The billion-dollar infrastructure deals powering the AI boom

A lot of computing power is needed to run an AI product -and while the technical industry racet to tap the power of AI models, there is a parallel race to build the infrastructure that will provide them with electricity. With a recent win, the Nvidia -CEO Jensen Huang estimated that between $ 3 trillion and $ 4 trillion will be spent on the end of the decade on AI infrastructure -with a lot of that money from AI companies. Along the way they put enormous pressure on power grilles and push the construction capacity of the industry to the limit.

Below we have created everything we know about the largest AI infrastructure projects, including large editions of Meta, Oracle, Microsoft, Google and OpenAi. We keep it informed as the tree continues and the figures climb even higher.

Microsoft’s investment of $ 1 billion in OpenAI

This is demonstrably the deal that caught the entire contemporary AI -boom: In 2019Microsoft has made an investment of $ 1 billion in a buzzy non-profit called OpenAi, best known for its association with Elon Musk. Crucial is that the deal Microsoft made the exclusive cloud provider for OpenAi – and as the demands of model training became more intense, more started to come from Microsoft’s investments In the form of Azure Cloud Credit Instead of cash.

It was a lot for both parties: Microsoft was able to claim more Azure sales and OpenAi received more money for the biggest expenses. In the years that followed, Microsoft would build his investment up to nearly $ 14 billion-one movement that has been set to pay enormously when OpenAi turns into a profit motive.

The cooperation between the two companies has been won more recently. In January, OpenAi announced that the Microsoft’s Cloud would no longer use exclusively, instead to give the company a right to give the first refusal to future infrastructure requirements, but if others could not meet their needs. More recently, Microsoft began to explore other foundation models to provide its AI products with electricity, as a result of which even more independence of the AI ​​giant was determined.

The OpenAI with Microsoft regulation was so successful that it has become a usual practice for AI services to register with a certain cloud provider. Anthropic has received $ 8 billion in investments from Amazon, while Make changes at kernel level About the company’s hardware to make it better suited for AI training. Google Cloud has also signed on smaller AI companies such as Lovable and Windsurf as ‘primary computer partners’, although those deals do not entail any investments. And even OpenAi went back to the well and received an investment of $ 100 billion from Nvidia in SeptemberGiving capacity to buy even more of the company’s GPUs.

The rise of Oracle

On June 30, 2025, Oracle unveiled in a SEC entering that it had signed a cloud services of $ 30 billion with a public partner; This is more than the company’s cloud income for the entire previous tax year. OpenAi was eventually unveiled as the partner and secure Oracle A place next to Google As one of OpenAi’s series of post-microsoft hosting partners. It is not surprising that the shares of the company are included.

A few months later it happened again. On September 10, Oracle unveiled a five -year -old deal of $ 300 billion for computing power, set up in 2027. Oracle’s shares climbed even higher, making founder Larry Ellison the richest man in the world. The huge scale of the deal is amazing: OpenAi does not have $ 300 billion to spend, so the figure assumes enormous growth for both companies and more than a little trust.

But before a single dollar is spent, the deal has already cemented Oracle as one of the leading AI infrastructure providers – and a financial force that must be taken into account.

Building the hyperscale data centers of tomorrow

For companies like Meta who already have Significant legacy infrastructureThe story is more complicated – although equally expensive. Mark Zuckerberg has said that Meta is planning to spend $ 600 billion on the American infrastructure Until the end of 2028.

In just the first half of 2025, the company spent $ 30 billion more than the previous year, largely driven by the growing AI ambitions of the company. Some of those expenses go to large ticket cloud contracts, such as a recent $ 10 billion deal with Google CloudBut more resources are poured into two huge new data centers.

A new site of 2,250 hectares in Louisiana, Dubynchronized HyperionAn estimated $ 10 billion to expand and to provide an estimated 5 gigawatt of computing power. In particular, the site contains an arrangement with a local nuclear power plant to handle the increased energy tax. A smaller site in Ohio, called Prometheus, is expected to come online in 2026, powered by natural gas.

That kind of Buildout comes with real environmental costs. The Xai van Elon Musk built his own hybrid data center and Power-Generation factory in South Memphis, Tennessee. The plant quickly became one of the largest emitters of smog -producing chemicals in the province, thanks to a series of natural gas turbines that Experts say the Clean Air Act is violating.

The Stargate Moonshot

Only two days after his second inauguration, President Trump announced a joint venture between Softbank, OpenAi and Oracle, intended to publish an AI infrastructure of $ 500 billion in the United States. Name “Stargate” after the 1994 film, the project arrived with incredible amounts of hype, where Trump called “the largest AI infrastructure project in history. Sam Altman seemed to agree and said:” I think this will be the most important project of this era. ”

In broad strokes, the plan that Softbank provides the financing was, with Oracle de Build -out with input from OpenAi. The supervision of it all was Trump, who promised to clear all regulatory obstacles that the build could delay. But there were doubts from the start, including Van Elon Musk, Altman’s Business Rival, who claimed that the project did not have the available funds.

As the hype has disappeared, the project has lost some momentum. In AugustBloomberg reported that the partners did not reach a consensus. Nevertheless, the project has progressed with the construction of Eight data centers in Abilene, TexasWith the construction of the last building that will be completed by the end of 2026.

This article was first published on September 22.