Top 5 Multi-Billion Dollar Deals OpenAI Signed so Far

OpenAI has entered a remarkable phase of deal-making, locking in multiple mega-contracts worth tens to hundreds of billions of dollars. These agreements reflect a fundamental shift in how AI development is financed and scaled: the frontier is no longer just algorithms and models, but compute infrastructure, data-centre scale, and long-term supplier partnerships.

OpenAI is essentially scouring the world for guaranteed access to GPUs, specialised hardware, cloud capacity and long-term financial backing. The significance is threefold: it anchors OpenAI’s ability to train and deploy ever-larger models; it signals to the market that compute scarcity is the bottleneck; and it ties

OpenAI into a web of strategic industrial partnerships that will reshape the AI supply chain. In short: OpenAI is no longer just a model-maker—it is building a compute empire.

Here are five of OpenAI’s biggest multi-billion-dollar deals to date, in (roughly) descending order of size, with date, value, and what each side gets.

1. OpenAI & Oracle Corporation

• When signed: Reported around September 2025.
• How much: About US$300 billion over ~five years.
• What OpenAI gets: Access to a massive cloud-infrastructure commitment from Oracle: computing capacity spanning roughly 4.5 gigawatts of power capacity (data-centre scale), capabilities to support model training and inference at scale beginning in roughly 2027.
• What Oracle gets: A guaranteed large-scale customer and revenue stream (tens of billions per year), strengthened positioning as a major cloud-AI supplier, and an anchor partner relationship with one of the most important AI companies. The deal also boosted Oracle’s revenue outlook dramatically.
• Why it matters: This deal is a statement that OpenAI is securing long-term compute capacity outside of its original partner network (e.g., Microsoft). A $300 billion commitment puts pressure on both parties to deliver next-generation infrastructure and cements Oracle’s role in the AI era.

2. OpenAI & NVIDIA Corporation

• When signed: Announced 22–23 September 2025.
• How much: Up to US$100 billion investment/partnership.
• What OpenAI gets: A commitment to deploy at least 10 gigawatts of NVIDIA systems in its data-centres, access to cutting-edge GPUs, and a strategic hardware partner aligned with its roadmap.
• What NVIDIA gets: A deep tie-in to OpenAI’s compute needs and model-build plans; access to a large, guaranteed buyer of its chips; and the opportunity to capture the AI-compute market at the infrastructure level.
• Why it matters: By investing so heavily, NVIDIA is entwined with OpenAI’s growth and model-deployment path. For OpenAI, securing GPU supply and hardware co-development is critical given the compute bottleneck in AI.

3. OpenAI & Advanced Micro Devices, Inc. (AMD)

• When signed: Announced early October 2025.
• How much: Described as a “multi-billion-dollar” deal (exact number not always disclosed, but the magnitude is very large and in the tens of billions) for supply of GPUs and other hardware.
• What OpenAI gets: Access to hundreds of thousands of AMD’s next-generation GPUs (Instinct MI450 series), deployment beginning in second half of 2026 with up to 6 gigawatts of capacity; also option to acquire ~10% stake in AMD via warrants.
• What AMD gets: A major anchor customer, very large hardware purchase commitment, potential rise in revenue (AMD said it expects over $100 billion in new revenue over four years from OpenAI and related customers).
• Why it matters: This deal shows OpenAI diversifying hardware suppliers beyond just NVIDIA. It also aligns OpenAI with hardware makers at a deeper level (equity, warrants). It underlines the scale of compute infrastructure needed.

4. OpenAI & CoreWeave, Inc.

• When signed: Several tranches—initial contract March 2025 (~US$11.9 billion), add-on May (~US$4 billion), expansion September ~US$6.5 billion.
• How much: Up to US$22.4 billion total across the partnership.
• What OpenAI gets: Cloud-infrastructure capacity from CoreWeave, one of the fastest-growing AI-cloud providers; helps OpenAI scale its datacentre buildout, particularly under its “Stargate” initiative.
• What CoreWeave gets: Long-term contract revenue, a strong anchor customer, and growth impetus in the AI-infrastructure space.
• Why it matters: Because the compute demands of OpenAI are so large, it is signing multiple parallel infrastructure partners rather than relying solely on one. The CoreWeave deal shows the breadth of its compute-architecture strategy, not just chip deals.

5. OpenAI & Amazon Web Services (AWS)

• When signed: Announced 3 November 2025.
• How much: US$38 billion, for a seven-year multi-year cloud partnership.
• What OpenAI gets: Access to hundreds of thousands of NVIDIA GB200/GB300 GPUs via AWS EC2 UltraServers, tens of millions of CPUs, and the ability to scale compute through 2026 and beyond.
• What AWS gets: A massive customer in OpenAI, prestige of hosting the compute for one of the world’s important AI companies, and capability to further entrench itself in the AI cloud market.
• Why it matters: This deal marks a strategic shift: until now OpenAI had strong association with Microsoft’s Azure cloud. Moving into AWS signals diversification and the scale of resources required to power its future efforts.