Strategy

tl;dr Everyone sees humanoids as a way to automate physical tasks. We see them as a vessel to enable remote physical work.

We watch stunning videos on X of humanoids folding cloth. But when we walk around the city, we see thousands of cars and a startling absence of humanoids. Why?

Autonomous humanoids are years away, and the reason is a massive data gap.

First, there's the staggering problem of quantity: the largest robotics dataset (video + movement) is a single human-year of experience. We need closer to 100 million human years to even begin covering the real world.

But even if we had all that data, it wouldn't be enough. The deeper problem isn't quantity; it's quality. The industry is fixated on kinematics - what a robot's movement looks like. This misses the most critical ingredient for physical intelligence: touch. Video captures what a task looks like, but can't feel the firmness of a tomato or the tension in a wire.

Today's methods won't close this gap. They are dead ends:

• Lab teleoperation: Too expensive, too sterile, not scalable.
• Simulation: Can't model the infinite complexities of real-world physics.
• Video: Captures motion but misses the physics. It sees, but it doesn't feel.

There is only one way to generate internet-scale physical data: provide a utility first. It is the only model that has ever worked. Tesla, for example, built a compelling car that people wanted to drive, generating real-world data at an unprecedented scale.

We therefore don't try to solve the physical intelligence problem in robotics without the key ingredients first. We build our business by letting humans and AI each do what they already do best. We are building the world's most advanced teleoperation stack that allows for seamless full-body control of humanoids while sitting at a desk with complete sensory feedback. Our AI co-pilot handles workflow reasoning - translating complex tasks into simple guidance for the operator, cutting training from months to days. Meanwhile the human provides the one thing AI lacks: true physical intelligence. This unlocks the ability to deploy humanoids at massive scale for economically valuable tasks today.

We deploy low-cost humanoids-as-a-service, each piloted in real-time by a human operator from a global workforce. This solves our customers' critical labor shortages today, and every completed task generates the one thing the industry lacks: rich, multi-modal physical data (synchronized video, motor commands, and force-feedback) as a natural byproduct. This allows us to generate physical intelligence data at unprecedented scale.

This brings us to the long game. The industry is waiting for a 'ChatGPT moment' in robotics. We are not. Our strategy wins whether that breakthrough comes tomorrow or in ten years, because our business is not a binary bet on full autonomy.

Even if that breakthrough never comes, we will have unlocked a world where physical work is no longer tied to location. A construction worker in Manila can operate a crane in Dubai. A technician in Ohio can repair machinery on a cruise ship. We will have built a world where opportunity is truly global.

By fusing human adaptability with AI's knowledge and robotics' location-independence, we are creating a superhuman, no longer limited by biological form. We are merging human intellect with a synthetic body to build something more.

We are building the next generation of humans: Homo stellaris.

Leander & Satrajit