In 2025, OpenAI struck deals worth about $1 trillion to build the infrastructure it needed to grow its artificial intelligence services. These agreements span data centers, chip supply, cloud services, and more. The move signals a bold push by OpenAI to scale rapidly, but also raises questions about funding and risks.
What the Deals Cover
- OpenAI has made large agreements with major tech companies, including AMD, Nvidia, Oracle, and cloud/data infrastructure providers.
- One recent deal with AMD involves deploying 6 gigawatts of compute power using AMD’s chips, with the first 1 gigawatt coming online by late 2026.
- Another major pact is with Nvidia, in which Nvidia will invest up to $100 billion and supply chips to OpenAI’s data centers.
- OpenAI also made a $30 billion per year cloud & infrastructure deal with Oracle to lease large amounts of compute and build new data centers. Financial Times
- Further contracts include expansions with cloud providers like CoreWeave and other infrastructure firms.
Why These Deals Matter
- AI systems like ChatGPT and newer models require massive computing power. To train, run, and scale these models, the hardware and data centers must be vast and reliable.
- By locking in infrastructure deals now, OpenAI ensures it has resources and partners ready as demand grows.
- These deals also tie together many big tech firms into a shared ecosystem — many are now both suppliers and stakeholders in OpenAI’s growth.
Challenges & Risks
- Financing gap: The $1 trillion figure is much larger than OpenAI’s current revenue. Analysts question whether OpenAI can fulfil all commitments.
- Execution risks: Building data centers, securing power, cooling, logistics — many moving parts must work smoothly.
- Energy demand: These AI systems consume huge amounts of electricity. Ensuring sustainable, affordable power is a challenge.
- Competition: Other tech giants like Google, Microsoft, and AI startups are also making big infrastructure bets.
What’s Next
OpenAI has already announced the Stargate project, a joint effort with Oracle, SoftBank, and other partners to invest up to $500 billion in AI infrastructure through 2029.
Going forward, we can expect more infrastructure announcements, further chip deals, and perhaps new partners joining OpenAI’s expansion mission.
Summary Table
| Topic | Key Fact / Detail |
|---|---|
| Total deal value in 2025 | ~$1 trillion in infrastructure / compute deals |
| Major partners | AMD, Nvidia, Oracle, CoreWeave, cloud & infrastructure firms. |
| AMD deal specifics | Deploy 6 GW of chips, first 1 GW by late 2026; OpenAI gets option for ~10% stake in AMD |
| Nvidia deal specifics | Up to $100 billion investment and chip supply agreement |
| Oracle deal specifics | $30 billion/year for compute leasing, new data centers for “Stargate” project Financial Times |
| Major challenges | Funding feasibility, execution, energy needs, competitive pressure |
