18 hours ago
23
While Microsoft Azure has been the market’s darling for AI-driven growth, largely thanks to its cozy relationship with OpenAI, analysts now expect Amazon Web Services to deliver superior returns on its AI capital expenditures over the coming years.
The numbers behind the shift
AWS currently commands roughly 30% of cloud infrastructure market share, compared to Microsoft’s approximately 21%. That gap has been a durable feature of the cloud landscape for years, even as Azure has posted faster revenue growth rates, often in the 30-40% year-over-year range.
Microsoft’s growth advantage has been largely attributed to its OpenAI integration, which turned Azure into the default on-ramp for enterprises experimenting with generative AI.
Amazon has announced plans to allocate $75 billion in capital expenditures for 2024, with AI infrastructure as the primary focus. AWS Bedrock, its managed AI service, reported 170% quarter-over-quarter customer spend growth in Q1 2026, with token volumes exceeding all previous totals.
Enterprise budgets are ballooning
The total cloud market is expected to reach $330 billion in 2024. TD Cowen surveys indicate that enterprise cloud spending attributed to generative AI is projected to grow fourfold over three years, with acceleration through 2026-2027.
Microsoft’s rapid Azure expansion has come with margin pressure, a natural consequence of scaling infrastructure ahead of demand. Until utilization rates improve across new data centers, the gap between capital deployed and revenue generated can weigh on returns.
What this means for crypto and decentralized compute
When AWS and Azure collectively pour over $100 billion into AI infrastructure in a single year, they’re validating a thesis that compute demand is exploding beyond what any single provider can efficiently serve. That’s exactly the demand gap that decentralized compute protocols, things like Render Network, Akash, and io.net, are designed to fill.
These protocols aggregate idle GPU capacity from distributed providers and offer it at lower price points than centralized cloud. For AI inference workloads, training runs that don’t require enterprise-grade guarantees, and cost-sensitive startups, decentralized alternatives become more attractive as centralized pricing reflects massive capex recovery needs.
The fourfold growth in enterprise AI spending projected by TD Cowen implies that demand will increasingly spill over into alternative providers.
Investors in the crypto space should watch two things closely. First, utilization metrics from decentralized compute protocols, not just token prices but actual GPU hours consumed and revenue generated. Second, enterprise AI budget surveys like TD Cowen’s, which serve as leading indicators for how much compute demand is heading toward non-traditional providers.
Construction timelines for hyperscale data centers run 18-24 months minimum, and power grid constraints are becoming a real bottleneck in key markets. That gap between AI demand growth and centralized supply growth is the window decentralized compute needs.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.
English (US) ·