SetupAI
繁
SourceStalecollected in 20m

Cathie Wood: Tesla Robotaxis Set for Massive Growth

Read original on Bloomberg Technology
#embodied-ai#autonomous-driving#transportation

Understand why top investors view embodied AI in transportation as the next major revenue frontier.

30-Second TL;DR

What Changed

Transportation is identified as the largest revenue opportunity for embodied AI.

Why It Matters

This signals a shift in investment focus toward physical-world AI applications, suggesting that autonomous fleets will become a major driver for AI hardware and software demand.

What To Do Next

Monitor Tesla's FSD (Full Self-Driving) progress and API updates to understand how autonomous fleet data is being leveraged for real-world navigation.

Who should care:Founders & Product Leaders

Key Points

  • •Transportation is identified as the largest revenue opportunity for embodied AI.
  • •Cathie Wood highlights Tesla as a leader in the autonomous vehicle space.
  • •Robotaxi services are expected to scale rapidly in the near future.
Key numbers$8$10$3490%

Deep Insight

Background and context from public sources — not the original article. 22 sources cited.

Enhanced Key Takeaways

  • •Cathie Wood's Ark Invest projects the global autonomous taxi market could reach $8 to $10 trillion by 2030, with some Ark analysts suggesting up to $34 trillion, and anticipates robotaxi operations could constitute 90% of Tesla's total enterprise value by 2029.
  • •Ark Invest predicts Tesla's robotaxi platform could achieve non-GAAP profit margins approaching 90% due to its software-driven recurring revenue model and near-zero incremental cost per ride.
  • •Tesla's vertical integration is identified as a key competitive advantage, potentially allowing its robotaxi services to operate at a significantly lower cost of approximately 25 cents per mile compared to rivals like Waymo, which are projected to have 50% higher costs.
  • •Tesla has aggressively expanded its robotaxi program, confirming deployment to seven major US cities in the first half of 2026, with Austin already operating fully driverless (unsupervised) services since December 2025.
  • •Tesla's Full Self-Driving (FSD) system, which underpins its robotaxi ambitions, has transitioned to a vision-only, end-to-end neural network architecture (FSD v12), learning directly from raw camera inputs and human driving behavior.

Competitor Analysis

Tesla
Technology Approach
Vision-only, end-to-end neural network (FSD v12), leverages existing vehicle fleet.
Operational Status / Key Features
Expanding to 7 new US cities in H1 2026; Austin operating unsupervised.
Projected Cost / Mile (at scale)
~25 cents
Waymo (Alphabet)
Technology Approach
LiDAR + radar + high-definition maps, purpose-built vehicles.
Operational Status / Key Features
Undisputed operational leader in the US; mature, consistent, offers paid driverless rides.
Projected Cost / Mile (at scale)
~38 cents (50% higher than Tesla's projection)
Zoox (Amazon)
Technology Approach
Purpose-built robotaxi vehicles.
Operational Status / Key Features
Newer, experiential, operating in smaller, defined areas (e.g., San Francisco).
Projected Cost / Mile (at scale)
Not specified
Apollo Go (Baidu)
Technology Approach
Not fully detailed, but uses autonomous driving technology.
Operational Status / Key Features
Strong presence in China, over 20 million cumulative trips and 190 million fully driverless kilometers by February 2026.
Projected Cost / Mile (at scale)
Not specified

Technical Deep Dive

  • Vision-Only Philosophy: Tesla's FSD system relies exclusively on cameras for perception, a controversial decision that contrasts with competitors using LiDAR and radar. The thesis is that if AI can see and reason like a human, it can scale globally without expensive sensors or pre-mapped environments.
  • Neural Network Stack: FSD is a pipeline of interconnected neural networks. It uses multi-camera input, synchronizing feeds into time-aligned sequences. Convolutional neural networks (CNNs) extract spatial features, with a shift towards video-based neural networks that analyze temporal continuity to understand motion and object persistence across frames.
  • Vector Space Representation: Perception data is converted into a vector space representation, where lanes become splines and vehicles are dynamic objects with trajectories, enabling better reasoning about the environment.
  • Occupancy Networks: A key breakthrough involves occupancy networks that predict which 3D space is occupied versus free, even when objects are partially hidden, improving occlusion handling and depth estimation.
  • End-to-End Learning: Tesla's FSD version 12 (v12) represents a fundamental shift to a fully neural, end-to-end learning architecture. The system directly calculates control commands (accelerator, brakes, steering) from raw camera inputs, learning from billions of kilometers of real driving situations and human reactions, rather than relying on thousands of hard-coded rules.
  • Training Infrastructure: Training FSD requires immense computational power. Tesla has historically used NVIDIA GPUs and has also developed its custom supercomputer, Dojo, optimized for video training workloads and neural network scaling.
  • Hardware Evolution: Tesla has progressed through several hardware generations (AP1, HW2, HW2.5, HW3, HW4). Hardware 3 (FSD Computer) was released in 2019, and Hardware 4 (FSD Computer 2), which is roughly three times more powerful and includes a new camera suite and upgraded radar, began rolling out in mid-2023.

Future ImplicationsAI analysis grounded in cited sources

Tesla's robotaxi service will achieve widespread national unsupervised operation in the US by the end of 2026.
Elon Musk has recently announced plans for nationwide expansion of unsupervised robotaxis in the US later this year, following successful implementation in Texas.
The cost per mile for robotaxi services will significantly undercut traditional ride-hailing, leading to mass adoption.
Ark Invest projects autonomous ride costs could fall to as low as 25 cents per mile, a substantial reduction compared to current ride-hailing services like Uber, which cost over $3 per mile.
Tesla's vertical integration and vision-only approach will enable it to achieve a dominant market share in the global robotaxi market.
Cathie Wood's core thesis is that Tesla's control over its manufacturing and supply chain, combined with its camera-centric AI, will result in the lowest cost structure and superior scalability, positioning it to capture a significant portion of the autonomous mobility market.

Timeline

2015-10
Tesla Autopilot launched, providing advanced driver-assistance features.
2020-10
Tesla launched the Full Self-Driving (FSD) Beta Program, making its self-driving software available on city streets for a limited number of consumers.
2022-11
Elon Musk announced the wide release of Full Self-Driving Beta to anyone in North America.
2025-06
Tesla launched its robotaxi pilot program in Austin, Texas, initially with safety monitors.
2025-12
The Austin robotaxi service began offering unsupervised (fully driverless) rides.
2026-01
Tesla announced plans to expand its Robotaxi program to seven new major US metropolitan areas within the first half of 2026.
Read original article on Bloomberg Technology

Weekly AI Recap

Read this week's curated digest of top AI events →

PostLinkedIn

AI-curated news aggregator. All content rights belong to original publishers.
Original source: Bloomberg Technology ↗

This is a summary, not the original. Read the source, or get the weekly briefing.

The weekly digest

One email a week. Unsubscribe anytime.