CoreWeave Tops Revenue Forecast on AI Cloud Surge, Posts Narrower Loss

CoreWeave’s Q2 Results: Revenue, Loss, and Customer Wins

CoreWeave reported second-quarter revenue of $2.58 billion, edging past the $2.56 billion consensus among analysts, while its adjusted loss per share narrowed to $1.03, compared with the $1.20 loss that Wall Street had anticipated. The results sent shares up nearly 10% in after-hours trading, as investors focused on the strength of the AI cloud computing market that the company serves.

The neocloud provider, which rents out infrastructure powered primarily by Nvidia GPUs, has been a direct beneficiary of enterprises’ growing appetite for AI workloads. CoreWeave has landed cloud capacity agreements with Meta and Anthropic this year, cementing its role as a go-to supplier for large-scale AI projects. The company’s revenue backlog rose to $104.2 billion as of June 30, from $99.4 billion at the end of March, and it disclosed more than $25 billion of net new customer commitments already secured in the current quarter.

To meet that demand, CoreWeave boosted capital expenditures to $9.4 billion in the June quarter, up from $6.8 billion in the prior three months and more than triple the $2.9 billion spent a year earlier. It activated eight new data centers, bringing its global active footprint to 51. Co-founder and CEO Michael Intrator said the second quarter marked an inflection point as scale began to translate into expanding operating leverage, signaling that margins should improve as revenues grow.

The results underscore the relentless pace of investment in AI infrastructure and the niche that specialized cloud providers occupy. While the backlog suggests many quarters of visible revenue, the company’s heavy spending means the path to sustainable profitability remains a key question for the quarters ahead.

What the CoreWeave Earnings Say About AI Infrastructure Demand

Where CoreWeave Sits in the AI Cloud Stack

CoreWeave’s differentiation stems from its deep relationship with Nvidia, giving it early and scaled access to the chipmaker’s most advanced GPUs. This specialist positioning has won it marquee customers such as Meta and Anthropic, which are racing to build and deploy large language models. The $104 billion backlog, up 5% in just one quarter, signals that enterprises view CoreWeave as a credible alternative to the hyperscale cloud incumbents—Amazon Web Services, Microsoft Azure, and Google Cloud—which also offer Nvidia instances but often bundle them with proprietary silicon or broader platform services.

The Capex Juggernaut and the Path to Profitability

CoreWeave’s capital expenditure of $9.4 billion in a single quarter dwarfs its quarterly revenue and highlights the cash-intensive nature of the AI infrastructure race. While the company is still reporting an adjusted loss, the gap between revenue and losses is narrowing, and the CEO’s “operating leverage” comment suggests that fixed costs are being spread over a rapidly expanding revenue base. However, whether this trajectory can continue without a major capital raise or debt injection will be crucial. If the promised scale translates into sustained improvements in gross margins and eventually positive free cash flow, the investment thesis strengthens; if not, the heavy capex could weigh on the stock.

Customer Concentration and Contract Visibility

The $104 billion backlog provides long-term revenue visibility, and the $25 billion in net new commitments so far in the third quarter indicates that demand remains robust. Yet, a significant portion of that backlog likely comes from a small number of hyperscale clients. While the contracts lock in capacity, they may contain clauses tied to performance or volume milestones, and any slowdown in AI spending at a single large customer could materially affect CoreWeave’s growth. The addition of eight data centers in a single quarter demonstrates the speed at which the company is trying to diversify its geographic and customer footprint, reducing over-reliance on any one region or client.

Implications for CoreWeave, Competitors, and AI Investors

For CoreWeave investors: The revenue beat and narrower loss are positive, but the $9.4 billion quarterly capex rate translates to an annual run rate of over $37 billion while revenue is at an annualized $10 billion—watch for commentary on financing plans and gross margin trends in the upcoming earnings call.

For competitors like Nebius and the hyperscalers: CoreWeave’s ability to convert $25 billion in net new commitments this quarter into actual revenue will be a litmus test for the neocloud model; if execution falters, it may slow the market’s appetite for specialist GPU clouds, but if it succeeds, it pressures incumbents to match the Nvidia-focused approach.

For AI enterprises and startups: CoreWeave’s expansion to 51 data centers improves the global availability of Nvidia-based compute, which could ease the GPU supply constraints that have hindered many AI projects—this may be a signal to lock in capacity earlier rather than later.

For the broader AI supply chain: The sheer scale of commitments suggests that underlying demand for AI chips and related infrastructure remains strong, which is a positive read-through for Nvidia’s own revenue growth and for data center equipment providers.