AI Infrastructure

SoftBank's Roze: Robots to Build AI Data Centers at $100B

SoftBank is assembling Roze, a robotics company to build US data centers, bundling ABB Robotics and energy assets and targeting a $100B IPO as soon as H2 2026.

SoftBank's Roze: Robots to Build AI Data Centers at $100B — article cover
On this page6 SECTIONS
  1. What Roze Is Being Built to Do
  2. The Case for a $100 Billion Valuation
  3. Internal Doubts and the Ghost of Zume
  4. Where Roze Fits in Son’s Empire
  5. Why the Rest of the Industry Should Care
  6. Sources

On April 29, 2026, the Financial Times reported that SoftBank is assembling a standalone company called Roze, which would deploy autonomous robots to speed up US data center construction — and is already eyeing an IPO. Some executives want the listing as early as the second half of 2026, at a targeted valuation of roughly $100 billion. CNBC and the Wall Street Journal followed; SoftBank declined to comment.

The story rests on a loop: AI needs data centers, data centers need construction labor, and skilled labor is the scarcest input in the American market. Masayoshi Son’s answer is to close the loop — have robots build the houses that AI lives in, then take that business public. It is a bet on robotics as the enabling layer for the entire AI buildout, not on any single robot product.

What Roze Is Being Built to Do

Per the FT and CNBC reports, Roze would be an AI and robotics company whose business is building data centers, using autonomous robots to make US server-farm construction more efficient. The consolidation direction is already visible: SoftBank could fold in its existing energy, land, and infrastructure assets, and ABB Robotics — which SoftBank agreed to acquire in October 2025 — is on the list of pieces to bundle.

The Case for a $100 Billion Valuation

The $100 billion figure is not picked from thin air. If Roze absorbs ABB Robotics plus SoftBank’s energy and land holdings, one listed company would own the full stack of AI infrastructure’s supply chain: the robots that build the facilities, the power that runs them, and the ground they sit on. Both the FT and CNBC report that Son is personally driving the plan.

The 2026 AI listing wave helps the audacity. From Cerebras re-filing for its IPO to a parade of infrastructure names, investor appetite for anything that monetizes AI demand is at a high.

Internal Doubts and the Ghost of Zume

Not everyone inside SoftBank is convinced. The FT notes that executives have doubts about both the valuation and the IPO timeline, with Middle East conflict uncertainty cited as a variable. TechCrunch adds an awkward precedent: SoftBank once poured hundreds of millions of dollars into Zume, the robot pizza startup that went bankrupt in 2023. “Robots doing physical-world work” is a business SoftBank has already fumbled once.

On the day the story broke, SoftBank shares slipped 0.9%, still up more than 18% year to date.

Where Roze Fits in Son’s Empire

Roze is less a standalone bet than the construction crew for Son’s AI portfolio. SoftBank has committed more than $30 billion to OpenAI; the Stargate project with OpenAI and Oracle is billed at $500 billion; and the group is building its own large-scale data centers, including in Ohio. The Vision Fund booked a $2.4 billion gain in the December quarter, helped largely by OpenAI — evidence that the strategy is already showing up in the numbers. When every line of the empire needs more buildings finished faster, “who builds them” stops being a procurement question and becomes group strategy. Son’s answer: own a company that builds.

Why the Rest of the Industry Should Care

Three things to watch. First, if construction robots actually show up on site, one of the biggest US data center bottlenecks — the skilled-trade shortage — gets partially relieved, and gigawatt-class projects have a shot at tighter timelines. Second, Roze’s bundle of robots-plus-energy-plus-land could become a template other conglomerates copy for their own listings. There is a parallel elsewhere: Jeff Bezos co-founded Project Prometheus, which plans to buy industrial firms and modernize them with AI, as TechCrunch reported in March. The wealthy are converging on the same idea — AI applied to heavy industry is the next margin pool. Third, mind the execution risk: the valuation is a target, the timeline is a hope, and even insiders are arguing about both. A company that has not officially launched is best treated as a stress test of Son’s storytelling, not a done deal.

Sources

AI-assisted summary compiled from the sources above, reviewed by a human before publishing.

FOUND_THIS_USEFUL?

Support more practical AI articles, tutorials, and build notes.

BUY_ME_A_COFFEE
SHAREXEMAIL