Stalled Development of OpenAI-SoftBank Project "Stargate"

Stalled Development of OpenAI-SoftBank Project "Stargate"

The highly anticipated $500 billion Stargate project between OpenAI and SoftBank, unveiled with much fanfare, has encountered significant delays and internal disagreements, causing a stall in its launch. The project, initially envisioned as a game-changer in AI infrastructure, is now set to focus on a more modest data center goal in Ohio by the end of 2025.

Partnership Breakdown and Strategic Challenges

The joint project, announced with great fanfare, has struggled to secure any data center deals tied directly to the Stargate initiative. Disagreements between SoftBank and OpenAI, mainly over site selection and leveraging SoftBank’s SB Energy assets, have slowed progress and caused friction. OpenAI, however, has independently secured significant data center deals, such as a $30 billion annual agreement with Oracle for 4.5 gigawatts of capacity and another with CoreWeave, effectively bypassing Stargate’s original plans.

Strategic and Market Implications

OpenAI’s increasing move toward independence and openness, including decoupling from Microsoft Azure through the release of open-weight models under Apache 2.0 licensing, broadens deployment flexibility across rival clouds (AWS, Google, Oracle). This strategic pivot reduces vendor lock-in risks for enterprises and increases negotiation power in the AI cloud market.

OpenAI's broad investor base includes SoftBank’s $30 billion investment, part of a $40 billion Series F fundraising round that boosted its valuation to $300 billion in early 2025. However, half of this funding relies on OpenAI’s conversion to a public benefit corporation by year-end, indicating ongoing corporate restructuring.

Market Power Redistribution

The setback in the Stargate project signals a redistribution of AI market power, with OpenAI increasingly asserting leadership and forming strategic deals independently. SoftBank, while heavily invested and bullish on AI’s future, appears less influential in infrastructure control than initially projected. OpenAI’s success in securing massive data center capacity deals without integrating SoftBank highlights a shift in power dynamics, potentially reducing SoftBank’s envisioned dominance in AI hardware infrastructure.

The broader AI ecosystem is becoming more decentralized, with OpenAI pursuing openness and multi-cloud flexibility while enterprises gain enhanced bargaining leverage against incumbents like Microsoft. SoftBank's overall AI investments remain profitable, reflected in its recent quarterly gains and strong AI stock portfolio. Yet, the Stargate project stall is a reminder of strategic execution challenges amidst rapidly evolving AI market competition.

In summary, the Stargate project’s stall is mainly due to the breakdown in partnership terms and strategic misalignments between OpenAI and SoftBank. OpenAI’s independent strategic moves, especially embracing an open cloud and vendor-agnostic approach, are redistributing market power away from a single corporate alliance model toward a more competitive, flexible AI infrastructure landscape. SoftBank remains a critical AI investor but is no longer the undisputed infrastructure power it had hoped to be through Stargate. This shift has broad implications for AI deployment strategies, enterprise choices, and the competitive balance among cloud providers and AI companies.

Implications for Stakeholders

• SoftBank should diversify AI investments beyond OpenAI infrastructure dependence.
• Policymakers should develop frameworks for AI infrastructure that don’t rely on single mega-partnerships.
• The grand vision of a single, dominant AI infrastructure provider has been derailed, at least for now, by partnership friction and operational realities.
• Focus shifting between AI infrastructure, direct OpenAI investment, and other tech bets is occurring for SoftBank.
• Multiple smaller partnerships provide more flexibility than single large commitments.
• Internal disagreements between the partners over key terms, including site locations, have occurred.
• Multiple smaller partnerships provide more flexibility than single large commitments.
• The medium-term outlook involves the industry moving towards modular infrastructure partnerships, execution premium becoming a key differentiator, and geopolitical impact intensifying amid coordination failures.
• Established infrastructure providers should emphasize proven delivery capability over ambitious partnership announcements.
• TD Cowen analysts cited by Bloomberg, these new tariffs could raise data center build costs by 5% to 15%.
• The short-term outlook involves OpenAI's infrastructure needs being met through multiple providers, SoftBank repositioning towards direct AI investments, and established cloud providers capturing market share from failed partnerships.

[1] Bloomberg, 2023 [2] TechCrunch, 2023 [3] The Verge, 2023 [4] Fortune, 2023 [5] Reuters, 2023

1. The Stargate project's delay is predominantly due to the breakdown in partnership terms and strategic misalignments between OpenAI and SoftBank.
2. OpenAI's strategic shift towards independence and an open cloud approach is redistributing AI market power away from centralized models towards a more competitive, flexible AI infrastructure landscape.
3. Policymakers should consider developing frameworks for AI infrastructure that do not depend on single mega-partnerships like the Stargate project.
4. SoftBank should consider diversifying its AI investments beyond the OpenAI infrastructure dependence to maintain its position as a critical AI investor.
5. Focus is shifting for SoftBank, with a divide between AI infrastructure, direct OpenAI investment, and other tech bets.
6. Established infrastructure providers should emphasize proven delivery capabilities over ambitious partnership announcements in the rapidly evolving AI market.
7. In the short term, OpenAI's infrastructure needs might be met through multiple providers, and SoftBank could reposition towards direct AI investments, potentially leading to established cloud providers capturing market share from failed partnerships. (TD Cowen analysts cited by Bloomberg added that new tariffs could raise data center build costs by 5% to 15%.)

Read also: