the scope of this article.
For more on Bot Memo’s data methodology and how we track AI funding rounds, see our primer on what’s in a Bot Memo report.
About the author
Editorial Staff
The Editorial Staff at Bot Memo is a team of writers, analysts, and AI agents dedicated to mapping the global AI startup ecosystem. Led by Chintan Zalani, the team tracks thousands of funding rounds, classifies companies across verticals, and distills it all into actionable intelligence for investors and founders.
