Solving a Key 'Data Problem' for Humanoid Robots

Editor's note: When most people think about AI, they think of chatbots and virtual assistants – things that live inside a computer. But as Jeff Brown from our corporate affiliate Brownstone Research explains, AI is also improving machines in the real world. In today's issue, adapted from Jeff's daily e-letter The Bleeding Edge, he discusses how this shift started with electric vehicles... and what it means for humanoid robots.

One of Tesla's (TSLA) key advantages in the autonomous-driving industry has been its data repository of real-world video, which it used to train its Full Self-Driving ("FSD") software.

Both the repository and Tesla's strategy for collecting that data distinguish it from its competitors.

You see, Tesla has designed and manufactured all of its electric vehicles ("EVs") with self-driving capabilities.

Every new Tesla EV comes with an FSD computer and cameras to capture a 360-degree view of its driving conditions.

Naturally, this increases the cost of manufacturing each EV. But having AI hardware at the ready in every EV enables Tesla to upsell many of its customers to self-driving software after their initial purchase or lease.

And as I'll explain, the real value Tesla has generated is the ability to collect vast amounts of real-world data...

Building the Data Foundation

This all began shortly after Tesla installed its Autopilot systems into EVs in late 2014. It was the less-advanced software that preceded FSD.

That's how, prior to FSD, Tesla managed to collect about 9 billion miles' worth of data from the millions of Teslas already on the road.

When a driver returned home, their EV would connect to Wi-Fi and automatically upload the data from their trip to Tesla. Those 9 billion miles became the foundation for the early training of Tesla's vision-based FSD software.

And as Tesla's "fleet" of EVs capable of autonomous driving got more advanced, its data-collection efforts grew.

Tesla has now collected around 14.6 billion miles' worth of FSD data... 5.7 billion of them from urban environments.

By way of comparison, Google's Waymo autonomous vehicles have collected data on only 220.6 million autonomous miles. That's about 1.5% of Tesla's total.

This is one of the key reasons that Tesla's Cybercabs and Robotaxis are so incredible. Those ride-hailing services use the same FSD software as Tesla's other vehicles.

And this software also became the foundation for Tesla's humanoid robot – Optimus.

The Index of Autonomy

There are more parallels between EVs and robots than you might think. After all, a self-driving car is nothing more than a robot on four wheels...

Which raises the question: How can companies collect the data needed to train humanoid robots without millions of prototypes operating in the real world?

It's a classic chicken-or-the-egg problem. EVs are one thing... Even when they weren't intelligent and capable of autonomous driving, they could – and did – operate like normal cars and collect useful data.

But a humanoid robot without intelligence and knowledge of the real world would serve no purpose at all.

With FSD as a foundation − and thousands of Optimus robots now working in Tesla's offices and factories collecting data – Tesla has a major advantage in this market.

But what about other robotics companies?

Recently, I was excited to see an announcement made by leading humanoid-robotics firm Figure AI. It was about Index – its program to build "the world's largest and most diverse physical dataset."

Index had been kept under wraps for about four months before the program was made public. It's a clever way for Figure to solve its data problem.

Index is a smartphone app that enables users – or, as Figure calls them, Creators – to get paid for recording themselves performing tasks and uploading videos back to Figure.

Figure's Index App | Source: App Store @Figure INDEX

Figure creates a list of tasks that it wants to collect data on. Wearing the video-recording device Figure provides, Creators can earn income by simply completing chores throughout the day.

As of August, Figure had already paid out $15 million to Creators for their data-collection efforts. And in September, Figure's CEO announced that Index had crossed 86,000 weekly active users.

It's pretty impressive what Figure has done in only six months. It has already collected more than 22 million videos of real-world tasks. And it has established a broad reach.

Each point of light in the image below represents the location where a video has been collected...

Creators Video Uploads | Source: Figure AI

Figure has committed to spending $1 billion in the next 12 months on data collection and computational resources to train its robots.

This is a massive sum for a startup that just formed in 2022. But it raised $1.5 billion in May 2025 in its Series C round, and it will almost certainly raise another large round within the next six months to continue accelerating growth.

Robots Will Walk Among Us

Robotics is a massive market opportunity worth trillions of dollars. Elon Musk has said that about 80% of Tesla's total value will one day be based on its Optimus business.

No other technology can easily solve labor shortages or transform labor markets, thereby improving quality of life on a global scale.

Industry, logistics, distribution, healthcare, agriculture, security, manufacturing, education, and many other sectors will be transformed by intelligent, general-purpose humanoid robots.

And the world will adopt them as quickly as they can be manufactured.

Mark my words: They will walk among us...

Regards,

Jeff Brown

Editor's note: Jeff has a fascinating secret to making money in AI – and right now, he's stepping forward with his latest market discovery. A new AI superchip 50 times faster than Nvidia's could change daily life as we know it... handling everyday tasks for millions of people. And he says the firm behind it could generate 1,000% gains for investors who get in before a pivotal catalyst on October 28.