SoftBank’s Son says AI revolution dwarfs dotcom boom By Investing.com

SoftBank will invest €75 billion (about $87 billion) to build AI infrastructure in France, including 3.1 gigawatts of data center capacity in Hauts-de-France by 2031 as part of a 5-gigawatt buildout. CEO Masayoshi Son said the AI boom could be 50x larger than the dotcom era and framed any market correction as a buying opportunity. The announcement, made alongside President Macron and Schneider Electric, reinforces Europe’s AI infrastructure race and could support related industrial and data-center names.

Analysis

This reads as a capital-allocation signal more than a product headline: the market is being told to treat AI infrastructure as a multi-cycle industrial buildout, not a speculative software theme. The second-order implication is that the beneficiaries are increasingly the picks-and-shovels names with power, grid, cooling, and industrial automation exposure, while pure-play AI software remains vulnerable to valuation compression if capex growth outpaces near-term monetization.

The France siting matters because it shifts the bottleneck from chips to electrons and permits. Over the next 12-24 months, the trade is less about model breakthroughs and more about who can secure low-cost power, interconnects, and local subsidies; that favors European utilities, grid equipment, and high-voltage infrastructure suppliers, while penalizing incumbents in regions where permitting is slower or power prices are higher. A multi-gigawatt commitment also increases concentration risk in the AI supply chain: any delay in transformer lead times, grid upgrades, or semiconductor supply could push revenue recognition right by quarters, not weeks.

The market is likely underpricing the financing risk. Large AI infrastructure projects can look reflexively bullish until rising rates, weaker enterprise spend, or a sentiment reset forces investors to distinguish between announced capex and economic returns; that creates a 6-18 month window where the stocks of enablers can outperform while the mega-cap AI complex can still derate. The contrarian view is that the current enthusiasm is not about near-term earnings — it is a public attempt to anchor long-duration expectations higher, which often precedes a rotation into infrastructure beneficiaries and away from crowded momentum names.

Geopolitics adds a separate layer: if tensions escalate, any energy-price spike would hurt data-center operating costs and could make power-intensive AI builds less attractive in Europe. So the cleanest expression is not a broad long AI basket, but a barbell: long infrastructure enablers with order-book visibility, short the most expensive AI software names that need flawless execution to justify multiples. The key catalyst to monitor is whether this project converts from announcement to grid connection and equipment orders within the next two quarters; if it slips, the whole thesis becomes a headline-only trade.

Key Decisions for Investors

• Long SU.PA / Schneider Electric into 6-12 months; pair against a basket of high-multiple AI software names if available. Risk/reward: infrastructure demand is more tangible than model monetization, but any delay in project approvals or grid interconnects is the main downside.
• Long European grid and electrical equipment exposure via NG.L / ABB / Siemens Energy over the next 3-9 months. Thesis: multi-gigawatt AI buildouts create a durable backlog in transformers, switchgear, and cooling systems before they create earnings for software vendors.
• Short the most crowded, premium-valued AI software proxy basket on any post-news strength; use 3-6 month horizon. Thesis: capex announcements support the theme, but not necessarily near-term revenue conversion, making valuation compression likely if execution slips.
• Buy out-of-the-money calls on select utility/infrastructure suppliers with direct AI datacenter exposure for 6-12 months. Preferred structure: limited premium paid to capture a rerating if the project triggers follow-on European AI capex.
• Avoid chasing broad AI beta immediately; wait for confirmation of power and permitting milestones over the next 1-2 quarters before adding risk. The best entry is on any headlines about delays, which would likely reset expectations and improve entry points.