Lights dimmed in the boardroom as the spreadsheet refreshed. I watched the free cash flow number blink negative, and you could feel the room inhale. That blink—$42.4 billion (≈€39 billion)—was the loudest sound of the week.
I’ll walk you through what happened, why it matters to your portfolio or product plans, and where the real danger signs are hiding. You don’t need to be an investor to see the angle; you just need to know which numbers to watch next.
On the trading floor, traders compared notes about Prime Day timing before the bell—then analysts saw the quarter’s revenue miss
Amazon reported $201.9 billion (≈€186 billion) in revenue, up 12% year over year, yet still below Wall Street’s forecasts. You probably felt the prickle of déjà vu if you watched this week’s hyperscaler earnings: cautious growth, heavy AI spending, and share-price wobble.
Amazon said Prime Day’s calendar shift moved meaningful sales into the prior quarter, softening year-over-year comps. That’s plausible, but it’s not the whole story: advertising and retail revenue both underperformed, suggesting friction at the company’s core commerce machine.
At a server farm, engineers pointed at racks of new hardware—then the finance team saw the capex spike
Free cash flow plunged to a negative $42.4 billion (≈€39 billion) from a negative $7.5 billion (≈€7 billion) in the prior quarter, while the company reported it spent an extra $94.8 billion (≈€87 billion) on property and equipment tied to AI build-out.
Amazon now expects roughly $220 billion (≈€202 billion) in capital spending this year. I’ll tell you plainly: that pace will torque investors’ patience. The build resembles like a dam bursting—massive inflows now, uncertain downstream returns later.
Why did Amazon burn $42.4 billion in Q3?
Because Amazon is buying compute, datacenters, and custom silicon at scale. CEO Andy Jassy pointed to Trainium—Amazon’s own AI accelerator chips—and Bedrock, its model marketplace. Those are long bets on reducing dependence on Nvidia and on selling AI services; they cost money now in exchange for potential margin relief later.
But spending without immediate revenue lift compresses free cash flow, and investors punish that mismatch. You should ask: is this capex buy-now for cheaper compute later, or capex that simply buys speculative market share?
During an earnings call, executives touted Alexa usage increases while investors frowned at ad revenue—then the stock dropped
Jassy defended the company’s AI strategy and flagged Alexa for Shopping: interactions were up 570% year over year, and U.S. customers using it reportedly spend 50% more per order than those who don’t. That’s a persuasive metric—you care about average order value.
Still, AI shopping agents are emerging across the industry. Meta launched Muse; others are building agentic assistants. Some analysts warn these agents could reroute purchase intent away from Amazon’s ad stack. Jassy insisted that Alexa will remain the primary in-house agent, complete with stored payment, shipping, and purchase history—a convenience moat he argued is hard to replicate.
Can Alexa save Amazon’s ad business?
Short answer: maybe, if adoption scales and clicks become conversions. You’ll want to monitor active users, order frequency, and ad attachment rates inside Alexa sessions. The headline stat—50% higher spend per order—sounds good, but small-sample or novelty effects can fade.
On the scoreboard of hyperscalers, traders compared notes while share prices slumped—then Microsoft stood alone
Of the four major hyperscalers that reported this cycle, only Microsoft posted positive free cash flow, and only Microsoft saw its shares climb after results. Meta showed negative free cash flow of $2.12 billion (≈€2 billion), and Alphabet reported negative free cash flow of $9.07 billion (≈€8 billion), the latter suffering its worst one-day drop in history.
Amazon’s shares fell about 7% after the release. That correlation—big AI bets plus negative cash flow equals investor churn—has been the market’s message this week. You should be asking whether Microsoft’s model (less capital intensity, bigger software margins) is the safer path or simply the shorter-term winner.
In the machine rooms, engineers debated chip strategy—then investors tracked competitive risk
Trainium aims to challenge Nvidia’s dominance; Bedrock aims to be a neutral marketplace for models. Those are legitimate tools in a long game. But you need to read the risk: owning silicon and data centers makes Amazon more vertically integrated—and more capital-heavy.
Think of Amazon’s strategy as a high-stakes chess match where every move requires a significant investment on the board. If those investments deliver proprietary advantages—cheaper inference, sticky developer adoption, higher margins—Amazon wins. If adoption lags, the spending becomes a visible vulnerability.
At the investor call, executives pointed to metrics while the market priced risk—then you decide what to watch next
I want you to watch three things: capex guidance and pace, ad-revenue trends inside retail, and adoption metrics for Alexa and Bedrock. Those will tell you whether the cash burn is a temporary phase or a structural problem for margins and free cash flow.
Tools and platforms in sight: Trainium vs. Nvidia’s GPUs, Bedrock competing with Google Cloud AI and Microsoft Azure AI, and Muse from Meta as a user-facing agent experiment. These companies are all rewriting the AI infrastructure playbook, but the bill is due now while the returns are still uncertain.
You’ll see loud claims on calls and glossy product demos; my job is to point out the metrics that matter. Which will you watch—daily cash flow, capex pace, or the ad-revenue trend that proves Alexa can convert browsers into buyers?