Amazon's free cash flow went $42.4 billion negative and it kept the AI budget anyway. The question that asks what you'd protect.
On October 29, 2026, TechCrunch reported that Amazon had missed analysts' expectations for both total revenue and earnings per share in its third-quarter results. The report had one bright spot: Amazon Web Services, which TechCrunch said "beat revenue expectations and grew a whopping 39% compared to the year-ago quarter." The number that unsettled investors sat elsewhere. Free cash flow for the trailing 12 months fell to a negative $42.4 billion, which TechCrunch described as the company "spending far more cash than it generated," against positive free cash flow of $14.8 billion for the trailing 12 months ended September 30, 2025. Amazon said the spending was driven largely by its investments in AI, and, according to TechCrunch, reiterated that it still expects to end 2026 with cash capital expenditures of about $220 billion. For scale, TechCrunch noted that Amazon forecast fourth-quarter net sales of between $240 billion and $247 billion, and observed that the company expects to spend almost as much on capital over the full year as it expects to bring in from sales in its largest quarter.
Read the report as a set of decisions rather than a scorecard. A company with cash flow that far in the red has not stopped choosing. It has chosen, in public and again, to hold one line of spending steady while every other number is allowed to disappoint. Whether the bet is right is a question for investors. What matters here is the shape of it: more ambition than the quarter can fund, a commitment the company declined to revise, and an explicit account of what is being protected and what is being let go. Engineering managers make a version of that call every quarter, usually at a scale where nobody writes a headline and often with far less clarity about the cost. That is why an earnings miss is also, exactly, an interview question.
Why they ask it
Every manager has cut scope. The question is not whether you did it but whether you did it on principle. Interviewers at senior levels are listening for three things. First, that you had criteria, and can state them, for deciding what the date actually required versus what had attached itself to the project along the way. Second, that you knew who the cut would land on and talked to them before deciding, not after. Third, that you treated the cut as a trade-off with a cost, recorded it, and owned the follow-up. A candidate who can describe all three has shown the interviewer how they will behave the next time a date and a backlog disagree, which at any company is soon.
The trap
The most common weak answer is a story about the date rather than the decision. It narrates the crunch, the late evenings, the heroic final week, and arrives at "we shipped on time" without ever saying what was removed or why. That is a story about effort, and the interviewer asked about judgment. The second trap is pretending the cut was free. "We dropped the reporting module and nobody missed it" sounds like a win, but if nobody missed it, it was padding rather than scope, and the interviewer will wonder what else in your plans is padding. The third trap is cutting whatever was easiest to cut, usually the work that was least started, rather than the work the outcome could most afford to lose. Those are often the same item, but not always, and the gap between them is precisely what the question is probing.
Applying STAR-T
Situation. A payments team was migrating off a processor whose contract ended on a fixed date, with no extension on offer. With weeks to go, the plan was tracking well behind, and the project had grown to include a new reconciliation dashboard and self-serve merchant refunds alongside the core cutover.
Task. The manager owned the ship decision and the relationship with finance and support, who were the internal users of the two features at risk.
Action. The manager listed every remaining work item and sorted it by a single question: does the contract date require this, or did it join the project because it was convenient to build at the same time? Processing transactions on the new provider and verifying that no historical data was lost were non-negotiable. The dashboard and self-serve refunds were not. Before cutting them, the manager sat with the finance lead and confirmed that reconciliation could run from a nightly export for a couple of months, and with support that refunds could be handled by ticket at current volume. The two features were removed from the release and written up with named owners and target dates. The data verification work, which was the slowest and least visible item on the board, was kept and staffed first.
Result. The cutover landed days before the contract ended with no lost transactions. The dashboard shipped later that quarter; self-serve refunds shipped the quarter after.
Trade-off. Finance ran reconciliation by hand for most of two months and support absorbed a steady stream of extra refund tickets. The manager said so in the retrospective, because a cut described as costless is not believed, and a manager who cannot name the cost is suspected of not having looked for it.
The follow-up that breaks weak answers
"What happened to the work you cut?" A weak answer treats a cut as a deletion and has nothing to say. A strong answer knows that scope removed from a date does not vanish; it moves, to a later release, to a manual process, or onto another team's plate, and the manager can say where it went and whether it ever shipped. That is the same discipline the Amazon report invites. When a company stops protecting cash generation in order to protect a spending plan, the cash does not disappear; it shows up as a negative number that someone has to explain. The candidate who can point to where their cut landed, and what it cost the people it landed on, has answered the question the interviewer was actually asking.
Score your answer against the director’s bar
Q: Describe a time you had to cut scope to hit a date, and how you decided what to cut.
Bank the scope cut you still think about, and rehearse it until the criteria come out before the heroics. Try it free →