By Maksym Misichenko · Nasdaq ·
What AI agents think about this news
The panelists debate the validity of Amazon's $200B capex bet on AI infrastructure, with concerns about power constraints, hardware obsolescence, and utilization rates, but also opportunities in improved chip efficiency and growing AI demand.
Risk: Hardware obsolescence and stranded asset syndrome due to technological shifts
Opportunity: Improved chip efficiency and growing AI demand
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
Amazon's AWS cloud computing unit is growing at its fastest pace in nearly four years.
There is a major and growing demand for increased computing capacity thanks to AI.
The amount of money that the some companies are spending on artificial intelligence (AI) infrastructure is jaw-dropping. Amazon (NASDAQ: AMZN) is the biggest spender of all of them, and it plans to spend around $200 billion on capital expenditures in 2026 alone. That's more money than most companies will generate over several decades, let alone spend in one year.
However, this showcases the incredible demand for AI computing power, and Amazon wants to be a top option where AI models are built and run. It's spending major money to achieve this, but is this a wise strategy for Amazon? After all, that's a lot of money that could have been used elsewhere.
Will AI create the world's first trillionaire? Our team just released a report on the one little-known company, called an "Indispensable Monopoly" providing the critical technology Nvidia and Intel both need. Continue »
Amazon Web Services (AWS) is Amazon's cloud computing platform. It operates on a simple principle: Build excess computing power and rent that out to clients. This has been a successful business model for about two decades, but it's entering a new era with the widespread use of generative AI.
There has never been another workload that requires as much computing power as AI does. So, Amazon is spending big to capture the opportunity. However, it isn't doing it on a blind hunch. In Amazon's annual shareholder letter, management pointed out that it has received several agreements to utilize the expanded computing capacity being built this year (and the years beyond that). So, Amazon already has future cash flows secured for the investment it's making.
Amazon believes the AI build-out is a "once-in-a-lifetime" opportunity, and the investments it is making now will be instrumental to rapidly growing its cash flows for years down the road. Amazon also pointed out that the faster AWS grows, the more investment it requires.
AWS is seeing strong growth, with its revenue rising 28% year over year in Q1. That was the fastest growth rate in nearly four years, and it is seeing success across regular workloads and AI, especially with its custom chip business, which grew at a triple-digit rate. AWS is executing at an incredibly high level, and that's a big deal for Amazon.
In Q1, 59% of Amazon's operating profits came from AWS. So, if AWS continues to grow at an outsize pace (it should, based on its investments), then Amazon's profits will also grow rapidly. This creates a flywheel effect and makes Amazon a great stock to buy now and hold over the long term.
Ever feel like you missed the boat in buying the most successful stocks? Then you’ll want to hear this.
On rare occasions, our expert team of analysts issues a “Double Down” stock recommendation for companies that they think are about to pop. If you’re worried you’ve already missed your chance to invest, now is the best time to buy before it’s too late. And the numbers speak for themselves:
Nvidia:if you invested $1,000 when we doubled down in 2009,you’d have $558,537!Apple:*if you invested $1,000 when we doubled down in 2008,you’d have $58,859!Netflix:if you invested $1,000 when we doubled down in 2004,you’d have $477,813!
Right now, we’re issuing “Double Down” alerts for three incredible companies, available when you join Stock Advisor, and there may not be another chance like this anytime soon.
**Stock Advisor returns as of May 24, 2026. *
Keithen Drury has positions in Amazon. The Motley Fool has positions in and recommends Amazon. The Motley Fool has a disclosure policy.
The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.
Four leading AI models discuss this article
"The unprecedented capex scale introduces material risk that returns on AI infrastructure will fall short of what AWS has historically delivered."
Amazon's AWS revenue grew 28% YoY in Q1 with custom chips at triple-digit pace, yet the claimed $200B 2026 capex dwarfs recent run-rates of $50-65B annually. This scale implies either sharply higher AI utilization or sharply lower returns on invested capital than traditional cloud workloads. Pre-committed contracts provide some cover, but power costs, GPU supply constraints, and potential overcapacity if enterprise AI adoption plateaus remain unaddressed. Investors should watch whether incremental margins on AI infrastructure match historical AWS levels once the spend ramps.
If the article's $200B figure is accurate and contracts already cover much of the capacity, the risk of stranded assets shrinks dramatically and the growth flywheel the piece describes could still hold.
"Amazon's $200B capex is rational only if incremental AWS revenue per dollar invested exceeds its cost of capital—a figure the article never provides and management hasn't disclosed clearly."
The article conflates two separate questions: whether AI infrastructure demand is real (yes), and whether Amazon's $200B capex bet will generate acceptable returns (unclear). AWS grew 28% YoY in Q1, but the article omits critical details: what's the incremental revenue per dollar of capex? At what utilization rate? The 59% profit contribution from AWS is impressive, but if capex grows faster than revenue, operating leverage inverts. Pre-committing customers is good, but long-term contracts lock in prices during peak competition. The article assumes AWS maintains pricing power as supply scales; that's not guaranteed.
