Its medium and long term profitability Neocloud Not clear at all. Even experts disagree on its future. These companies are mainly dedicated to Rent computing power From their data centers to artificial intelligence (AI) laboratories that need infrastructure to train and run their models. Some of the most important ones are CoreWeave, Lambda, Together AI, TensorWave or Nebius. The first four are American and the last is in the Netherlands.
The explosion of AI that we’ve been experiencing for four years now and the fever to build more data centers around the planet invites us to find out. Neocloud They must be doing very well. And yes, at least some of them are doing well. Dutch Nebius earned $582 million in the second quarter of 2026, although its annual revenue had already reached $3 billion at the end of June. CoreWeave, for its part, had a turnover of $2,575 million in the same period, more than double that of a year earlier.
Also, AI laboratories have achieved some very juicy promises with some of them. OpenAI, Thinking Machines Lab, Perplexity, and Cognition have more than $140 billion in deals between them. Neocloud Processing power Crusoe. Some experts, such as Nvidia CEO Jensen Huang, believe these companies have a tremendous future. However, others, such as SpaceX, Tesla and XAI leader Elon Musk, argue that many of them will go bankrupt in the medium term.
A huge investment requires adequate financing
Neither Elon Musk nor Jensen Huang are uninterested in this area. Nvidia leads the market in chips for training and inferring AI models, so labs and companies that own data centers will have unimaginable benefits ahead of the AI catharsis we’re experiencing. And xAI has turned some of its infrastructure into a business that rents computing power to third parties.
Jensen Huang maintains that the difficulty some of these companies are having in obtaining liquidity is temporary
Be that as it may, it’s interesting to see what arguments these two tech industry leaders use to support their thesis. Elon Musk believes that companies that are exclusively dedicated to renting infrastructure to train and run AI models will have a tough time in the future. When it comes to getting financing. According to him, SpaceX raised $40 billion in debt to buy AI chips because of its credit rating and multiple sources of revenue.
As I mentioned a few lines above, Jensen Huang positions himself on the other side of the scale. Nvidia’s leader assured that demand for computing power will eventually support the business Neocloud. Huang argues that the difficulty some of these companies are having in obtaining liquidity is temporary, and that demand for computing services is so high that it will trigger the emergence of new forms of financing.
On August 10, Nvidia announced an agreement with six financial institutions, including BlackRock, Goldman Sachs, and Blackstone, to combine more than 500 billion dollars On third-party capital to finance AI infrastructure. If it is good, it will help, no doubt, but we must not ignore that the stock markets are reacting negatively to the possibility of these companies’ debt becoming unsustainable. We won’t have to wait long to see this as 2027 is likely to be a defining year in this sector.
Picture | Nebius
More information SCMP
In Xataka | China has plans to win the AI war against the US. And Dipsik’s champion
