The narrative is relentless: AWS is back, firing on all cylinders with AI tailwinds and a $142 billion run-rate. Q4 2025 delivered $35.6 billion in revenue, up 24% year-over-year—the fastest growth in 13 quarters. Amazon's 2025 full-year AWS revenue hit $128.7 billion, up 19%. Analysts trumpet acceleration into 2026. Brutal truth? This is not expansion. It's the last gasp of bloat before enterprises finish optimizing their cloud spend and force hyperscalers into real margin compression and share erosion.
Enterprises didn't suddenly rediscover appetite for unlimited cloud. They spent the past two years mastering FinOps, rightsizing instances, shifting to savings plans, and killing zombie workloads. Cloud cost optimization isn't a side project—it's board-level mandate. Organizations waste 21% of cloud budgets on average, but the survivors have clawed that down aggressively. The result: AWS growth slowed to the high teens for much of 2025 before the Q4 optics bump, even as the overall cloud market expanded 30% in Q4 to $119 billion. AWS's market share has already slid from nearly 50% in 2018 to around 32% by late 2025. Rivals like Google Cloud (growing ~50% in some quarters) and Azure (39%) are eating the incremental dollars.
Data doesn't lie. In early 2025 quarters, AWS posted 17-20% growth on a massive base while competitors posted 30%+ on smaller ones. By Q4, AWS hit 24%—impressive until you realize it's still trailing the market's 30% clip and relies heavily on committed AI workloads that enterprises are now scrutinizing line-by-line. Amazon itself is dumping $200 billion in capex for 2026, much of it AI infrastructure with monetization delayed to 2027-2028. Customers? They're optimizing first, committing second. The backlog sounds massive, but it masks deferred decisions and negotiated discounts that erode effective pricing power.
This is classic late-cycle cloud behavior. Enterprises optimized ruthlessly during the 2023-2024 slowdown—rightsizing, moving to spot/reserved instances, and implementing automated governance. Now, with AI adding new spend layers, CFOs aren't repeating the mistake. FinOps maturity has skyrocketed; tools and processes that once saved 10-15% are now standard, delivering sustained 20%+ efficiency gains. AWS isn't losing customers en masse—it's losing pricing leverage and incremental wallet share as workloads get leaner and multi-cloud strategies distribute risk.
The contrarian reality: AWS's 'growth' is increasingly a function of its own scale and AI hype rather than organic enterprise expansion. On a $142 billion base, even modest optimization across the customer base shaves billions off potential revenue. Meanwhile, capex intensity is exploding—$200 billion in 2026 versus ~$125 billion prior—front-loading costs while returns lag. Operating income for AWS rose to $12.5 billion in Q4, but sustaining margins on optimized spend plus heavy AI buildout will prove brutal. Competitors with fresher architectures and aggressive AI positioning are capturing the high-growth tail.
Investors buying the re-acceleration story are ignoring the denominator effect and the optimization wave. Cloud isn't dying—it's maturing into a cost-disciplined utility. Enterprises have learned the hard way that unchecked spend delivers diminishing returns. AWS remains dominant by inertia, but dominance on a slowing growth trajectory with ballooning capex is a value trap, not a compounding machine.