Amazon Outage Map

Community Discussion

Tips? Frustrations? Share them here. Useful comments include a description of the problem, city and postal code.

Beware of "support numbers" or "recovery" accounts that might be posted below. Make sure to report and downvote those comments. Avoid posting your personal information.

Amazon Issues Reports

Latest outage, problems and issue reports in social media:

• 
  Emre Yılmaz (@easyparser_emre) reported 5 minutes ago

  everyone thinks scraping amazon is a blocking problem. get past the anti-bot and you're done. that's the easy part. the real problem starts after you already have the html.

• 
  Johnny N. (@Sole_Republican) reported 6 minutes ago

  I am seeing a lot of toxic negativity on the category of building with AI in the Amazon space, so let me give you guys a sneak peek into what I got going on. As you've been following me a while, these screenshots are not from the traditional software I've shared in my past, this one? All me. Here's the benefit I love so much about AI and I think everyone should be taking advantage of, there's a lot of issues I think we are going to be able to keep on solving with AI, but here's one we just solved. First thing first, all our data was scattered everywhere between Inventory Labs, Amazon, SellerSnap, Salesforce, Airtable. It was everywhere. So what I done first was brought everything in-house, good, nice, I have all my data. It feels good, but this is how any business should be run, you should be able to see the data of your business at any given time. Ok, everything became one system, from there my next issue was I didn't want to click into the UI again, I want to just add the ASIN right into a PO that goes into the UI automatically. Done, did that, once we open an ASIN link, we no longer have to go into the UI to see data or add it to a PO, everything was one click and we at the Amazon link, that's all that matters, no more going back and forth between softwares. AI allows you to customize your business the way you want, and you should absolutely be building with it, someone's going to do it, doesn't matter if you do it or not. Feature that your current software or system doesn't support? If you can dream it, it can be made into existence.

• 
  🗡️ Lore Master Chris 🇺🇸 (@CWilson2197) reported 9 minutes ago

  @LostHistory9 They can’t because Amazon would rather throw millions of dollars down the drain than admit that the “chuds” were right the whole time.

