CoreWeave

A $1.5B IPO in March 2025 put CoreWeave at the center of the AI-compute buildout, followed by major customer commitments and debt-market scrutiny.

Who they are

CoreWeave is a cloud provider of Nvidia GPU-based AI computing capacity whose coverage centers on the financing, power and data-center infrastructure needed to serve large AI customers. It became a public company under ticker CRWV in 2025 and appears alongside Nvidia, Microsoft, Meta, OpenAI, Anthropic and data-center operator Core Scientific.

The recent arc

Coverage peaked around CoreWeave’s 2025 public-market transition: it filed to list on Nasdaq after reporting 2024 revenue of $1.92 billion and an $863 million net loss, then raised $1.5 billion at $40 a share in a scaled-back IPO. Reporting subsequently focused on converting AI demand into contracted capacity, including an approximately $4 billion OpenAI agreement through 2029 alongside an earlier $12 billion deal and the proposed roughly $9 billion all-stock acquisition of Core Scientific to expand data-center capacity.

In 2026, the story shifted from listing and expansion plans to the scale and financing of delivered infrastructure. Nvidia added a $2 billion investment tied to an effort to exceed 5 GW of AI computing capacity by 2030; Meta expanded its commitment by $21 billion on top of a prior $14.2 billion deal; and CoreWeave announced a multiyear Anthropic agreement while saying it had 43 active data centers. Its August results reported $2.58 billion in Q2 revenue, up 112% year over year, a $104 billion backlog and 1.5 GW of contracted power capacity. Later coverage also flagged a CoreWeave-tied data center’s $1.1 billion junk-bond sale and reports that it was considering chip-price hedges.

The tension

The central tension is between extraordinary contracted demand for AI compute and the capital intensity of supplying it. CoreWeave’s customer and investor ties to Nvidia, plus large commitments from Meta, OpenAI and Anthropic, support the expansion narrative, but debt moved into SPVs, higher-yield financing for a tied data center, and consideration of memory and storage-price hedges show how exposed the model is to infrastructure costs, financing conditions and customers’ AI spending.

Why it matters

CoreWeave is a visible test of whether specialized GPU clouds can turn AI workload commitments into durable infrastructure economics at public-market scale. If its backlog, power buildout and major customer contracts translate into sustained utilization, it could validate an alternative route to AI capacity outside the largest hyperscalers; if demand expectations or funding costs weaken, its reliance on rapid capacity additions and complex financing could become the defining constraint.

Related: Nvidia · IPO · Core Scientific · Meta commits to spending additional $21B on AI cloud infrastructure fr · Nvidia invests $2B more in CoreWeave to speed up an effort to add 5GW+ · CoreWeave says it has signed a multiyear deal with Anthropic, includin