SoftBank Group is racing to assemble $22.5 billion owed to OpenAI before December 31, deploying aggressive asset sales and potential borrowing to complete one of the largest AI infrastructure investments in history. The scramble underscores how capital access now defines competitive advantage in artificial intelligence.
The obligation stems from SoftBank's April agreement to invest up to $40 billion in OpenAI at a $300 billion valuation, contingent on the AI company completing its transition to a for-profit structure. That restructuring concluded in October, triggering payment terms requiring the remaining $22.5 billion by year-end after an initial tranche was delivered.
SoftBank CEO Masayoshi Son has already sold the firm's entire $5.8 billion stake in AI chip leader Nvidia, offloaded $4.8 billion of T-Mobile US shares, and implemented staff reductions to marshal funds. Son has also slowed most dealmaking at SoftBank's Vision Fund, requiring his explicit approval for any transaction exceeding $50 million.
Strategic Stakes
Completing the deal would deepen SoftBank's influence across the AI supply chain, including participation in the Stargate Project—a proposed $500 billion initiative with OpenAI, Oracle, and other partners to build large-scale AI data centers across the United States.
SoftBank secured the OpenAI investment when the company was valued at $300 billion. Since then, OpenAI's valuation has climbed dramatically, with the company now in talks to raise additional funding from investors including Amazon that could triple its valuation to approximately $900 billion. This trajectory would deliver SoftBank significant paper gains once the transaction completes.
SoftBank may tap margin loans backed by its ownership of semiconductor firm Arm Holdings to bridge any remaining gap. The company recently expanded its margin loan capacity by $6.5 billion, bringing total undrawn capacity to $11.5 billion. Arm's stock has tripled since its IPO, providing additional collateral headroom.
Another capital pool sits on a longer timeline. SoftBank planned to take PayPay, its payments app operator, public this month. That IPO slipped after regulatory delays and is now targeted for first quarter 2026, expected to raise more than $20 billion.
Industry Implications
Rising training and deployment costs are reshaping AI economics. The sector now resembles capital-intensive industries like energy and telecommunications, characterized by long investment cycles and persistent spending. Capital depth affects bargaining power with chip suppliers, data center operators, and cloud partners.
Tech giants have moved similarly. Meta and Microsoft are committing unprecedented sums to data center buildouts, often sharing risk through partnerships. The rush to build infrastructure has prompted companies to bring in deep-pocketed partners to spread financial exposure.
For SoftBank, meeting the deadline would reinforce its role as a central force in the AI ecosystem. Private market discussions already point to OpenAI valuations well above recent rounds, offering substantial paper gains if the deal closes on schedule. Missing the date would signal execution challenges during a critical period for AI infrastructure investment.
The scramble captures a fundamental shift across AI. Technical breakthroughs still matter, but staying power now matters equally. In this phase, companies that continue funding infrastructure set the competitive pace.
