$1.7 Billion Bet on 'Physical AI': What It Means When Large Models Get Hands and Feet
Atoms, founded by Uber co-founder Travis Kalanick, raised $1.7 billion to build AI that can see, touch, and move in the real world. Here's what 'Physical AI' actually means and why it matters.
$1.7 Billion for an AI That Can Actually Do Things
Atoms, a company founded by Uber co-founder Travis Kalanick, recently closed a $1.7 billion funding round with a clear goal: build "Physical AI."
In short, this means AI that can do more than chat or write code — it can see, touch, and move. Think controlling robots, driving cars, operating robotic arms, and even understanding the laws of physics in the real world.
This is quite different from the chatbots most people know, like ChatGPT or China's Wenxin Yiyan (Baidu's large language model). Those are "digital AI" — they live on screens and output text, images, or code. Physical AI, by contrast, is meant to step out of the screen and into the real world, interacting directly with objects, environments, and people.
It's not just about building a bigger "brain." It's about giving AI "eyes" and "hands."
What Exactly Is Physical AI?
At its core, Physical AI combines large AI models with physical systems. You can think of it in a few ways:
- Embodied AI: Giving AI a body (like a robot) so it can move, grasp objects, and avoid obstacles in real environments.
- Autonomous driving: A self-driving car is itself a "physical AI system" — it must perceive road conditions in real time, make decisions, and control the steering wheel.
- Industrial robots + AI: Traditional robotic arms repeat the same motions. Add AI, and they can adapt to different tasks — like sorting packages of different shapes and sizes.
A concrete example: If you ask a digital AI "pour me a glass of water," it can only list the steps. But a robot powered by Physical AI would walk over, pick up the glass, fill it, and bring it back — all while avoiding your cat along the way.
Another way to look at it: Physical AI isn't just a smarter chatbot. It's about giving AI "physical common sense" — an understanding of gravity, friction, object shapes, and spatial relationships. It's the leap from knowing to doing.
What Are the Big Tech Giants Racing For?
Atoms' $1.7 billion round is not an isolated case. Major tech companies worldwide are pouring money into Physical AI:
- Tesla: Elon Musk is building electric vehicles, a humanoid robot called Optimus, and even a dedicated chip factory (Terafab) for it. Recent quarterly reports show AI and robotics investments have weighed on cash flow — but he's pressing on.
- OpenAI: Plans to invest nearly $750 billion in computing power by 2030, and recently committed $20 billion to a new data center in Georgia, USA.
- AMD: Announced up to $50 billion in investment in AI company Anthropic, along with GPUs delivering 2 gigawatts of compute capacity.
- SpaceX AI: Musk's AI division is expanding data centers in Texas.
The deeper picture: Behind these moves is a shared belief — the marginal returns of purely digital AI are diminishing. No matter how smart a chatbot gets, it can't replace factory assembly lines, logistics warehouses, or in-home services. Physical AI is where AI truly meets the real economy.
The key takeaway: Competition in Physical AI is essentially a three-way race across computing power + hardware + real-world use cases. Whoever can make AI run reliably, cheaply, and at scale in the physical world will win.
How Does This Affect Ordinary People?
You might be thinking: What does this have to do with me? I don't run a robotics company.
Actually, it has a lot to do with you.
In the next 3–5 years, Physical AI could transform these areas:
- Logistics and delivery: Autonomous vehicles and drones will become more common, making deliveries faster and cheaper.
- Home services: Think of robot vacuums, upgraded — able to pick up toys, fold laundry, and assist elderly family members.
- Manufacturing: AI-powered robots in factories can work 24/7, adapt to different tasks, and no longer need to be reprogrammed for every change.
- Healthcare support: Surgical robots become more precise; rehabilitation robots adjust their force based on a patient's condition.
But there are trade-offs: The rollout of Physical AI also means some repetitive physical jobs may be displaced — warehouse sorters, long-haul truck drivers, basic assembly workers, for example.
That said, don't panic: Technological change has never happened overnight. Just as ATMs didn't eliminate bank tellers but changed what they do, Physical AI is more likely to transform jobs rather than eliminate them — for instance, shifting from "driving a truck" to "monitoring an autonomous fleet."
What Should Ordinary People Do?
Facing the wave of Physical AI, there's no need to panic — but it does help to keep your knowledge up to date.
A few suggestions:
- Focus on human–machine collaboration skills: The future workplace isn't "humans vs. AI" — it's "people who use AI vs. people who don't." Learning to work alongside AI tools and robotic systems is a real advantage.
- Watch for skill-transition opportunities: If you work in repetitive physical labor, consider getting ahead of the curve by learning about equipment maintenance, system monitoring, or data analysis — the "human–machine interface" roles of the future.
- Stay rational about capital hype: A $1.7 billion funding round looks impressive, but Physical AI has high technical barriers and long deployment cycles. Don't get swept up in "concept hype" — invest carefully.
A broader perspective: Looking back at history, every breakthrough in "AI + the physical world" has come alongside infrastructure upgrades. Autonomous driving needs 5G and high-definition maps; embodied AI needs low-cost sensors and edge computing. The explosion of Physical AI may depend even more on infrastructure than we realize.
A Few Things to Watch Out For
- Don't confuse "Physical AI" with "Artificial General Intelligence (AGI)": The former is about deploying AI in specific real-world scenarios; the latter means "as capable as a human in every way." AGI remains a long-term goal.
- Hot capital ≠ mature technology: $1.7 billion is a lot, but hardware costs, safety verification, and real-world adaptation for Physical AI are all tough challenges.
- Safety and ethics can't be an afterthought: An AI that can physically act carries far greater consequences if it makes a mistake than a chatbot does — think autonomous vehicle accidents or robotic misoperations. Regulation and standards must keep pace.
The analysis above is for informational purposes only and does not constitute investment advice.
One-sentence summary to share: Physical AI is the key step that turns AI from "able to talk" into "able to act." Over the next few years, it will gradually reshape logistics, manufacturing, and home services — changing not just technology, but how we work.
Join the conversation: Which part of your life do you think Physical AI will change first — delivery, household robots, or something else? Share your take.