Artificial Intelligence (AI) is expensive. As a technology, it constantly needs more things to advance: more power, more data centers, more memory, more money.
As early as 2024, OpenAI CEO Sam Altman said that the development cost of ChatGPT-4, the latest version of the large language model (LLM) at that time, exceeded $100 million.
Will artificial intelligence create the world’s first trillionaire? Our team just released a report on a little-known company that has been described as an “essential monopoly” that provides critical technology that both Nvidia and Intel need. continue”
Many companies want to develop artificial intelligence, whether for commercial use or for their own internal purposes. But many either cannot afford it or would rather avoid spending hundreds of millions or even billions of dollars to develop their own AI infrastructure.
Fortunately for them, there are core weaving (NASDAQ: CRWV)which provides potential AI developers with the tools and hardware they need to develop AI without having to bear all the costs associated with it themselves.
CoreWeave offers a range of products targeted at AI developers. For example, there’s CoreWeave ARENA, which is basically a rentable AI lab on the company’s cloud.
In it, customers can test the performance of their AI programs to see how efficiently it scales and how much it costs. It runs on CoreWeave’s own data center network, which spans North America and Europe. By accessing CoreWeave’s system, start-up AI developers don’t have to spend a lot of money to build their own data centers.
The company also offers a service called Slurm on Kubernetes, or SUNK, an industry first. It allows customers to train their artificial intelligence programs using CoreWeave’s network. This is important because when OpenAI launched the first version of ChatGPT, training new AI models wasn’t cheap, and it’s only getting more expensive.
A study by Epoch AI shows that the cost of training new AI models has increased by 2 to 3 times per year over the past eight years. If this pace continues, it could cost more than $1 billion by 2027 just to train an LL.M.
But with CoreWeave, smaller AI companies can, in some cases, train AI faster and cheaper than their competitors. French artificial intelligence developer Mistral AI trained its namesake artificial intelligence program entirely on CoreWeave’s cloud, training it 2.5 times faster than it could manage on its own.
It’s not just small companies that use CoreWeave. international business machine It is also a major customer, working with CoreWeave to develop and train its Granite model 1.8 times faster than before using the CoreWeave service.
OpenAI also uses CoreWeave to help create its own models. This makes sense; they know firsthand how expensive it is to develop artificial intelligence.
How does this translate into CoreWeave’s financial results? In fact, very good.
CoreWeave ends 2025 with incredible results.
Although the company fell short of profitability with an adjusted net loss of $606 million, its revenue increased 168% from 2024 to $5.1 billion. New business is also pouring in.
The company’s order backlog grew 342% to $66.8 billion by the end of 2025. Therefore, the demand for its services is obviously very high.
More than half of the backlog is for long-term deals of 25 months or more, so revenue will continue to grow significantly going forward.
CoreWeave expects better results in 2026 and set full-year guidance at $12 to $13 billion. The lower guidance is more than double 2025 total revenue.
A bit optimistic? perhaps.
But even as concerns about a looming AI bubble kept investors on edge throughout 2025, CoreWeave still managed to grow at an impressive rate. With new deals emerging at an alarming rate, CoreWeave’s growth guidance is likely to be met.
There’s no shortage of demand, and CoreWeave already has long-term partnerships with tech giants like IBM.
Even with losses, CoreWeave already looks strong. If it can become profitable in the relatively short term, it will likely remain an important part of the AI market infrastructure.
Now look at this, it’s down from its all-time high in June 2025, but still up 81% from its March 2025 initial public offering (IPO). With the growth numbers it’s releasing, CoreWeave appears to be hitting its stride.
Before buying CoreWeave stock, consider the following factors:
this Motley Fool Stock Advisor The analytics team has just identified what they believe is 10 Best Stocks Investors can buy now… and CoreWeave isn’t one of them. The 10 stocks selected could generate huge returns in the coming years.
consider when Netflix This list was created on December 17, 2004… If you invested $1,000 when we recommended, You will have $522,791!* or when NVIDIA This list was created on April 15, 2005… If you invested $1,000 when we recommended, You will have $1,132,678!*
Now, it’s worth noting stock advisor The overall average return is 952% — outperformed the market compared to the S&P 500’s 191%. Don’t miss the latest top 10 list, available via stock advisorand join an investment community built by individual investors for individual investors.
See 10 stocks »
*Stock Advisor returned on March 11, 2026.
James Hires has no position in any of the stocks mentioned. The Motley Fool has jobs and recommendations for International Business Machines Corporation. The Motley Fool has a disclosure policy.
This artificial intelligence (AI) stock has a $66 billion backlog. Here’s why it will soar in 2026. Originally posted by The Motley Fool