CoreWeave reported Q3 revenue of $1.4 billion, up 134% year-over-year, despite operating in what the company describes as a "highly supply-constrained environment." The AI infrastructure provider added over $25 billion in revenue backlog during the quarter, bringing its total remaining performance obligations to more than $55 billion.
The company reached $50 billion in RPO faster than any cloud provider in history. CoreWeave now serves triple the number of customers generating over $100 million in annual revenue compared to the previous year, indicating enterprise AI workloads are consolidating around specialized infrastructure rather than traditional cloud platforms.
The growth trajectory mirrors broader momentum in AI-optimized computing. NVIDIA, CoreWeave's primary GPU supplier, continues outperforming the semiconductor sector as enterprises race to deploy large language models and generative AI applications. Traditional hyperscalers like AWS and Azure offer GPU compute, but specialized providers are capturing high-value enterprise customers with purpose-built infrastructure and direct allocation of scarce chip supply.
CoreWeave's backlog composition suggests sustained demand through 2026 and beyond. The $55 billion figure represents multi-year contracted commitments, providing revenue visibility that public cloud providers rarely disclose. With supply constraints persisting across the NVIDIA H100 and upcoming Blackwell GPU lines, CoreWeave's ability to secure chip allocations gives it a structural advantage over smaller competitors.
The company's unicorn valuation reflects investor confidence that specialized AI clouds will capture meaningful share from general-purpose platforms. As model training and inference workloads grow more compute-intensive, enterprises are prioritizing performance and availability over the flexibility of traditional cloud architectures.
Q1-Q2 2026 will test whether CoreWeave maintains triple-digit growth as comparisons toughen and new GPU capacity comes online. Customer concentration remains a risk—if a handful of $100M+ accounts represent most revenue, any churn could materially impact results. Still, the 134% growth rate and accelerating backlog additions suggest enterprise AI infrastructure spend is only beginning to scale.