Recommended Free Tools
Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
Yes, the OpenAI–Oracle deal is real—but the headline needs precision. The companies first announced an agreement on July 22, 2025, to develop up to 4.5 gigawatts of additional U.S. data-center capacity for OpenAI’s Stargate project. OpenAI later described the partnership as worth more than $300 billion over five years.
That does not mean OpenAI paid Oracle $300 billion upfront, that Oracle alone is funding the entire project, or that all 4.5 GW is already operating. The agreement sits inside Stargate’s broader, separate plan to invest up to $500 billion over four years in U.S. AI infrastructure.
The short answer
OpenAI and Oracle agreed to expand Stargate through up to 4.5 GW of additional U.S. data-center capacity. OpenAI subsequently characterized the two-company partnership as exceeding $300 billion over five years.
The most accurate description is therefore a multi-year commercial and infrastructure partnership, not a confirmed $300 billion cash payment or a single Oracle-owned data center. The public announcements do not disclose the complete contract, payment schedule, minimum purchase obligations, cancellation provisions, or final financing structure.
#1 Best Overall
For households and investors, the important distinction is between a headline contract value and money that has already been spent or recognized as revenue. Much of the infrastructure still requires power connections, construction, equipment, financing and deployment.
OpenAI’s July 2025 announcement said the additional capacity, combined with the initial Stargate site in Abilene, Texas, would put more than 5 GW under development and support more than 2 million chips. Those figures were company estimates, not an independent audit.
What each major number means
| Figure | What it refers to | What it does not prove |
|---|---|---|
| 4.5 GW | Additional U.S. data-center capacity covered by the OpenAI–Oracle agreement announced in July 2025. | It is not a direct measure of model performance, available computing power or operating capacity. |
| More than $300 billion | OpenAI’s later description of the value of its partnership with Oracle over five years. | It is not evidence of a $300 billion upfront payment or a fully disclosed guaranteed revenue contract. |
| $500 billion | The broader Stargate infrastructure investment target announced in January 2025 over four years. | It is not the value of the Oracle agreement and was not described as fully funded on day one. |
| Nearly 7 GW | OpenAI’s reported broader Stargate capacity plan in September 2025. | It does not mean nearly 7 GW was already operational. |
| More than $400 billion | OpenAI’s reported planned Stargate investment over three years as of September 2025. | It should not be combined mechanically with the $300 billion Oracle figure. |
A gigawatt measures electrical power capacity. It does not directly tell readers how many useful computations a facility can perform. Actual output depends on the chips installed, networking, cooling, software efficiency, utilization and the availability of electricity.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11What is Stargate?
Stargate is an OpenAI-centered AI infrastructure initiative involving data centers, power, chips, networking, cloud services, financing and construction partners. It is not simply a cloud subscription or one facility.
When Stargate was announced on January 21, 2025, OpenAI and SoftBank described a plan to invest $500 billion over four years in U.S. AI infrastructure, with $100 billion intended for immediate deployment. SoftBank and OpenAI were identified as lead partners. SoftBank was assigned financial responsibility, while OpenAI was assigned operational responsibility.
The initial announcement named Oracle, Nvidia, Arm, Microsoft and OpenAI as technology partners. The first buildout was associated with Texas, with additional sites under evaluation. The original Stargate announcement did not mean that all planned infrastructure had already been financed, constructed or powered.
A timeline of the deal
- January 21, 2025: OpenAI and SoftBank announce Stargate, including the four-year, $500 billion infrastructure target.
- July 22, 2025: OpenAI announces an agreement with Oracle for up to 4.5 GW of additional U.S. Stargate data-center capacity.
- July 2025: OpenAI says Oracle has begun delivering Nvidia GB200-family racks to the Abilene, Texas, site and that early training and inference workloads have started there.
- September 23, 2025: OpenAI describes the Oracle partnership as exceeding $300 billion over five years and announces additional Stargate sites.
- October 22, 2025: OpenAI identifies the Midwest site as being in Wisconsin and names Vantage as Oracle’s development partner there.
- March–June 2026: Oracle reports rapid cloud growth, large total remaining performance obligations and additional financing activity. Those corporate figures cannot be attributed entirely to OpenAI.
