The Q2 numbers behind the hike
According to Amazon's Q2 2026 earnings call, reported by finance.biggo.com, AWS revenue rocketed 36.7% to $42.2 billion for the quarter. The company also disclosed that its AI-related annual revenue run rate has now topped $25 billion — a figure that puts Amazon's AI business alone larger than the total revenue of many Fortune 500 companies.
The results beat Wall Street expectations and sent Amazon shares soaring. TradingKey reported the stock jumped 10% in after-hours trading, while Barron's and TechStock² noted that shares were seen rising as much as 12% as investors digested the stronger-than-expected AWS profit margins.
The surge in AWS profitability was critical to investor confidence. Despite the eye-watering capital expenditure, markets focused on the fact that AWS is generating enough operating profit to make the massive investment look sustainable — at least for now.
Why $220 billion and counting
CNBC reported that Amazon attributed the capex hike partly to rising memory costs — a reference to the surging price of high-bandwidth memory (HBM) and other specialized chips essential for AI training and inference. Memory has become a bottleneck across the AI hardware supply chain, with demand far outstripping the limited production capacity of suppliers like SK Hynix and Micron.
But the bigger driver is sheer demand. Fortune reported that CEO Andy Jassy told analysts that even at $220 billion for the year, Amazon "still won't have enough capacity to meet demand." AWS is reportedly turning away or delaying workloads because it cannot bring data center capacity online fast enough to satisfy customers racing to deploy AI applications.
That admission is remarkable for a company of Amazon's scale. It signals that the constraint on AI growth is no longer just about chips or models — it is about physical infrastructure: data center buildings, power connections, cooling systems, and the electrical grid capacity to feed them all.
The funding question Jassy dodged
One detail that caught analysts' attention was Jassy's response when asked how Amazon plans to finance the additional $20 billion. According to TradingView, when pressed on the call, Jassy offered only: "Nothing to share at this time."
That non-answer is significant. Amazon has already issued substantial debt for AI infrastructure — including a reported $25 billion bond issuance earlier in July 2026 — and the company's total AI-related capital commitments now dwarf its historical spending on fulfillment infrastructure. Investors are increasingly scrutinizing whether the returns on AI infrastructure will justify the enormous outlays, or whether companies are building capacity that could sit underutilized if AI demand disappoints.
For now, the AWS revenue numbers are doing the talking. A 36.7% growth rate on a base of over $40 billion per quarter is extraordinary for a business of that size, and it suggests that enterprise AI adoption is translating into real, recurring cloud revenue — not just pilot projects and proofs of concept.
The hyperscaler spending war
Amazon's $220 billion figure places it at the top of the hyperscaler spending hierarchy, but it is far from alone. Microsoft, Google, and Meta have all announced similarly massive AI infrastructure budgets for 2026. The collective capital expenditure of the top four cloud and AI companies now exceeds $700 billion annually — a sum larger than the GDP of many countries.
As reported by 디지털투데이, the "cloud big three" — Amazon, Microsoft, and Google — all posted record growth in the same quarter despite persistent talk of an "AI bubble." Each is pressing ahead with infrastructure investment, effectively betting that demand for AI compute will continue to compound for years.
The risk, as The Washington Post noted in a separate analysis published the same day, is that America's biggest companies are "burning cash on AI" in ways that could prove risky if the expected returns fail to materialize at the projected scale.
What it means for the AI industry
Amazon's spending decision has ripple effects across the entire AI ecosystem. Every dollar of AWS capex flows to chipmakers like Nvidia and AMD, to data center construction firms, to power utilities, and to the broader supply chain. It also signals to Amazon's cloud customers that capacity constraints — not model quality — may be the defining challenge of the next 12 to 18 months.
For startups and enterprises building on AWS, Jassy's warning about insufficient capacity is a practical concern: AI workloads may face longer provisioning times, higher spot prices, and regional availability constraints. Companies that lock in reserved capacity now may have a significant advantage over those that wait.
The bottom line is that Amazon has chosen to go all-in. The $220 billion bet is a declaration that the AI infrastructure race is a winner-take-most contest — and that Amazon intends to be the one standing on top.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
