Meta Platforms is doubling down on artificial intelligence infrastructure. The company has signed a massive deal with Nebius Group NV , committing to spend up to $27 billion over the next five years to secure advanced computing capacity needed to build and run its AI systems.
The agreement highlights how aggressively major technology companies are investing in data centers and chips as the race to build powerful AI models intensifies.
What the deal includes
The agreement is structured in two parts, giving Meta both guaranteed computing power and the option to expand further as demand grows.
- $12 billion in dedicated AI infrastructure that Nebius will provide exclusively to Meta starting in early 2027
- Up to $15 billion in additional computing capacity that Meta can purchase from infrastructure Nebius is building for other customers
- The total commitment could reach $27 billion , making it one of the largest single infrastructure contracts Meta has signed
This arrangement gives Meta long term access to specialized data centers designed specifically for training and running large AI models .
Why Meta is spending so heavily
Artificial intelligence has become Meta’s top strategic priority . The company is building its own advanced AI models while integrating AI tools across its platforms including Facebook, Instagram, and messaging services.
To support this push, Meta needs enormous computing power.
- Training large AI models requires thousands of high performance GPUs
- Running AI services for billions of users requires massive data center capacity
- Companies want to secure infrastructure years in advance to avoid shortages
By locking in capacity now, Meta ensures it will have the computing resources required as AI adoption accelerates.
Part of a much larger AI spending wave
Meta is not alone in making these investments. The biggest tech companies are expected to collectively spend around $650 billion on AI infrastructure in 2026 , according to industry estimates.
Much of that spending is going toward:
- Building new AI focused data centers
- Purchasing advanced GPUs
- Developing custom AI chips
- Securing long term cloud infrastructure agreements
Meta has already signed several other large partnerships this year to support its AI strategy.
- Multi billion dollar infrastructure partnerships with Nvidia and AMD
- Development of in house AI chips
- Expansion of internal data centers
CEO Mark Zuckerberg has also said the company plans to spend about $600 billion on US infrastructure projects by 2028 .
Who Nebius is and why it matters
Nebius is a relatively new but fast growing player in the cloud infrastructure market.
- The company is headquartered in Amsterdam
- It was spun off from Yandex in 2024
- Its business focuses on building AI optimized data centers
Companies like Nebius are often called “neoclouds.” Unlike traditional cloud providers such as Amazon or Google, these firms focus specifically on AI workloads.
Nebius also has a strategic partnership with Nvidia , which supplies the powerful chips needed to train modern AI models.
This focus on AI infrastructure has driven strong investor interest. Nebius shares jumped about 15 percent in premarket trading after news of the Meta deal, and the stock has nearly quadrupled over the past year .
Nvidia’s role in the AI infrastructure boom
Nvidia has played a major role in the rise of companies like Nebius. The chipmaker has been actively financing cloud startups that purchase its GPUs.
Recent investments include:
- $2 billion investment in Nebius
- $2 billion investment in CoreWeave , another AI focused cloud provider
- Participation in a $2 billion funding round for UK based Nscale
- A massive $30 billion investment in OpenAI
Some critics argue that these investments create a circular ecosystem where Nvidia funds companies that then spend heavily on its chips. Supporters say the strategy accelerates the expansion of AI infrastructure worldwide.
What this means for the AI race
The Meta Nebius deal shows just how expensive the AI race has become.
Building competitive AI systems now requires:
- Advanced semiconductor chips
- Massive data centers
- Long term infrastructure commitments
- Billions of dollars in capital
For companies like Meta, securing infrastructure early could make the difference between leading the next generation of AI services or falling behind rivals like Google, Microsoft, and OpenAI.
One thing is becoming clear across the tech industry: the future of AI will be built not only with algorithms, but with enormous amounts of computing power.