What Oracle is providing
Oracle’s role extends beyond renting ordinary virtual machines. The partnership involves several layers of the AI infrastructure stack:
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Fix the driver behind crashes, sound loss and screen glitches3Repair Windows errors before they cause bigger problems- Data-center development and operations.
- Oracle Cloud Infrastructure capacity.
- Nvidia GPU systems.
- High-speed networking and related systems infrastructure.
- Power, cooling, storage and facility operations.
- A cloud environment for training and running OpenAI models.
OpenAI said Oracle had begun delivering Nvidia GB200 racks to Abilene in June 2025 and that early training and inference workloads had begun. Those statements came from OpenAI and should be treated as company-reported progress rather than an independently audited performance report.
Building an AI data center involves much more than purchasing GPUs. A project also needs land, permits, electricity interconnection, substations, transmission capacity, cooling systems, fiber connectivity, construction labor, financing and operating staff. Any one of those elements can delay usable capacity.
Where the facilities are located
| Location | Reported role or partners | Development-stage qualification |
|---|---|---|
| Abilene, Texas | Flagship Stargate site associated with OpenAI and Oracle. | OpenAI reported that portions were operating and that early workloads had begun. |
| Shackelford County, Texas | One of the additional Stargate locations announced by OpenAI. | Announced or under development; not equivalent to fully operational capacity. |
| Doña Ana County, New Mexico | Additional Stargate site. | Planned or under development according to the public announcements. |
| Wisconsin | Midwest site involving Oracle and Vantage. | Announced development; public announcements do not establish that the full planned capacity was operating. |
| Lordstown, Ohio | Site developed through a SoftBank-related partnership. | Part of the broader Stargate program, not necessarily the Oracle-specific 4.5 GW alone. |
| Milam County, Texas | Another site associated with the broader Stargate expansion. | Planned or under development rather than automatically available compute. |
OpenAI did not initially disclose the full locations and funding details for the additional 4.5 GW. The public site announcements also cover capacity at different stages: some operating in part, some under construction and some planned or scalable.
Why OpenAI needs so much capacity
OpenAI needs infrastructure for two fundamentally different workloads.
Training
Training is the process of building and refining AI models. It can require large clusters of GPUs working together for extended periods. Training workloads demand high-speed networking, stable power and carefully coordinated storage and software.
Inference
Inference is the process of serving responses after a model has been trained. It is tied directly to user and application demand. A surge in usage can require additional capacity even when no new frontier model is being trained.
OpenAI therefore needs a large, reliable supply of computing capacity rather than access to a single facility. Geographic distribution can improve resilience and help place capacity near available power and network infrastructure, although operating across multiple locations also increases technical complexity.
Why Oracle wants the partnership
For Oracle, OpenAI offers the kind of large anchor customer that can justify major investment in cloud infrastructure. The agreement could help Oracle:
Free tools Windows power users keep installed
One-click scans. No signup required.
- Expand Oracle Cloud Infrastructure’s position in AI training and inference.
- Secure long-term demand for data centers, GPUs and networking.
- Compete more directly with Microsoft Azure, Amazon Web Services and Google Cloud.
- Increase cloud revenue and future contracted obligations.
- Accelerate construction and financing of AI-focused facilities.
Oracle’s fiscal 2026 results reported $67.4 billion in total revenue, $34 billion in cloud revenue and $18.1 billion in cloud infrastructure revenue, with infrastructure revenue up 77% year over year. Oracle also reported total remaining performance obligations of $638 billion at the end of the fourth quarter.
Those are company-wide figures. They do not mean that $638 billion of remaining obligations came from OpenAI or that Oracle’s entire cloud growth was generated by Stargate. Oracle’s official fiscal 2026 results should be read in that context.
Is Oracle replacing Microsoft?
No. The public announcements describe expansion and diversification, not a clean Oracle-for-Microsoft switch.
OpenAI said Microsoft would continue providing cloud services to OpenAI, including through Stargate. The original Stargate announcement also said OpenAI would continue increasing its Azure consumption while adding computing capacity from Stargate partners.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
The strategic result is a multi-cloud arrangement. OpenAI gains access to additional infrastructure and may reduce dependence on a single provider. Microsoft remains a major cloud partner, while Oracle becomes a significant additional source of capacity.
