SoftBank's $500 Billion AI Dream Falters: Inside the Struggles of the Stargate Project

A Grand Vision at Risk of Collapse

In January 2025, SoftBank Group’s charismatic CEO Masayoshi Son stood at the White House and unveiled an audacious vision: a \$500 billion investment into artificial intelligence (AI) infrastructure over the next four years, in collaboration with OpenAI. The plan was to build massive AI-centric data centers across the United States, powered by dedicated energy facilities—what Son boldly claimed would be a transformative force for the future of humanity. But now, only months later, that towering vision is showing signs of faltering.

According to a July 21st report by The Wall Street Journal, the “Stargate” initiative—a joint mega-project between SoftBank and OpenAI—has run into serious roadblocks. Originally slated to inject \$100 billion immediately into development, the project has now been drastically scaled down. Instead of launching full-scale infrastructure, the goal has shifted to constructing a modest data center in Ohio by the end of 2025.

At the heart of the stall are disagreements between SoftBank and OpenAI, particularly regarding the location of key data center sites. While internal discussions continue, progress has slowed significantly. Despite this, both companies issued a rebuttal to WSJ’s claims, asserting that Stargate is “no longer just a vision, but rapidly becoming reality.” They cited ongoing evaluations across multiple U.S. states and emphasized that some projects are already underway.

SoftBank’s CFO Yoshimitsu Goto offered a more optimistic picture back in May, stating that over 100 candidate locations were being considered and that the first three sites were likely to be in Texas. He claimed the initiative was “very close to realization.” Yet, WSJ’s report hints at a more sobering reality: momentum is waning, logistics are clashing, and despite bold claims, progress is lagging behind schedule.

Meanwhile, SoftBank has deepened its financial entanglement with OpenAI. In April, it announced an additional \$30 billion in funding, bringing its total investment to a staggering \$32.2 billion—overtaking Microsoft as OpenAI’s top financial backer. However, this growing stake could become a liability. Microsoft is reportedly reevaluating its own partnership with OpenAI, leaving SoftBank to shoulder an even greater share of the risk.

What Undercode Say:

The Stargate project was initially positioned as a moonshot—ambitious, game-changing, and futuristic. But like many moonshots, it is now crashing into the hard wall of reality. The challenges SoftBank and OpenAI face are not just about location scouting or construction timelines; they reflect deeper tensions in the AI race itself.

SoftBank is walking a tightrope. On one hand, it aims to emerge as a global leader in AI infrastructure, making bets bigger than any of its contemporaries. On the other hand, its increasing financial commitment to OpenAI places it in a vulnerable position. If OpenAI’s vision stumbles or pivots sharply, SoftBank may find itself overextended.

The slowdown in Stargate’s rollout underscores the complexity of deploying large-scale AI infrastructure. This isn’t just about stacking servers—it’s about securing power supplies, obtaining permits, coordinating across state jurisdictions, and building the physical and digital spine for next-gen computation. It’s a logistical nightmare, even with billions on the table.

OpenAI, while clearly a technological trailblazer, appears increasingly selective about its partnerships and long-term vision. Their disagreements with SoftBank over something as fundamental as site selection point to deeper strategic misalignment. That’s concerning, especially when billions have already been exchanged.

SoftBank’s eagerness to overtake Microsoft in its investment volume could backfire. Microsoft has integrated OpenAI into its product suite (from Azure to Copilot), which helps hedge its bets. SoftBank, however, has no such diversified ecosystem—making it more exposed if OpenAI’s trajectory falters.

Yet, the project is not doomed. The reduction in scope may be strategic—a recalibration rather than a collapse. Building a smaller data center in Ohio first could serve as a prototype or proof-of-concept. Once operational, it might help fast-track regulatory approvals and instill investor confidence. But the clock is ticking. AI infrastructure demands urgency, and any delay is a missed opportunity.

In essence, the Stargate story is a microcosm of the broader AI arms race—where vision, capital, and execution must align perfectly. As of now, SoftBank is learning the hard way that even the deepest pockets can’t buy perfect coordination.

🔍 Fact Checker Results

✅ WSJ did publish a report on July 21, 2025, outlining the stalled progress of the Stargate project.
✅ SoftBank has officially announced a \$500 billion AI investment plan in collaboration with OpenAI earlier this year.
❌ Claims that full-scale infrastructure would be built immediately are now retracted—the target is a smaller facility in Ohio.

📊 Prediction

If SoftBank fails to operationalize the first phase of Stargate by mid-2026, OpenAI may seek alternative infrastructure partners—potentially reviving deeper collaborations with Microsoft or even AWS. SoftBank’s heavy investments could become stranded assets, especially if AI compute demand shifts toward more modular, decentralized solutions. However, if the Ohio pilot succeeds and scales efficiently, SoftBank might reclaim momentum and redefine itself as a dominant AI infrastructure provider across North America.