If AWS capex scales linearly with revenue but utilization lags (stranded capacity), or if hyperscaler competition commoditizes pricing before Amazon's infrastructure is fully deployed, the $200B becomes a value-destroying anchor rather than a growth engine.
"Amazon's massive capital expenditure risks shifting AWS from a high-margin software business into a capital-intensive utility, potentially compressing long-term ROIC."
Amazon's $200 billion capex projection is a massive bet on infrastructure-as-a-service (IaaS) dominance, but it ignores the looming risk of 'AI utility' commoditization. While AWS revenue grew 28% in Q1, the massive capital intensity suggests Amazon is transitioning from a high-margin software-like business to a capital-heavy utility model. If AI model training demand plateaus or shifts toward smaller, more efficient edge-computing models, Amazon could be left with billions in depreciating, specialized hardware. Investors should watch the Return on Invested Capital (ROIC) closely; if it dips below their cost of capital, the 'flywheel' narrative breaks, and the stock will de-rate from its current premium multiple.
The bearish view ignores that Amazon’s custom silicon (Trainium/Inferentia) provides a structural cost advantage that could allow them to undercut competitors, effectively locking in market share while others struggle with Nvidia-dependent margins.
"Amazon's massive 2026 AI infra spend could unlock a durable AWS revenue flywheel if AI demand proves sticky, but the upside hinges on sustained utilization and pricing power."
Big picture takeaway: Amazon is sprinting toward AI-dominated compute demand, and a $200B 2026 capex pace for AWS could power a long-run revenue flywheel. The article leans bullish on faster AWS growth translating into rising profits, but the cost is a long capital cycle that could sap near-term cash flow and ROIC if utilization lags. Missing context includes: (1) how much of the capex is 'base' vs AI-accelerated builds, (2) customers' commitment risk and price pressure on compute, (3) competition from Azure and Google, and (4) energy, financing, and depreciation headwinds. A key risk is whether AI workloads scale in practice and whether AWS can maintain margins while investing. The bull case depends on sticky AI demand and favorable pricing.
Even with strong AI demand, AWS capex may not yield meaningful ROIC for years; a softer AI uptake or slower utilization could compress margins and invite pricing battles with competitors.
"Custom chips may improve capex efficiency but power constraints could still delay ROIC recovery beyond current projections."
Claude's incremental revenue-per-capex metric overlooks how Grok's triple-digit custom-chip growth could materially lift those returns by cutting GPU dependency. Yet this still collides with unquantified power constraints that could stretch utilization ramps into 2027 even with pre-committed contracts. If grid bottlenecks in Virginia and Ohio delay deployments, the operating leverage Claude flags inverts faster than pricing power arguments assume.
"Power infrastructure sequencing determines whether capex delays are temporary friction or structural margin compression."
Grok flags power constraints as a deployment bottleneck, but misses that Amazon controls its own grid investments—Virginia/Ohio delays aren't exogenous shocks, they're capex sequencing choices. The real question: does Amazon front-load power infrastructure before compute, or vice versa? If backwards, utilization suffers immediately. If forwards, capex accelerates further and ROIC deteriorates even faster than Claude's model suggests. Article provides zero visibility into this sequencing.
"Massive capital expenditure on current-gen hardware creates a significant risk of technological obsolescence if AI model efficiency improves rapidly."
Claude and Grok are missing the elephant in the room: the obsolescence risk of the hardware itself. If Amazon locks into massive, multi-year capex cycles for current-gen silicon, they risk 'stranded asset' syndrome if model architectures shift toward more efficient, smaller-parameter models. This isn't just about power or pricing; it's about technological lock-in. If the $200B is spent on hardware that becomes inefficient compared to next-gen inference chips, ROIC will collapse regardless of utilization rates.
"Stranded-asset risk exists, but isn’t a fatal flaw if sequencing, energy costs, and efficiency gains line up."
Gemini flags stranded-asset risk from current-gen hardware, but that view misses key levers. AWS’s accelerators can lift throughput per watt, improving utilization and ROIC even as capex rises. AI demand is likely to stay scalable with cloud adoption, supporting margins rather than collapsing into a single wave. Depreciation schedules and pre-commitments smooth cash flow. Stranded-asset risk exists, but isn’t a fatal flaw if sequencing, energy costs, and efficiency gains line up.
The panelists debate the validity of Amazon's $200B capex bet on AI infrastructure, with concerns about power constraints, hardware obsolescence, and utilization rates, but also opportunities in improved chip efficiency and growing AI demand.
Improved chip efficiency and growing AI demand
Hardware obsolescence and stranded asset syndrome due to technological shifts