SoftBank Doubles Down on AI Infrastructure Amid Strategic Hurdles

SoftBank reports $20.7B profit, fueling a $180B AI fund. While announcing 5G and chip advances, its 'Stargate' data center project faces delays and soaring costs.

SoftBank Group is intensifying its strategic pivot toward artificial intelligence infrastructure, underscored by two recent technical announcements. However, the ambitious vision is encountering significant implementation challenges, revealing the complexity of translating strategy into concrete reality.

Financial Performance and Strategic Capital Allocation

The conglomerate's latest financial results provide substantial fuel for its expansive plans. For the nine-month period ending December 2025, SoftBank reported a net profit of $20.7 billion—a figure approximately five times larger than the prior year's result. This financial strength supports a monumental $180 billion fund earmarked for AI and technology investments.

The broader strategy, championed by CEO Masayoshi Son, operates on the premise that "Artificial Super Intelligence" will significantly surpass human intelligence within a decade. This belief is backed by substantial planned investments, including an estimated $40 billion into OpenAI and $2 billion directed toward Intel.

Technical Milestones: 5G and Next-Gen Chips

On the technical front, SoftBank, in collaboration with partners Murata and CLPA, announced a breakthrough in industrial 5G connectivity. The consortium successfully demonstrated Time-Sensitive Networking (TSN) over a 5G network for the first time. Laboratory tests achieved an average time synchronization of 122 nanoseconds, well under the 3GPP standard's limit of 900 nanoseconds.

Furthermore, the partners validated the operation of industrial equipment compliant with the "CC-Link IE TSN Class B" standard over the wireless network for a continuous period exceeding six hours. The full results are scheduled for presentation at MWC Barcelona in 2026.

Concurrently, SoftBank is advancing its AI data center capabilities. Following a recent $350 million funding round for SambaNova Systems and the launch of its SN50 AI chip, SoftBank is positioned to be the first major customer to deploy these processors. The SN50 hardware is designed to run agentic AI models at a significantly lower cost compared to traditional GPU approaches. These chips are slated for integration into the next generation of AI data centers in Japan.

Vision Meets Reality: Delays and Rising Costs

Despite clear progress, the path forward is not without obstacles. The highly publicized "Stargate" project—an offshore AI data center initiative involving OpenAI, Oracle, and MGX—is reportedly facing delays. Disagreements concerning ownership, financing, and construction have slowed development. Consequently, the objective of achieving 10 gigawatts of capacity by the end of 2025 has been missed.

Compounding the issue is severe cost pressure. Estimates for the required computing expenditure have surged from $450 billion to $665 billion. In the interim, OpenAI is increasingly relying on alternative cloud providers.

Market Snapshot

Shares of SoftBank's domestic telecommunications unit closed at 209.00 Japanese Yen in the latest trading session.