Waabi Raises $750 Million for Physical AI: Canada's Largest Tech Investment

Hello HaWkers, while the world talks about chatbots and LLMs, a Canadian startup has just raised $750 million for something different: AI that operates in the physical world. Waabi, founded by legendary researcher Raquel Urtasun, raised the largest tech investment round in Canadian history.

The focus? Autonomous trucks and robotaxis. The technology? A radically different approach called Physical AI. Let's understand why investors are betting so heavily on this.

What Is Physical AI

Physical AI is the next frontier after LLMs.

Definition

LLMs (what we have today):
- Processes text, code, images
- Operates in the digital world
- Input: data, Output: data

Physical AI (the next wave):
- Processes the real world in real time
- Operates in the physical world
- Input: sensors, Output: physical actions

The Fundamental Difference

# Traditional LLM
def chatbot_response(text_input):
    # Processes text
    # Generates text
    return generated_text

# Physical AI
def autonomous_vehicle(sensor_data):
    # 360 degrees of cameras
    # Lidar, radar, ultrasound
    # GPS position, maps
    # Prediction of other vehicles' behavior
    # Decision in milliseconds
    return steering_angle, acceleration, braking

The difference? LLM errors generate bad text. Physical AI errors can have real physical consequences.

Waabi's Revolutionary Approach

What makes Waabi different from other autonomous vehicle companies.

One Brain, Multiple Vehicles

Traditional approach (Tesla, Waymo):
- Specific model for each vehicle type
- Truck = different system than car
- Massive retraining for each context

Waabi approach:
- A single AI "brain"
- Works on trucks, cars, robotaxis
- Transfers knowledge between vehicles

Simulation at Scale

# The secret: billions of virtual miles

waabi_simulation = {
    "miles_simulated": "10+ billion",
    "scenarios_tested": [
        "Pedestrian crossing unexpectedly",
        "Car cutting off truck",
        "Emergency vehicle approaching",
        "Construction zone navigation",
        "Adverse weather conditions",
        "Night driving",
        "Sensor failure scenarios"
    ],
    "physics_accuracy": "99.9%",
    "cost_vs_real_testing": "0.001%"
}

# Waabi can simulate years of driving in hours

Raquel Urtasun: The Founder

Background:
- Former head of autonomous vehicle research at Uber
- CS Professor at the University of Toronto
- One of the most cited researchers in computer vision
- Pioneer in deep learning for autonomous driving

Why it matters:
- Unmatched technical credibility
- Can attract the best researchers
- Investors trust the technical vision

The Investment Numbers

Let's get into the round details.

Series C of $750 Million

Amount: $750 million USD
Type: Series C
Led by: Khosla Ventures, NVIDIA
Participation: a16z, Tiger Global, Uber

Use of funds:
- Test fleet expansion
- Engineer hiring
- Manufacturer partnerships
- Simulation infrastructure

Estimated Valuation

Previous rounds:
- Series A: $83M (2021)
- Series B: $200M (2023)
- Series C: $750M (2026)

Estimated valuation: $5-8 billion

Why Investors Are Excited

potential_market = {
    "trucking_logistics": "$800 billion/year (US only)",
    "robotaxi": "$300 billion/year (projected 2030)",
    "last_mile_delivery": "$150 billion/year",
    "industrial_autonomous": "$100 billion/year"
}

# Waabi can address all of them with a single platform

Autonomous Trucks: The First Market

Why trucks before passenger cars.

The Trucking Problem

Driver crisis:
- The US has a deficit of 80,000+ truck drivers
- Average driver age: 55 years
- Few young people want the profession
- Labor costs rising 8%/year

The Autonomous Solution

Benefits of autonomous trucks:
- Operates 24/7 (trucks don't sleep)
- Long-distance routes (highway driving)
- More predictable than urban environment
- Clear ROI for fleet operators

Hub-to-Hub Model

Waabi's operating model:

[Warehouse A] → Human driver → [Highway Hub]
                                      ↓
                            [Autonomous Truck]
                            (Highway, 500+ miles)
                                      ↓
                               [Highway Hub]
                                      ↓
                        Human driver → [Warehouse B]

Advantages:
- Humans handle urban "first/last mile"
- AI does the easy and repetitive part
- Gradual transition, not disruptive

The Technology Behind It

Technical aspects that make Waabi work.

Sensor Fusion

# Fusion of multiple sensors
class WaabiPerception:
    def __init__(self):
        self.sensors = {
            "cameras": 12,  # 360 degrees, multiple distances
            "lidar": 4,     # Long, medium, short range
            "radar": 6,     # Works in rain/snow
            "ultrasonic": 8 # Short distance
        }

    def perceive(self):
        # Each sensor has different strengths
        camera_data = self.cameras.detect_objects()
        lidar_data = self.lidar.measure_distances()
        radar_data = self.radar.detect_movement()

