Analysis
CoreWeave shares jumped roughly 19-20% this week after the company reported second-quarter revenue of $2.58 billion, more than double the $1.21 billion it posted a year earlier, with losses coming in better than analysts feared. The stock climbed further after the company disclosed a $104 billion revenue backlog -- contracted future revenue anchored by accelerating AI cloud demand and rapid capacity deployment across its GPU-focused infrastructure, and lined up another $2.6 billion in financing toward more than $30 billion of planned 2026 capital spending.
Goldman Sachs analyst Gabriela Borges raised the firm's price target on CoreWeave to $139 from $121 while maintaining a Neutral rating, noting that demand is outpacing supply, GPU pricing remains firm across chip generations, and there's a credible path to higher margins as new capacity comes online.
What CoreWeave actually is
CoreWeave began as a cryptocurrency-mining operation before pivoting entirely to renting Nvidia GPU capacity to AI labs and enterprises, a model sometimes called a "neocloud" -- specialized AI infrastructure providers that compete with the hyperscalers' own cloud offerings by moving faster and specializing narrower. The $104 billion backlog figure is the number that matters most for the business model: it represents contracted future revenue, not current-quarter revenue, meaning CoreWeave's growth trajectory is largely locked in via multi-year customer commitments rather than dependent on quarter-to-quarter demand fluctuations.
The competitive landscape
- CoreWeave -- $2.58B Q2 revenue, $104B backlog: GPU-focused neocloud infrastructure provider. Competitors: Nebius, Lambda, and hyperscaler-native offerings from Microsoft Azure, AWS and Google Cloud, all competing for the same Nvidia GPU allocation and AI-lab customer base. Source
CoreWeave's advantage over the hyperscalers is speed of capacity deployment and specialization; its disadvantage is balance-sheet scale against Microsoft, Google and Amazon, all of whom can absorb short-term margin pressure that would be existential for a company financing $30 billion of 2026 capex primarily through debt and specialized lending arrangements.
The stock's post-earnings pop also arrived in the same week Nvidia guided Q3 revenue to $108 billion and Marvell, a supplier one layer removed from CoreWeave's own GPU demand, got sold off despite beating estimates -- a reminder that investor enthusiasm for AI infrastructure right now is concentrated unevenly, rewarding companies closest to Nvidia's own guidance beat far more than companies merely adjacent to the AI capex cycle.
The counterweight
A revenue backlog is a promise, not cash in hand -- it depends on customers honoring multi-year contracts even if their own AI spending plans change, and CoreWeave's business model requires continuous heavy capital investment to keep pace with that backlog, meaning the company remains structurally dependent on capital markets staying open and cheap. CoreWeave's own stock has been volatile in both directions this year on debt-market sentiment shifts, a reminder that a company this financially levered to the AI buildout is a leveraged bet on the cycle continuing, not a hedge against it slowing.