Multi-cloud infrastructure can improve bargaining power and resilience, but it creates its own costs. OpenAI must manage data movement, security, identity, networking, software portability, orchestration and workload placement across providers.
What does the $300 billion actually represent?
The available official wording establishes that OpenAI described the Oracle partnership as exceeding $300 billion over five years. It does not, by itself, establish the commercial mechanics behind that figure.
The figure could include multiple components such as cloud services, infrastructure capacity, equipment, operations and long-term commitments. The public record does not provide a complete breakdown.
It does not publicly establish:
- The exact payment schedule.
- Minimum purchase or take-or-pay obligations.
- Cancellation rights.
- The exact amount allocated to construction, GPUs, networking, power or cloud services.
- Whether the entire figure is guaranteed revenue.
- Whether all planned capacity will be delivered on the same timetable.
- Who owns every GPU and other major infrastructure asset.
For personal-finance readers, this distinction is familiar from other large business commitments: a headline contract value is not the same as cash already paid, revenue already recognized or profit already earned.
How the infrastructure may be financed
It would be misleading to say that OpenAI simply paid Oracle $300 billion upfront. The public record points to a more complicated mix of possible funding sources, including:
- Oracle capital spending.
- Customer commitments and potential prepayments.
- Debt and equity financing.
- Project-level financing.
- Data-center developers and infrastructure partners.
- GPU procurement and equipment financing.
- Customer-supplied equipment in some arrangements.
Oracle’s filings discussed customer prepayments or equipment supplied directly by customers in connection with large infrastructure commitments. The filing did not identify every customer or show that the entire amount related to OpenAI.
Oracle also said it planned to raise up to $50 billion in debt and equity financing during calendar 2026 and had raised $30 billion through investment-grade bonds and mandatory convertible preferred stock at the time of its March 2026 filing. That financing was described as supporting Oracle’s broader cloud and AI expansion, not as Stargate-only funding.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Oracle’s SEC filing is therefore useful evidence of the scale and complexity of the financing environment, but it is not proof that Oracle has already spent $300 billion on Stargate.
What about the reported $30 billion annual contract?
Reporting connected Oracle’s previously undisclosed customer contract—described as worth roughly $30 billion in annual revenue beginning in fiscal 2028—with OpenAI. However, that identification came through reporting and unnamed sources rather than a complete public Oracle customer disclosure.
The careful formulation is that Oracle disclosed a major cloud-services agreement expected to generate roughly $30 billion in annual revenue, and reporting linked at least part of that arrangement to OpenAI’s Stargate capacity deal. It should not be presented as a formally confirmed $30 billion annual payment guaranteed by OpenAI unless a primary filing explicitly establishes that point.
What has actually been delivered?
The announcements indicate that the Stargate buildout was progressing, but they do not show that the entire 4.5 GW was immediately available.
OpenAI reported that:
- Parts of the Abilene facility were up and running.
- Oracle had begun delivering Nvidia GB200-family racks.
- Early training and inference workloads had started.
- Additional sites were planned or under development.
Readers should distinguish among:
- Operational capacity: power and equipment that can actually run workloads.
- Capacity under construction: facilities being built but not yet fully usable.
- Contracted capacity: capacity covered by a commercial agreement.
- Announced capacity: a public plan that may still require financing, permits or grid access.
- Expansion potential: future capacity that can be added if conditions permit.
A planned 4.5 GW is therefore not the same as 4.5 GW of live AI computing capacity.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.The main risks
Power and grid constraints
AI data centers consume enormous amounts of electricity. Projects may face delays in grid interconnection, transmission upgrades, local permitting, fuel supply, water availability or backup-generation requirements.
The electricity demand can also create costs for utilities and communities. Questions include who pays for grid upgrades, whether local reliability is affected and whether power comes partly from natural-gas generation or other sources.
Construction and financing risk
Large facilities require substantial capital before they generate revenue. Higher interest rates, construction inflation, supply-chain disruption or delayed customer demand could reduce returns or extend project timelines.
Oracle’s exposure is also concentrated. A small number of very large AI customers may account for a significant share of new demand, making contract enforceability and customer financial capacity important.
