SB Energy IPO filing reveals $430 billion AI infrastructure backlog

The scale of this filing underscores the extreme capital requirements for the next phase of AI infrastructure, which directly underpins the generative video and streaming recommendation engines of the future. By vertically integrating power generation with data center development, SoftBank is attempting to solve the energy bottleneck that currently limits large-scale compute expansion. This move creates a tight financial loop between Nvidia, OpenAI, and SoftBank, potentially crowding out smaller infrastructure players who lack similar backstops. Watch for the final IPO pricing and the specific terms of the residual value guarantee, as these will set the benchmark for future AI-focused infrastructure valuations.

SoftBank Group has been aggressively consolidating its AI infrastructure portfolio ahead of the SB Energy IPO filing. In June 2026, Nokia and Indosat Ooredoo Hutchison announced a GPU-accelerated AI-RAN partnership in Indonesia, expanding the Nokia-NVIDIA architecture already adopted by T-Mobile US, SoftBank, and Vodafone, signaling that SoftBank's telecom operations are already embedded in the same GPU-heavy infrastructure thesis that underpins SB Energy's data center ambitions. The broader pattern shows SoftBank positioning itself as a full-stack AI infrastructure provider, from radio access networks to hyperscale compute facilities, with Nvidia serving as the connective tissue across both domains.

The financial structure of SB Energy's IPO reflects a new model for AI infrastructure capital formation that diverges sharply from traditional utility or data center REIT approaches. Nokia's Autonomous Network Fabric is delivering automation rates higher than 90 percent and service delivery times of four hours or less for operators, demonstrating the operational efficiency gains that justify massive infrastructure investment at the network layer. For SB Energy, the $105 billion Nvidia residual value guarantee functions as a similar de-risking mechanism, but at the power and physical infrastructure layer rather than the software orchestration layer. This creates a template where chip vendors underwrite the demand side of infrastructure projects, effectively converting hardware purchase commitments into project finance collateral.

Ericsson's competing vision for AI-era infrastructure provides useful contrast to SoftBank's approach. Ericsson described the network as becoming an intelligent fabric connecting devices, agents, sensors, cars, and edge nodes, with CTO Ekudden highlighting that uplink traffic could triple over the next five years driven by AI glasses, persistent voice interaction, and real-time video. In roughly a third of operator networks today, uplink growth is already outpacing downlink growth by 50 percent. These traffic projections validate the demand assumptions underlying SB Energy's 10-gigawatt PORTS-Pike campus and similar projects, though Ericsson's distributed fabric model implies different infrastructure requirements than the centralized hyperscale approach SoftBank is pursuing with OpenAI.