• 
  daniel bakalarz (@danbakalarz) reported 11 minutes ago

  Now that $googl, $msft, $meta, and $amzn have all delivered their CYQ2 calls, here's my aggregate-level takeaways on how the ground shifted this quarter. First, the dollars are buying less. $45bn of the 2026 plans is management-admitted input-price inflation, not capacity. Amazon's entire $20bn raise (from $200bn to $220bn) was attributed to the higher cost of memory. And $25bn of Microsoft's $190bn CY26 plan was "the impact of higher component pricing." That's 6% of a combined $732bn buying zero incremental compute! And it's a floor, because Alphabet and Meta haven't disclosed their cost content. So, headline capex growth now materially overstates physical capacity growth. Second, the mix has quietly gone defensive. A data center is really two purchases: the building, which takes years and can't be cancelled once started, and the chips, which have short lead times and can simply not be reordered. MSFT has gone from 50% to 67% chips in three quarters, and its long-lived dollars are DOWN 22% vs FY26 Q1 while total capex is UP 17%. Hood said the implication out loud: "if the demand environment changes, you just slow down what is, in fact, the largest component." Two-thirds of the largest capex program in corporate history is now in the category you can switch off in a quarter. GOOGL is at 60% servers on the same logic. META appears to be doing the opposite: buying shells, deferring servers to 2028. BUT, it's the same trade from the other side: shells are what you can't get on short notice, servers are most of the money, so it locks up the scarce input and leaves the big cheque unwritten. Nobody is spending less. They've just rearranged the spend so that spending less later is a decision instead of a renegotiation. (Worth noting separately, because it's optics rather than flexibility: MSFT extended dc useful lives from 15 to 25 years and is moving future leases from finance to operating. Reported CY26 capex drops from $190bn to $175bn with zero change in actual spend). Third, every one of them is now being explicit about the off-ramp. Hood, unprompted, went further than the mix itself: "The investment into land and data center builds is actually quite flexible." That's a CFO explaining on an open call how she would stop. Meta's Li gets to the same place by sequencing rather than mix: the plan is built to maximise 2026 and 2027 capacity while retaining "the ability to make server decisions when we come to being able to evaluate our actual needs in 2028 and beyond." Hood controls what she buys, Li controls when. And GOOGL is renting third-party capacity in Q3 rather than committing capital to it. Pieces of this have been building for a few quarters, but this is the first window where all four have said it out loud. Fourth, FCF is now the binding constraint. Amazon TTM FCF: -$7.6bn, vs +$18.2bn a year ago, that's a $25.8bn swing, and notably Jassy pre-framed it a full quarter before it arrived. META at +$784mm (statistically zero). GOOGL negative for the quarter, funded by $49.6bn equity and $85bn+ debt in a year. MSFT alone committed to staying FCF-positive in FY27, which is a capex ceiling wearing a different hat (see Hood's comments above). Fifth, and this is the one that ties it together. Points two and three describe an option. Meta is the first company visibly using it. In October 2025 Susan Li volunteered, unprompted, that 2026 capex dollar growth would be "notably larger" than 2025. This quarter she declined to characterise 2027 at all: "we aren't providing a specific outlook for 2027 CapEx at this time. Infrastructure planning remains highly dynamic." In fairness, none of the four has ever given an N+1 dollar figure at a Q2 call, so some of this is just the calendar. But META is the cleanest test of the whole thesis, because it's the only one with no third-party backlog (i.e. $130-145bn of 2026 spend against zero contracted external demand), justified now by the capacity being "extremely valuable" because it's scarce. That's a mark-to-market argument about the asset, not a cash-flow argument about the business. My bet is that if any of these commitments bends first, it's the one with nothing underneath it. The Jan. 2027 call will be the test: a 2027 range and a ceiling above $145bn would mean this Q's refusal meant nothing. Neither, and it was a retreat. My takeaway is that the prisoner's dilemma that has governed this ramp since Q3-24 is taking a different shape. The original bind was that nobody could slow down alone. Spending was the only safe move regardless of what the numbers said. What's changed is NOT the conviction on demand. That's clearly intact: cloud is accelerating, margins are expanding during the build (AWS operating margin 39%, up 650bp YoY, while spending $220bn), and contracted backlog across the three cloud vendors is ~$1.7tn. What's changed is that all four have now rhetorically built themselves a cheap exit. That's rational in a dilemma. IOW, an option that nobody has to exercise costs almost nothing, but it stops being unilateral disarmament once everyone holds one. Eight quarters in, this is the first real rhetorical deviation. Not lower capex. Just the end of the assumption that it could only go one way.

• 
  Debra McGuire (@DebraMc85536001) reported 13 minutes ago

  @helinotes They often get marked down quickly at retailers such as Sam's Club. Used book stores and little free libraries are great resources. Amazon shaves a bit off its books at release. But books ARE expensive.

• 
  SkyFy (@sky_roney) reported 16 minutes ago

  @gofishh77 Amazon can up their theft loss total for tax season. What a terrible way to start your day.

• 
  Chetan Sharma (@Chetansh2406) reported 16 minutes ago

  @AmazonHelp @AmazonHelp Contact support is not trying to solve my problem and your links are not valid.

• 
  fiery storm fox (@FlamingFurry) reported 17 minutes ago

  @amazonmusic @ShaboozeysJeans @slayyyter Amazon, until you fix your app on Xbox live, and issue a patch guarantee that it will finally work after 7 years of it being a piece of garbage, I will continue to complain to microsoft, either demanding that it be patched or delisted, it is complete garbage and does not work.