GPU supply and obsolescence
AI infrastructure depends on GPU availability, networking equipment, cooling systems and specialized software. Hardware can become economically less attractive before a long-term facility contract expires. The GPUs planned for one phase of the buildout should not automatically be assumed to represent every later phase.
OpenAI’s ability to absorb the capacity
OpenAI must generate enough demand and revenue to use or pay for a very large amount of infrastructure. Its commitments are substantial relative to its historical revenue base. The economics depend on user growth, enterprise adoption, model pricing, utilization and the terms of its commercial arrangements.
Environmental and community effects
Residents near these facilities may experience construction traffic, noise, land-use changes and pressure on electricity or water resources. Job estimates also require careful interpretation: promised employment may include temporary construction work, indirect jobs, permanent operating roles or a mixture of all three.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →OpenAI estimated that the additional 4.5 GW could create more than 100,000 U.S. jobs. That figure has not been independently verified and should not be read as a guarantee of 100,000 permanent positions.
What the deal means for major partners
- OpenAI: Gains additional training and inference capacity and greater infrastructure diversity, but takes on major long-term financial and operational commitments.
- Oracle: Gains a potentially transformative AI cloud customer and an opportunity to scale OCI, while taking on construction, financing and customer-concentration risk.
- Microsoft: Remains a major OpenAI cloud partner, although OpenAI’s additional providers may reduce Azure’s share of incremental workloads.
- Nvidia: Benefits from demand for GPUs, networking and accelerated-computing systems, while facing the normal risks of supply constraints and technology cycles.
- SoftBank: Remains central to Stargate’s financing and broader development structure, even though the Oracle agreement is a separate partnership within the project.
- CoreWeave and other infrastructure providers: Represent additional sources of specialized AI capacity rather than evidence that Stargate is limited to Oracle alone.
What investors and households should watch
The deal is not a retail product, and individuals cannot buy Stargate capacity directly. Its personal-finance relevance is indirect: it affects the business prospects and risk profiles of major technology, cloud, semiconductor and infrastructure companies.
Useful indicators include:
- Whether announced sites receive power and reach operational status.
- Oracle’s cloud revenue, capital spending, debt and remaining performance obligations.
- Whether large customer commitments turn into recognized revenue and cash flow.
- OpenAI’s ability to monetize model usage at sufficient margins.
- GPU availability, pricing and replacement cycles.
- Utility, permitting and environmental developments near proposed facilities.
- Whether the broader Stargate investment targets are revised, delayed or reduced.
Oracle’s total remaining performance obligations should not be treated as guaranteed profit, cash received or OpenAI-only revenue. Likewise, a company announcement of planned capacity is not the same as an independently verified project completion report.
What is known—and what is not
| Publicly established | Not fully established by the public announcements |
|---|---|
| OpenAI and Oracle agreed to develop up to 4.5 GW of additional U.S. capacity. | The complete contract text and payment schedule. |
| OpenAI later described the partnership as exceeding $300 billion over five years. | Guaranteed minimum spending, take-or-pay provisions and cancellation rights. |
| The arrangement sits within the larger Stargate initiative. | The precise allocation of the $300 billion among cloud services, equipment, construction and operations. |
| Stargate’s original target was $500 billion over four years. | The final financing structure for every site. |
| Parts of Abilene were reported to be operating, with early workloads started. | The final delivery schedule for every announced site. |
| Microsoft continues to provide OpenAI cloud services. | Whether every planned gigawatt will become operational on schedule. |
| Oracle’s company-wide cloud business and financing activity have expanded. | Oracle’s exact margin, ownership of every asset and OpenAI’s guaranteed usage level. |
Bottom line
OpenAI and Oracle did not announce a simple $300 billion check. They announced a large, multi-year infrastructure and cloud partnership centered on up to 4.5 GW of additional U.S. AI data-center capacity. OpenAI later described the partnership as exceeding $300 billion over five years.
Recommended Free Tools
The deal matters because AI growth is increasingly constrained by physical infrastructure: electricity, land, cooling, networking, chips, construction and financing. But the headline value should not be confused with cash already spent, operating capacity already online or guaranteed Oracle profit.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
