Amazon posts negative free cash flow of $42.4 billion, holds 2026 capex unchanged
- ID
- 40815
- Status
- summarized
- Published
- 30 Oct 2026, 6:24 AM
- Fetched
- 30 Oct 2026, 6:52 AM
- Provider
- CNBC Technology
- Category
- technology
- Original URL
- https://www.cnbc.com/2026/10/29/amazon-amzn-q3-earnings-report-2026.html
- Source URL
- https://www.cnbc.com/id/19854910/device/rss/rss.html
Summary
- Score
- 5.5
- Created
- 30 Oct 2026, 6:52 AM
- Tags
- Audience
- developersai_ml_learnerssaas_startup_founders
What happened
This CNBC piece is a pre-earnings preview published before Amazon's Q3 2026 call, not the results write-up: it lists analyst consensus of $1.96 adjusted EPS on $202.31B revenue, with AWS expected at $45.48B (38% growth, a fifth straight quarter of acceleration) and advertising at $20.82B. The most concrete operational detail is CEO Andy Jassy saying Amazon's cloud capacity is "largely sold out" for next year, with significant commitments already booked for 2028, while some capacity is held back for on-demand and Bedrock/EC2 inference. Note the headline's $42.4B negative free cash flow and unchanged 2026 capex figures do not appear anywhere in the supplied text.
Why it matters
If you are planning to scale inference or training workloads on AWS next year, the signal here is allocation, not price: Jassy says capacity is largely committed through 2027 with 2028 already being booked, leaving only a slice for on-demand and Bedrock/EC2 inference. That argues for locking in commitments or designing a multi-provider fallback now rather than assuming you can burst on demand later — and for treating the headline FCF/capex numbers as unverified until you see the actual results, since they are absent from this text.
Discussion angle
If hyperscaler capacity is effectively pre-sold into 2028, what does that mean for a Malaysian team's cloud plan — reserve early, or architect for portability across providers and regions?