CoreWeave Stock Forecast 2026: Revenue Projected at $12.4 Billion, Up 142% — If the Build-Out Goes to Plan

Go past Core.At Weave’s headquarters in Livingston, New Jersey, a suburban office complex close to the kind of corporate parks that used to house pharmaceutical back offices and insurance companies, you wouldn’t think you were looking at a company at the epicenter of one of the most significant infrastructure buildouts in recent technology history. The structure doesn’t make an announcement. However, something quite different from insurance is taking place inside and in data centers located throughout the United States and Europe. The contracts arriving in April 2026 appear to be confirmation that the infrastructure wager the company made years ago is paying off. CoreWeave is operating the GPU clusters that an increasing portion of the AI industry depends on.

In 2017, the business was founded as Atlantic Crypto, a GPU mining company that responded to market demands. CoreWeave had shifted to AI cloud computing by 2019, long before the AI compute boom turned the GPU market into something akin to a national strategic resource.

In hindsight, that timing is more important than it may have seemed at the time. During a time when practically no one else was considering it that way, the company developed expertise in high-density GPU infrastructure and Nvidia relationships. CoreWeave had an advantage over its more recent rivals by the time the demand materialized. Bernstein’s analysts recognized that CoreWeave has “by far the strongest commercial machine” among what the industry now refers to as neoclouds—GPU-first providers that operate alongside but apart from AWS, Azure, and Google Cloud—even though they continued to rate the stock as Underperform.

It has been difficult to process all of the contract news over the last few weeks at once. Meta increased its $14 billion commitment to CoreWeave’s platform to $21 billion through 2032. The Motley Fool interpreted it as an indication that the stock “might finally be turning the corner”—carefully worded language for a result that seems less speculative as the backlog increases.

Anthropic revealed a multi-year agreement to run its Claude AI models on CoreWeave infrastructure the day after Meta’s announcement. Anthropic’s status as one of the more significant frontier AI labs gives the announcement significance beyond its monetary value, even though financial details were not made public. Subsequently, Jane Street, a quantitative trading company that isn’t usually linked to AI cloud news, signed a $6 billion expansion contract and invested $1 billion in CoreWeave Class A shares at a price of $109 apiece. The final component is what distinguishes a vote of confidence from a customer relationship.

The revenue figures have been improving at a rate that challenges credibility. Revenue for the entire year 2024 was $1.92 billion. The total for 2025 was $5.13 billion, a 168% increase. The analyst consensus for 2026 is about $12.4 billion, which would be an additional gain of 142%. The longer-term bull case is driven by the revenue backlog, which was $66.8 billion at the last report, up 342% year over year. Goldman Sachs is modeling CoreWeave’s growth trajectory through the end of the decade using that backlog, which represents contracted future revenue. The company’s Q4 2025 quarterly revenue of $1.57 billion, which increased by 110% year over year, provides the 2026 projections with at least a head start.

All of this optimism is counterbalanced by the debt load, which should be treated honestly. With $4 billion in convertible senior notes and $1.75 billion in senior notes issued in private offerings earlier this month, CoreWeave’s total debt has surpassed $14 billion. On April 16, the company revealed that it plans to offer an extra $1 billion in senior notes with a 9.75% maturity date of 2031. This represents significant borrowing costs for infrastructure that won’t fully yield returns for years. With a net loss of about $1.2 billion in 2025, the business is not anticipated to turn a profit anytime soon. Despite raising their target price, Bernstein’s analysts maintained their Underperform rating, citing dilution risk from convertible bonds and the potential that the market is assigning valuations that need near-perfect execution to justify. It is not unreasonable to be skeptical.

Analyst notes and investor commentary frequently make the Tesla comparison, and it’s not totally incorrect. Before its unit economics began to turn a profit, Tesla spent money and accumulated debt during years of infrastructure investment. In technology, there is a clear pattern of significant early capital expenditures followed by high-margin recurring revenue. The management of CoreWeave has stated that they anticipate GPU costs to eventually break even or better and that the capex curve has a natural ceiling because, in contrast to streaming services or social networks, you can only construct a certain number of data centers before supply meets contracted demand. Although the exact date of that ceiling’s arrival is still unknown, the $66.8 billion backlog indicates that it hasn’t been reached yet. Whatever else, it’s one of the most fascinating stories in technology at the moment to watch this company manage the gap between the contracts it’s signing and the profitability it hasn’t reached.