SoftBank Eyes $100 Billion IPO for Roze, a New AI Robotics Unit to Build Data Centers

--- headline: "SoftBank Eyes $100 Billion IPO for Roze, a New AI Robotics Unit to Build Data Centers" slug: softbank-roze-100b-robotics-ipo category: business story_number: "06" date: 2026-05-05 author: The Vault AI sources: - name: CNBC url: https://www.cnbc.com/2026/04/30/softbank-roze-ai-robotics-ipo-100-billion-ft-report.html domain: cnbc.com - name: TechCrunch url: https://techcrunch.com/2026/04/29/softbank-is-creating-a-robotics-company-that-builds-data-centers-and-already-eyeing-a-100b-ipo/ domain: techcrunch.com - name: Yahoo Finance url: https://finance.yahoo.com/sectors/technology/articles/softbank-plans-100b-u-ipo-115700056.html domain: finance.yahoo.com - name: Tom's Hardware url: https://www.tomshardware.com/tech-industry/softbank-plans-robotics-and-ai-firm-in-the-us-to-build-data-centers-aims-for-usd100-billion-valuation-and-an-ipo-this-year domain: tomshardware.com - name: Benzinga url: https://www.benzinga.com/markets/ipos/26/04/52168304/softbank-eyes-2026-ipo-for-new-ai-robotics-venture-targets-up-to-100-billion-valuation-report domain: benzinga.com - name: NewsCase url: https://www.newscase.com/softbanks-grand-ai-wager-a-robot-builder-a-debt-pile-and-a-100-billion-ipo-dream/ domain: newscase.com ---

# SoftBank Eyes \$100 Billion IPO for Roze, a New AI Robotics Unit to Build Data Centers

Masayoshi Son is placing his biggest bet yet on the physical backbone of artificial intelligence. SoftBank Group is preparing to spin out and publicly list a new U.S.-based company called Roze that will deploy autonomous robots to construct data centers, targeting a valuation of roughly \$100 billion in what would rank among the largest pre-revenue technology IPOs in history, according to a Financial Times report confirmed by multiple outlets late last month.

The Vision: Robots That Build the AI Factory

The premise behind Roze is deceptively simple and enormously ambitious. The global AI boom has created insatiable demand for data center capacity, but the construction industry cannot keep pace. Skilled labor shortages, permitting delays, and the sheer scale of planned buildouts—measured in gigawatts, not square feet—have turned physical infrastructure into the binding constraint on AI progress. Roze aims to break that bottleneck by automating welding, assembly, heavy lifting, and other construction tasks with robots that can operate around the clock, bypassing the shift constraints of human crews.

"SoftBank's next frontier is Physical AI," Son said when he announced the $5.4 billion acquisition of ABB's robotics division in October 2025. "Together with ABB Robotics, we will unite world-class technology and talent under our shared vision to fuse Artificial Super Intelligence and robotics."

That acquisition now appears to have been the first piece of a much larger puzzle. Roze will consolidate several major SoftBank assets under a single corporate umbrella: ABB Robotics, the \$5.4 billion hardware acquisition that provides industrial robot arms, controllers, and software; Ampere Computing, the U.S. chip designer SoftBank acquired for \$6.5 billion; and DigitalBridge Group, the data center investment platform SoftBank is acquiring for approximately \$4.0 billion in enterprise value. DigitalBridge's portfolio holdings, including AtlasEdge and Vantage Data Centers, would provide Roze with an existing physical footprint and operational expertise in facility management.

Stargate and the Structural Advantage

Roze does not exist in isolation. SoftBank has poured roughly \$30 billion into OpenAI and is a cornerstone partner in the Stargate data center project—the \$500 billion initiative Son announced alongside Sam Altman and Larry Ellison at the White House in January 2025. That consortium, which also includes Oracle and MGX, plans to build up to 10 gigawatts of AI compute capacity across the United States. SoftBank's position inside Stargate gives Roze something no competing robotics or construction-tech startup can replicate: first refusal on tens of gigawatts of build capacity that are already capitalized and waiting on labor.

In total, SoftBank's AI commitments—including the December 2025 OpenAI investment, the April 2026 follow-on tranche, and the Roze consolidation—represent more than \$50 billion of fresh capital deployed in under five months. The sheer velocity underscores Son's conviction that the AI infrastructure buildout has at least a decade left to run, and that whoever owns the construction layer captures recurring economics that the chip and model layers cannot.

The Skeptics' Case

Not everyone is convinced. Some of Son's own executives have privately questioned whether both the timeline—targeting the second half of 2026 for the IPO—and the \$100 billion valuation are realistic, particularly given choppy equity markets and geopolitical headwinds. Several people familiar with internal discussions told reporters that 2027 may be a more achievable launch window.

"The lesson of Katerra and WeWork is that capital cannot fix construction by itself," one infrastructure investor told reporters covering the announcement. "Roze will need to ship robots that work in unstructured job sites, in heat, in dust, around live electrical work. That is a much harder problem than warehouse logistics."

The Katerra comparison stings. SoftBank's Vision Fund poured hundreds of millions into the modular construction startup, which promised to revolutionize building with technology and factory-produced components before collapsing into bankruptcy in 2021. WeWork, another Vision Fund darling that sought a sky-high IPO valuation before imploding, looms even larger in investor memory. Both failures illustrate the risk of applying tech-sector valuation multiples to industries with deep physical-world complexity.

Roze, as of late April, has no revenue, no completed data center, and no publicly demonstrated construction robot. A \$100 billion valuation on those fundamentals would make it one of the most expensive blank-check bets the public markets have ever been asked to underwrite.

Why It Matters

Despite the skepticism, the strategic logic is hard to dismiss outright. The AI industry's infrastructure needs are real, quantifiable, and growing. Hyperscalers including Microsoft, Google, Amazon, and Meta have collectively committed hundreds of billions to data center expansion over the next several years. Labor constraints are not theoretical—they are already delaying projects. If Roze can deliver robots that meaningfully accelerate construction timelines, the addressable market is enormous.

Son's deeper thesis is that vertically integrating chips (Ampere), robots (ABB), data center operations (DigitalBridge), and AI models (via the OpenAI partnership) creates a flywheel that no point-solution competitor can match. Whether that flywheel spins or stalls will depend on execution in the most literal sense: can the robots actually build?

What to Watch Next

The IPO timeline remains fluid. If SoftBank files in the second half of 2026, it would signal extraordinary confidence in both the technology and the market environment. A slip to 2027 would not necessarily be a negative—it could reflect a pragmatic decision to demonstrate working prototypes before asking public investors for \$100 billion in confidence. Either way, Roze is now the highest-stakes test of whether Masayoshi Son's AI spending spree can convert ambition into infrastructure, and infrastructure into returns.