The world often changes faster than we can predict, and how people choose to build through that change matters as much as what they build.
Neither idea is new to us. They are stitched into how we operate, and point to the same conclusion. Innovating through change is often where the best work happens, so long as it is done with discipline.
That matters in a moment when ambition is everywhere and scale is not. Plenty of companies talk about AI. The difference is who delivers it.
This is how CoreWeave became The Essential Cloud for AI. It’s how we earned the trust of foundation model providers, AI natives, and enterprises across the world. And it’s why we have conviction about our role in a global economy reshaped by technology.
And that discipline produced another incredible quarter for our business. We are growing profitably on an operating basis, and we’re doing it at scale.
- Revenue reached $3.7 billion, up 168% year-over-year, which is above the high end of our guidance, with more than $1 billion added quarter over quarter.
- Our adjusted operating income is $379 million, well above our guidance of $200 million to $260 million. On top of that, our margins moved into the low double digits a quarter earlier than planned.
- Revenue backlog grew to approximately $133 billion.
- The global footprint of our active power increased to more than 1.7 gigawatts, doubling our live footprint since the start of the year.
- We continued to extend our platform for customers, unifying our technology under CoreWeave Forge, enabling them to build, run, and continuously improve AI in a single environment.
Of course, none of this happened in a quiet market. Narratives shifted, sentiment swung, and people scrutinized the buildout of AI. What our latest results show is a business executing the right way, at speed, and with customers who keep choosing us.
As we look to the future, we are confident in where we stand. Here is why:
The Ecosystem is Not a Monolith
Some clouds were developed within larger, established businesses. CoreWeave built a purpose-built AI cloud from scratch against contracted demand.
Companies in this market differ in their technology, their customer contracts, their relationships with the communities where they operate, and how they finance growth. They should be measured against those differences. Because that decides who will last.
There will be waves of industry consolidation, shifting competition, and evolving public sentiment. That is how new industries form. This one happens to be both digital and physical, and it is built in real places, among real neighbors.
Where and how you build matters. The companies that succeed will be the ones who earn the trust of the communities who host them.
So the question is not only who is building, but who is building well. Supply remains constrained. Demand remains relentless. The key tests are who builds against real demand rather than speculation, who is deploying at scale, who has a diverse set of contracted customer commitments, and who has a record of executing at speed responsibly.
Scale Matters
We have more than 1.7 gigawatts of active power online today, across a global footprint of data centers. That includes new markets we've entered, such as Indonesia and India, while growing our footprint in Denmark and Spain. That is a fundamentally different position from holding a pipeline of sites still waiting on a grid connection, and it matters because energized capacity is capacity customers can use now.
Customers don’t come to CoreWeave to rent GPUs. They come to run their AI workloads on our fully-integrated AI cloud — from compute, storage, and networking, through the full application layer — all at hyperscaler scale and across our globally interconnected connected footprint.
Put plainly: Getting thousands of GPUs into a data center is a tough challenge. Operationalizing hundreds of thousands of them to work together as one system is even harder. The chips have to be connected, the right software and engineering is required to orchestrate them, and the whole platform has to keep running while customers put real workloads on it. That is what sets CoreWeave’s AI cloud apart, and it is what our customers find indispensable.
We have created an advantage that is increasingly difficult to replicate. Our fleet of infrastructure is the foundation of our AI cloud and an investment for the long term. It’s becoming a meaningful source of capital in its own right as more of it comes off contract every year and generates cash flow without new spending. And if barriers to building rise, the strategic value of what we've already operationalized grows.
Building early was only the start. The race for capacity will ultimately give way to a measurement of capability, and the companies on the far side of this buildout will be judged on what they built when the noise was loudest. How did customers turn an idea into a reality? How reliably does it run? How much work does it get out of every ounce of compute? What were customers able to do with it?
How We Built It
From the outset, a central principle has guided CoreWeave. Every decision is tied to an optimal economic choice: how we fund, what we charge, and how fast we build. Customer contracts come first, and investment follows. Because CoreWeave’s investments are tied to committed demand, loans for the GPUs and other assets we finance are backed by contracted cash flows. That gives lenders confidence to lend against them, and customers typically help fund the buildout through prepayments.
Our financing is drawn from a broad and diverse set of sources – from prepayments to delayed-draw term loans, to our growing operating cash flow. It’s why we will not use the unsecured debt market to fund our business at current yields. Nor do we need to.
The same discipline sets our pace. It lets us calibrate investment to customer demand, the regulatory landscape, and the capital markets. It is how we reached more than 1.7 gigawatts of active power, and how we are delivering capacity ahead of plan.
CoreWeave is already a company of consequence. Nine of the top 10 AI foundation model providers run on our AI cloud. AI natives use our platform like Wayve and Cognition. And traditional enterprises are following, including Caterpillar, Nissan, Jane Street, and many others. They see what AI can do for their businesses and for the work their people do every day.
Our platform combines leading price performance and reliability with the innovation, software, and development tools that enterprises depend on in one connected environment, open across models, frameworks, and other clouds. Customers use it to accelerate their AI strategies by bringing applications to market faster and improving their performance with every release.
This is how better technology becomes better business. Customers increasingly start with our software, and our performance makes us central to how they build and run AI.
Will It Be Worth It?
That is the question beneath the others, and we think the answer lies in people more than in technology. What makes AI powerful is what it lets people do. The limiting factor on what a person can achieve is no longer access to knowledge, capital, or credentials. It is access to the technology that makes action possible.
At the heart of CoreWeave is a fundamental belief that discovery is a human pursuit, and that AI is going to advance the ability to explore and innovate.
That holds for the labs pushing the frontier. It holds for enterprises applying AI to problems humanity has struggled with for millennia. It holds for a small business owner looking for new ways to connect with their customers, and for people who have limited technical background but are now able to act on their singular ideas.
When the dust settles, this will be about more than the number of chips bought or data centers built. This is a clarifying moment about what we can accomplish when human ingenuity pairs with technology. The point of all this building is to power the ideas still to come.
We have built the AI cloud those ideas deserve.
