SoftBank FY 2025: $64B OpenAI Gamble

SoftBank Group reported strong FY25 results, with revenue up 8% to ¥7,038.7 billion and free cash flow of ¥633.6 billion. The company is pivoting toward AI infrastructure, targeting ¥9 trillion in revenue and ¥1.7 trillion in operating income by FY30, supported by Enterprise Cloud & AI expansion. Its 13% OpenAI stake adds strategic leverage and potential valuation upside in the global AI ecosystem.

Analysis

The market is likely underestimating how much of SoftBank’s next leg is a financing and platform-control story rather than a simple earnings-upgrade story. If management can keep converting operating cash flow into asset-backed AI infrastructure, the company becomes a quasi-picks-and-shovels lever on compute scarcity, with returns driven as much by contract duration and capacity allocation as by headline revenue growth. That creates a potential re-rating path more similar to infrastructure owners than to a traditional holding company.

The biggest second-order winner is likely the AI supply chain outside SoftBank’s own equity stake: GPU/HBM suppliers, datacenter power equipment, optical networking, and grid-adjacent beneficiaries. A credible FY30 AI buildout implies multi-year demand visibility, which can support pricing power and tighter utilization across the ecosystem even if end-market AI software monetization remains uneven. Conversely, large cloud incumbents and private AI financiers may face margin pressure if SoftBank uses its capital base to subsidize early capacity capture and lock up strategic partnerships.

The key risk is that the narrative is front-loaded while execution is back-ended. The market can grant premium multiples for AI optionality for 6–18 months, but capital intensity, dilution, and valuation sensitivity of the OpenAI stake become much more important over a 2–4 year horizon if monetization lags infrastructure spend. The contrarian view is that consensus may be overvaluing the OpenAI stake as a clean mark-to-market asset, when in practice its value depends on governance, model leadership retention, and the availability of downstream compute — all of which could compress if AI capex saturates or financing conditions tighten.

Near term, the likely catalyst stack is not the annual number itself but any evidence of signed capacity, financing structure, or strategic partnership that reduces perceived execution risk. If SoftBank can show repeatable deployment economics, the stock can continue to rerate; if not, the market may eventually separate cash generation from speculative AI asset inflation and cap upside. The most attractive setup is to own the infrastructure enablers while fading the most expensive AI beneficiaries that lack control over the physical bottlenecks.

Key Decisions for Investors

• Long SFTBY on 6–12 month horizon into AI capex visibility: target a rerating if management proves financing discipline and signed capacity; trim if premium to look-through NAV expands beyond 25-30% without execution milestones.
• Pair trade: long semiconductor/power infrastructure beneficiaries vs short high-multiple AI software with weak monetization visibility over 3–9 months; the better trade is where revenue is tied to actual buildout, not narrative.
• Add exposure to datacenter/power chain leaders on any post-earnings consolidation; use a 3–6 month window because capacity commitments should filter through order books before revenue shows up.
• Avoid chasing the OpenAI-linked upside alone; if SoftBank’s AI thesis is being priced like a simple option, consider reducing exposure after any sharp post-results move unless there is a fresh catalyst on funding or deployment terms.
• For more tactical positioning, buy medium-dated calls on SFTBY only on pullbacks, not breakouts; this preserves upside to AI re-rating while limiting premium paid for already-positive sentiment.