• 
  Herpderpmcderp (@Herpderpmcderp) reported 18 minutes ago

  @AmazonHelp @SCShipyards Laura, they fix it yet so you can search for Made in America yet? Or does that still error out unlike when you search Made in China or Made in India?

• 
  𓂀 𓋹 (@essenceoflee__) reported 19 minutes ago

  I’m becoming increasingly frustrated with my experience shopping with @amazon . There seems to be a recurring issue with packages being delivered to the wrong address, and it raises serious concerns about the quality and accountability of the delivery drivers being hired.

• 
  BitKey (@Bit_key99) reported 20 minutes ago

  AI earnings are supporting stocks, but rising oil and Treasury yields are quietly reducing the market’s margin for error. Amazon and Microsoft showed that heavy AI spending can still be rewarded when revenue and cash flow follow. Apple showed the opposite lesson: good numbers are not enough when expectations are already high. The bigger risk now is outside tech. Disruption in the Strait of Hormuz is pushing energy and inflation concerns back into the market, while long-term Treasury yields are near multi-year highs. That does not automatically end the AI trade. But it makes weak companies and expensive narratives more vulnerable. I am watching whether QQQ holds its earnings-driven gains while oil and yields stay elevated. Price confirmation matters more than the headline. #Nasdaq #AI #Oil

• 
  John Coffee Hays (@26thAndGuad) reported 20 minutes ago

  @jarvis_best Want watches? 1.) Buy a Casio F-91W. It’s $20, it’s weightless, it tells time and a favorite of both sides of the GWOT. May or may not be the best-selling watch of all time. 2.) Do you like it? A.) Yes, buy a Casio A158 if you want the exact same watch in metal, an A168 if you want a slightly different look without the world’s worst light. B.) No, buy a Timex Expedition Scout. I like the paper-colored dial. Black is nice too. Hold down the crown, it lights up. 3.) Okay, time to go to the beach. Non-negotiable. A.) Casio G-Shock GWM5610-1. Solar powered. Indestructible. 6 band atomic timekeeping. Men want you. Women want to be you. Just roll with it. B.) Casio Duro. It’s a real (quartz) diver. Great dial. Date window. Great bezel. Great case. Awful strap. Comically bad lume. Buy a NATO ******** Amazon. Don’t dive in the dark. 4.) Go buy the other one. It’s like $100. You’ll like it. I promise. 5.) Go watch a bunch of Teddy Baldassarre videos and become a mechanical watch geek. Sorry for ruining your life.

• 
  Brijesh (@Brijesh2498) reported 22 minutes ago

  Ordered a replacement Amazon Pay ICICI FASTag and updated my address via @amazonIN support. Despite confirmation, @BlueDart_ delivered it to my old address! Huge communication failure between Amazon, @ICICIBank_Care & BlueDart. Please fix this ASAP! #FASTag

• 
  lil retard (@comic) reported 24 minutes ago

  @HomoBitcoinicus @dotkrueger Yeah and that’s their own closed loop system… hints my original post. Settlement can happen at the platform layer not the banking layer because the funds are already in their system. That’s why Amazon can issue you cash bank in store credit and pay high % cash back when using their card.

• 
  Lincoln 🇿🇦 (@Presidentlin) reported 25 minutes ago

  Insert image from (@lilyraynyc) I know he is talking about frontier AI discussions (whatever that is) The world is such a big place, all your friends, family, and their families could basically stop using Google, and nothing will really change. Other than ChatGPT, every other standalone chatbot is below Bing. Sample sizes are weird. Honestly speaking Google might find that TikTok and Amazon are bigger threats to its search monopoly than chatbots. It works in Google's favour for everyone to think no one is using them, they can tell Mr regulator, " Search is full of competitors, please don't regulate us". On pricing Google thought the direction for Flash was going up not down, and it did seem that way for a few months, but DeepSeek and OAI have reset that. Most of OAI pricing was to preempt any Xai or Meta moves, but they will just go even lower. The opp is to go after Sol since OAI didn't touch that.