A new analysis suggests that the massive spending by AI companies is turning cloud providers like Amazon Web Services (AWS) into highly profitable 'AI tollgates'.
The key driver behind this shift is the emergence of enormous, long-term contracts. Recently, AI leader Anthropic committed over $100 billion across ten years to secure up to 5 gigawatts of computing power from AWS. Not to be outdone, OpenAI also expanded its own AWS deal by about $100 billion. These aren't just vague promises; they are 'anchor-tenant' agreements that provide a predictable, multi-year revenue stream for AWS, transforming abstract 'AI tailwinds' into concrete financial figures.
Interestingly, this isn't an exclusive AWS story. The market is shaping up to be a multi-lane highway. Anthropic is also securing massive capacity from Google Cloud (GCP) for its specialized TPU chips, and OpenAI has moved beyond its initial exclusive relationship with Microsoft Azure. This diversification means that both AWS and GCP are becoming essential tollgates for the frontier AI industry, competing to attract the biggest players.
To put this in perspective, Wells Fargo estimates these deals could add about $25 billion to AWS's revenue in 2026 alone, which is nearly 20% of its entire 2025 revenue. For GCP, the contribution could be around $9 billion. At AWS's current profit margins, that $25 billion in revenue could translate to nearly $9 billion in operating income from just two clients. This is the kind of impact that gets investors' attention.
These developments didn't happen overnight. They are built on a foundation of earlier strategic moves, such as Amazon's initial $4 billion investment in Anthropic and its proactive securing of a massive nuclear power deal to ensure it has the energy to run these AI factories. These steps laid the groundwork for the mega-deals we see today.
However, two significant hurdles remain. First, there's execution risk: building data centers and securing grid-scale power is a monumental task fraught with potential delays. Second, there's regulatory risk. The U.S. Federal Trade Commission (FTC) is already scrutinizing these tight partnerships between cloud and AI companies. Any future rulings against exclusivity could weaken the 'tollgate' power of these providers. The narrative is clear, but the road ahead depends on execution and navigating policy.
- Glossary
- Anchor Tenant: A primary, large customer whose long-term commitment provides a stable financial foundation for a larger project, much like a major department store in a shopping mall.
- Hyperscaler: A large-scale cloud computing provider, such as AWS, Google Cloud, and Microsoft Azure, capable of providing enormous compute, storage, and networking resources globally.
- AI Tollgate: A metaphor describing a cloud provider that AI companies must pay to access the massive computing power required to train and run advanced AI models.