Meta plans to launch Meta Compute to compete with AWS and Azure

Meta, the parent company of Facebook, Instagram, and WhatsApp, is seriously evaluating the creation of a cloud computing business that would compete with giants like Amazon Web Services (AWS), Microsoft Azure, and Google Cloud. According to a Bloomberg report, the project, internally known as Meta Compute, would allow the company to rent out its excess computing capacity, especially GPUs intended for artificial intelligence. The news has had a strong impact on markets: Meta's shares rose between 9% and 10% following the rumors, reflecting investor confidence that this strategy can alleviate concerns about high AI spending, which has pressured the stock price lower in recent quarters.

What happened?

The news, initially reported by Bloomberg and picked up by TechRadar, indicates that Meta plans to sell idle computing capacity from its data centers. The company has invested heavily in AI infrastructure, with projected spending of between $125 billion and $145 billion in 2026 alone. This figure is comparable to the GDP of countries like Hungary or Kuwait, and represents a significant increase from the $37 billion invested in 2024. Given the possibility that these resources may be underutilized between workloads, Meta seeks to monetize them by offering cloud services to third parties. The business model is not new: SpaceX has already closed deals with Anthropic and Google Cloud to sell its excess computing capacity, generating estimated revenues in the hundreds of millions of dollars. However, Meta's scale could be much larger, given the volume of investment in data centers: Meta operates more than 20 data centers globally, with plans to expand to new sites in the United States, Europe, and Asia.

Mark Zuckerberg himself, Meta's CEO, did not rule out the idea during a call with investors: 'It's definitely on the table,' he said. Although the company has not officially confirmed the launch, the rumors have caused a 9-10% increase in its stock price, reflecting investor confidence that this strategy can alleviate concerns about high AI spending. Historically, Meta has been cautious about diversifying its revenue sources, which depend on digital advertising for more than 98%. A cloud business would reduce that dependence and improve operating margins, which were 35% in 2025, below Google's 40% and far from AWS's 60%.

Why is it important?

Meta's entry into the cloud market would represent a significant shift in the current landscape, dominated by three major players: AWS (32% market share), Microsoft Azure (23%), and Google Cloud (11%). Meta possesses massive infrastructure and AI expertise, which would allow it to offer competitive services. Moreover, the move comes at a time when even giants like Microsoft struggle to meet internal demand: GitHub, owned by Microsoft, recently turned to AWS for additional capacity because Azure could not cover its needs. This fact, reported by TechRadar, underscores the global shortage of AI computing capacity. According to IDC estimates, demand for AI GPUs will grow at a compound annual rate of 40% until 2030, while the supply of chips from Nvidia and AMD can barely keep pace. In this context, Meta could offer an attractive alternative, especially for startups and mid-sized companies facing long waiting lists at established providers.

The business model is not new: SpaceX has already closed deals with Anthropic and Google Cloud to sell its excess computing capacity. However, Meta's scale could be much larger, given the volume of investment in data centers. Additionally, Meta owns its own ecosystem of AI models, such as LLaMA, which has been downloaded more than 300 million times since its launch in 2023. Integrating these models as part of the cloud offering could create a network effect, attracting developers and researchers looking to train and deploy open-source models.

Consequences and outlook

If Meta Compute materializes, it could offer customers access to GPUs for AI training and inference, Meta's own models (such as LLaMA), or the ability to host custom models. This would diversify Meta's revenue, reduce its dependence on advertising, and improve the efficiency of its infrastructure. According to Bernstein, a cloud business could generate additional revenue of between $5 billion and $10 billion annually by 2028, representing 5-10% of projected total revenue. However, the cloud market is extremely competitive and requires high standards of security, compliance, and reliability. Meta will have to demonstrate that it can match or exceed established providers, which have spent decades building relationships with enterprise customers and certifications such as ISO 27001, SOC 2, and FedRAMP. Additionally, the company faces regulatory and privacy challenges, especially in Europe, where the General Data Protection Regulation (GDPR) imposes restrictions on data processing outside the EU. Meta has already had conflicts with European authorities over data transfers between the US and Europe, which could complicate its cloud offering in that market.

Another risk is the cannibalization of its own advertising business: if Meta offers cloud infrastructure to direct competitors like TikTok or Snap, it could leak strategic data or strengthen rivals. Therefore, Meta Compute is likely to focus on non-competing customers, such as healthcare, finance, or manufacturing companies. Additionally, the company will need to invest in a sales and support team, increasing short-term operating costs. Compared to similar moves, such as Google Cloud taking more than a decade to become profitable, Meta may need patience. However, the urgency to justify AI spending accelerates timelines.

What should readers know?

Meta Compute is not yet a reality, but the signals are strong. The company needs to justify its massive AI investment, and a cloud business is a logical path. Analysts believe Meta could succeed quickly due to the shortage of computing capacity in the market. For example, the waiting list for Nvidia's H100 GPUs on AWS is 3 to 6 months, and Meta could offer immediate access to its clusters. However, the project could take months or years to materialize, and its viability will depend on execution and customer adoption. For businesses, this means a new option in the cloud market, potentially cheaper if Meta decides to compete on price. For investors, it is a sign that Meta is seeking to monetize its assets, which could improve its margins in the long term. However, regulatory and operational risks should not be underestimated: if Meta fails to meet security standards or faces privacy fines, the project could fail. In summary, Meta Compute is a strategic bet that could redefine the balance of power in the cloud, but its success is not guaranteed.