        # Fusion creates a more reliable view
        fused_perception = self.fuse(
            camera_data,
            lidar_data,
            radar_data
        )

        return fused_perception

Prediction Engine

# Predicting the behavior of other agents
class BehaviorPrediction:
    def predict_trajectories(self, detected_objects):
        predictions = []

        for obj in detected_objects:
            if obj.type == 'vehicle':
                # Where will this car be in 3 seconds?
                trajectory = self.predict_vehicle(
                    current_position=obj.position,
                    current_velocity=obj.velocity,
                    road_context=self.map.get_context(obj.position),
                    historical_behavior=obj.track_history
                )

            elif obj.type == 'pedestrian':
                # Pedestrians are more unpredictable
                trajectory = self.predict_pedestrian(
                    current_position=obj.position,
                    body_orientation=obj.orientation,
                    crosswalk_nearby=self.map.check_crosswalk(obj.position)
                )

            predictions.append(trajectory)

        return predictions

Safety Architecture

# Multiple layers of safety
class SafetySystem:
    def __init__(self):
        self.layers = [
            PrimaryAI(),           # Primary driving AI
            SecondaryAI(),         # Independent backup AI
            RuleBasedSystem(),     # Deterministic rules
            HardwareFailsafe()     # Last resort in hardware
        ]

    def compute_action(self, perception):
        primary_action = self.layers[0].compute(perception)
        secondary_action = self.layers[1].compute(perception)

        # Cross-verification
        if self.actions_agree(primary_action, secondary_action):
            # Rule layer validates
            if self.layers[2].validate(primary_action):
                return primary_action

        # Disagreement = conservative action
        return self.safe_stop()

Comparison With Competitors

Where Waabi stands in the market.

Waymo (Alphabet/Google)

Waymo:
- Largest robotaxi fleet
- Operating in Phoenix, SF
- Billions invested
- Focus: urban robotaxi

Waabi vs Waymo:
- Waabi focuses on trucking first
- Waabi uses more simulation, less real fleet
- Waabi has a more asset-light approach

Tesla FSD

Tesla:
- Largest dataset in the world (billions of miles)
- Vision-only (no lidar)
- Sells to consumers
- Approach: scale over precision

Waabi vs Tesla:
- Waabi uses lidar + cameras
- Waabi focuses on B2B, not consumer
- Waabi prioritizes safety over scale

Aurora

Aurora:
- Also focuses on trucking
- Partnership with PACCAR, Volvo
- Merged with Uber ATG

Waabi vs Aurora:
- Direct competitors in trucking
- Waabi has more recent tech
- Aurora is closer to commercialization

Implications for Developers

What the rise of Physical AI means for software developers.

New Career Opportunities

physical_ai_careers = {
    "ML Engineer - Perception": {
        "skills": ["Computer Vision", "Point Cloud Processing", "Sensor Fusion"],
        "salary": "$200-350k"
    },
    "Simulation Engineer": {
        "skills": ["Physics Engines", "Game Engines", "Distributed Systems"],
        "salary": "$180-300k"
    },
    "Safety Engineer": {
        "skills": ["Formal Verification", "Fault Analysis", "Testing"],
        "salary": "$170-280k"
    },
    "Robotics Software Engineer": {
        "skills": ["ROS", "C++", "Real-time Systems"],
        "salary": "$160-280k"
    }
}

Technology Stack

Dominant languages:
- C++ (critical performance)
- Python (ML, prototyping)
- Rust (safety + performance)

Frameworks:
- ROS/ROS2 (robotics middleware)
- PyTorch (ML models)
- CUDA (GPU acceleration)
- Unreal/Unity (simulation)

Infrastructure:
- Kubernetes (orchestration)
- Kafka (data streaming)
- S3/GCS (petabytes of data)

How to Get Into the Field

# Path to Physical AI

# 1. Fundamentals
- Linear Algebra, Probability, Statistics
- Computer Vision basics
- Machine Learning fundamentals

# 2. Specialization
- Robotic perception (SLAM, sensor fusion)
- Motion planning algorithms
- Real-time systems

# 3. Practice
- Projects with ROS
- Kaggle competitions (autonomous driving)
- Contribute to CARLA simulator

# 4. Differentiator
- Experience with real hardware
- Understanding of safety standards (ISO 26262)
- Publications at conferences (CVPR, ICRA)

The Future of Physical AI

Where this is heading.

Beyond Vehicles

Physical AI goes far beyond cars:

Agriculture:
- Autonomous tractors
- Robotic harvesting
- Monitoring drones

Logistics:
- Autonomous warehouses
- Drone deliveries
- Last-mile robots

Construction:
- Autonomous equipment
- Robot inspection
- 3D printing of structures

Healthcare:
- Advanced surgical robots
- Assistive exoskeletons
- Care robots

Expected Timeline

2026-2027:
- Autonomous trucks on limited routes
- Robotaxis expand (Waymo, Cruise)
- Waabi begins commercial operations

2028-2030:
- Autonomous trucks go mainstream
- Robotaxis in more cities
- Autonomous deliveries become common

2030+:
- Level 4 personal autonomous cars
- Complete logistics transformation
- Global regulation established

Conclusion

The $750 million investment in Waabi signals that Physical AI is the next great frontier. After conquering the digital world with LLMs, AI is now advancing into the physical world.

For developers, this opens up a whole new universe of opportunities. The skills are different: less web, more embedded systems; fewer APIs, more sensors; less "eventual consistency", more "real-time or death".

The competition for Physical AI talent is just beginning. Those who position themselves now will be at the forefront of the next revolution.

If you want to understand more about how AI is transforming different industries, I recommend checking out another article: Snowflake and OpenAI: $200 Million Deal where we explore how AI is transforming the world of enterprise data.