In today's Cloud Wars Minute, I explain why hyperscalers are rewriting the rules of deal-making to build the next generation of AI infrastructure.

Highlights

00:01 — We are seeing the beginnings here of an incredible round of innovation, not just in technology, but in deal-making, partnerships, alliances, and financing, all by the hyperscalers trying to meet this insatiable AI demand. We're seeing these companies undertake some very innovative, bold, distinctive new strategies to build the capability and capacity to get these AI data centers built out to meet this insatiable demand.

00:49 — Google Cloud did a joint venture with Blackstone, in which Blackstone invested $5 billion into the joint venture. We have seen Amazon issue a series of debt and bond offerings totaling over $100 billion. AWS has said that in calendar year 2026 it will spend $200 billion on CapEx, most of which is going into AI data centers. Oracle announced $50 billion in debt and equity financing.

01:57 — This funding, this raising of funds to build out the data centers, is because there is, among these hyperscalers, over $2 trillion in committed contracted business. While Oracle right now is the smallest by revenue of the hyperscalers, it has the largest backlog, and in order to meet that, it has to spend a lot of money to build the capacity.

02:46 — Microsoft is using proceeds from its brilliant early relationship with OpenAI to help secure some of the funding. Under a newly restructured agreement between the two companies, Microsoft now will receive 20% of OpenAI revenues for the next few years. Plus, Microsoft has a huge ownership stake in OpenAI.

04:17 — Remarkable things are going on here as the technology buildout by all these companies has helped create this incredible demand. What we're seeing now is extraordinary efforts by the hyperscalers to combine with other companies, move into different industries, and do everything possible — at staggering expense — to meet this insatiable customer demand for AI.

Visit Cloud Wars for more.