The AI race has hit a wall, and it’s not what anyone expected. It’s not smarter algorithms or bigger datasets. It’s the unglamorous stuff underneath, like power grids, data centers, and fiber networks.
SoftBank just made a roughly $4 billion bet, including debt, that infrastructure, not innovation, will determine who wins. On Monday, the Japanese conglomerate announced it is acquiring DigitalBridge Group, one of the world’s largest digital infrastructure platforms, for $16 per share in cash.
The deal capped weeks of speculation and sent DigitalBridge shares surging more than 50% in premarket trading, a clear signal of how hungry investors are for exposure to AI-critical infrastructure. The transaction is expected to close in the second half of 2026, pending regulatory and shareholder approvals.
Kelechi Edeh
DigitalBridge manages roughly $108 billion in digital infrastructure assets spanning data centers, cell towers, fiber networks, and edge facilities across North America, Europe, the Middle East, and Asia. For Masayoshi Son, SoftBank’s CEO, this move isn’t about expanding a portfolio. It’s about control. While much of the tech industry is focused on building better AI models, Son is buying the physical backbone those models depend on to run at scale, and the numbers explain why.
Global data-center capacity will need to grow by about 130% by 2030 just to keep up with projected demand. That implies building the equivalent of roughly 1,000 hyperscale facilities within five years. But power is the bigger constraint. Data-center electricity demand is expected to reach around 68 gigawatts by 2027, close to the entire generation capacity of California. In key markets like Northern Virginia, one of the world’s most important data-center hubs, new projects face grid connection wait times of five to seven years.
How does DigitalBridge fit into SoftBank’s Stargate AI plans?
The DigitalBridge acquisition fits directly into Son’s broader vision through the Stargate Project. Unveiled earlier this year alongside OpenAI, Oracle, and Abu Dhabi-based MGX, Stargate is a $500 billion effort to build large-scale AI infrastructure across the United States. The partners have already announced five data-center sites, with plans to scale capacity from roughly 7 gigawatts toward 10 gigawatts over the next several years.
That level of ambition requires more than capital. It requires land, power agreements, cooling systems, fiber connectivity, and the regulatory relationships to move quickly. DigitalBridge already controls many of those pieces. By acquiring the platform outright, SoftBank avoids the delays and coordination costs that come with building from scratch. It's effectively buying speed in a market where time and access to power are the scarcest resources.
The timing is also brutal for anyone on the outside. Amazon has committed about $86 billion to AI data centers in 2025 alone. Meta plans to spend between $60 billion and $65 billion on AI-optimized facilities. Google and Microsoft are racing to lock down capacity wherever they can find it. Most of these companies are building or leasing infrastructure. SoftBank, by contrast, just bought the operator that owns and manages it.
What happens to AI infrastructure if the AI bubble bursts?
The scale of AI infrastructure spending naturally raises questions about sustainability. Tech companies are projected to have spent around $400 billion on AI infrastructure in 2025, a level of capital expenditure with few historical parallels. To justify that investment, some analysts estimate AI would need to generate roughly $2 trillion in annual revenue by 2030. That figure exceeds the combined 2024 revenues of Amazon, Apple, Google, Microsoft, Meta, and Nvidia.
But infrastructure behaves differently from software during downturns. Data centers, fiber networks, and power connections don't vanish when valuations collapse. They retain utility. Even if today’s AI leaders stumble or entire business models fail, the physical assets being built will still be used by someone. That asymmetry is central to SoftBank’s bet. Rather than trying to predict which AI model will dominate, it's betting that whoever survives will need somewhere to run.
“As AI transforms industries worldwide, we need more compute, connectivity, power, and scalable infrastructure,” Son said in a statement announcing the deal, adding that the acquisition would strengthen the foundation for next-generation AI data centers.
Kelechi Edeh
The move also fits a broader shift in SoftBank’s strategy. In November, the company sold its entire Nvidia stake for $5.83 billion to help fund a $30 billion commitment to OpenAI. That signaled a move away from betting on individual components toward betting on platforms. The DigitalBridge deal takes the logic one step further, from platforms to the physical infrastructure that those platforms cannot function without.
DigitalBridge will continue operating as a separate platform under CEO Marc Ganzi once the acquisition closes. For SoftBank, however, the calculation is already clear. Leadership in AI will not come solely from building smarter models. It'll come from controlling the infrastructure that those models need to scale. By acquiring DigitalBridge, SoftBank is no longer just financing the AI race. It's now positioning itself to own the ground everyone else has to build on.
