Sources: SoftBank plans to create AI and robotics company Roze in the US for building data centers, and may list it as early as 2026 at an up to $100B valuation
Masayoshi Son plots IPO for business named Roze as soon as this year — SoftBank is planning to create and list an AI …
Context & Ripple Effects
SoftBank’s earlier coverage centers on turning AI assets and ambitions into financeable vehicles: it pursued Arm’s U.S. listing, Arm moved toward an AI-chip division, and SoftBank recently sought financing backed by its OpenAI holdings. Roze extends that arc from chip and model-adjacent exposure into planned U.S. data-center construction.
The proposed company would give Masayoshi Son’s AI strategy a dedicated operating and potential public-market vehicle rather than leaving the effort solely within SoftBank’s existing holdings.
First-order effects
- SoftBank would create Roze as a U.S. AI-and-robotics company focused on building data centers, placing that infrastructure effort in a separate entity.
- A potential listing would create a standalone valuation and financing route for Roze, while making investors assess its data-center plan independently of SoftBank’s broader portfolio.
Second-order effects
- Separating the project could make it easier for SoftBank to match infrastructure spending with dedicated equity financing, rather than relying only on parent-level resources or asset-backed borrowing.
- If Roze proceeds with construction, it would concentrate SoftBank’s demand for the equipment, compute, and other inputs required to operate data centers, linking its AI investment strategy more directly to physical infrastructure execution.
Third-order effects
- The move points to AI infrastructure increasingly being organized as a distinct, financeable asset class: investors may be asked to fund dedicated compute-buildout vehicles rather than only software, chip, or conglomerate exposure.
- Whether this model persists will depend on whether public markets sustain high valuations for infrastructure plans before the underlying facilities are operating at scale.
The trend: AI investment is broadening from ownership of models and chips toward separately financed, publicly marketable vehicles for the capital-intensive infrastructure behind them.