SoftBank’s new robotics company is really a data center builder — and it’s already planning a $100B IPO

SoftBank is at it again — spinning up yet another ambitious venture that blurs the line between tech sectors. This time it’s a robotics company that builds data centers. And yes, you read that right: robots building the infrastructure that powers AI, while AI presumably helps design and run those robots.

The company isn’t even public yet, but according to sources familiar with the matter, SoftBank is already eyeing a $100 billion IPO. That’s not a typo. One hundred billion dollars. For context, that would put it in the same ballpark as some of the largest tech IPOs in history, rivaling Alibaba’s $25 billion debut and Saudi Aramco’s $29.4 billion listing — though valuation and raise size are different things.

Let’s step back and look at what SoftBank is actually doing here. The core idea is straightforward enough: data centers are the physical backbone of the AI boom, and building them is expensive, slow, and labor-intensive. SoftBank wants to automate that process using robotics. Think autonomous construction vehicles, robotic arms for rack installation, and AI-driven logistics to manage the supply chain. It’s not exactly sexy, but it’s the kind of mundane, high-volume work that robotics excels at.

What makes this interesting — and a bit eyebrow-raising — is the valuation. $100 billion for a company that hasn’t shipped a single data center yet? That’s either visionary or delusional, depending on your tolerance for SoftBank’s track record. The firm has a history of swinging for the fences with WeWork, Arm, and various SPACs. Some paid off, some didn’t. This feels like another moonshot, but one with more concrete revenue potential.

I’ve seen a few startups try this approach before. There’s a company called Built Robotics that does autonomous earthmoving equipment for construction sites. Another called Dusty Robotics focuses on construction layout. But none of them are aiming to build entire data centers from scratch. SoftBank’s scale is different — they’re not just selling robots; they’re selling a complete solution. They want to own the process end-to-end, from pouring concrete to plugging in servers.

That vertical integration is smart, but it also means they’re competing with established construction firms, data center operators like Equinix and Digital Realty, and potentially even hyperscalers like Amazon and Google who already build their own facilities. Those hyperscalers have deep pockets and their own automation R&D. SoftBank’s play here is to move faster and offer a standardized, robot-built product that’s cheaper and faster than traditional methods.

Will it work? Hard to say. The timing is right — demand for data centers is exploding thanks to AI workloads, and labor shortages in construction are real. But $100 billion is a lot of faith to place in a company that hasn’t proven its model yet. I suspect SoftBank is banking on the narrative: “robots building the AI infrastructure” is a story that sells, especially to investors who are already drunk on AI hype.

One thing I’ll give SoftBank credit for: they’re not afraid to think big. Most robotics companies are happy to sell a few thousand units and call it a day. SoftBank is trying to reshape an entire industry. Whether that ends in triumph or a spectacular crash remains to be seen, but it won’t be boring.

For now, I’m watching this one closely. If the IPO materializes at that valuation, it’ll be a bellwether for how much appetite there really is for AI-adjacent infrastructure plays. If it doesn’t, well, SoftBank has a habit of floating ambitious numbers and then walking them back. Either way, the idea of robots building data centers is no longer science fiction — it’s a business plan.