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:

• 
  SHAKA🤞🔥 (@GulfamAli0) reported 6 seconds ago

  @AmazonHelp "I have followed you, but DM is not working. Can you please DM me first?"

• 
  Aman Kumar (@amanaryan23) reported 6 seconds ago

  The thing engineers who crack Google, Microsoft, Meta, Amazon, Apple, Uber etc in their first attempt do that nobody in your prep group is talking about. The ones who crack it on attempt #1 share one habit that almost nobody else has. The answer: They prepare after taking the mock interviews, not just the interview. After every mock interview, they write down: → Where exactly did I slow down? → What did I say when I got stuck? → Did I clarify before coding or just dive in? → Did I talk through tradeoffs or just pick one? → What would an ideal candidate have done differently here? Most people finish a mock and think: “that felt okay” or “that was bad.” The engineers who crack big tech on attempt #1 finish a mock and think: “Here are the 3 specific things I will do differently next time.” That’s not motivation. That’s a system. Systems and correct preparation beat just hard work and talent. Every time. That’s why mock interviews are so important and valuable and that can be a difference between an offer or failure. Are you doing post-mock debriefs? Be honest below 👇

• 
  Jonah 📦 (@jonahhodges_) reported 4 minutes ago

  Day 1: Listing audit. We pull up every product they have on Amazon. → Check the images. → Check the copy. → Check the backend search terms. Most of the time it's a mess. One blurry image, bullet points written a decade ago, zero A+ content, no videos. We document everything that's broken and build the fix list.

• 
  Brian Phan (@brianphanU) reported 5 minutes ago

  The 2012 gTLD winners look even quieter. .google. .amazon. .apple. .bmw. .audi. You don't see them. That's the point. Every Google product, every Amazon login, every BMW dealer page now resolves through a namespace those companies own and lock. 14 years of compounding moat.

• 
  Brandon 🇺🇸 (@Bishop0051) reported 6 minutes ago

  @REDWRITER @eazy_peach If it ever goes on Amazon for free I'll have to watch tho. If all is as it claims to be it will be a thumbs down

• 
  Minesh Bhadauriya (@minesh956) reported 9 minutes ago

  @AmazonHelp Issue Not resolved yet, serious lack of support from Amazon

• 
  Lou from Long Island (@lou_fiore) reported 12 minutes ago

  Hey @AmazonHelp, I’ve been locked out of my account for 8+ days after a hack. Your team added a passkey to "protect" me, but it was set up for a device I don’t own. I’m stuck in a support loop that keeps promising a 24-hour fix that never happens! Tkt# P427804364 #AmazonSupport

• 
  DK (@deekey1101) reported 19 minutes ago

  @AmazonHelp Have already done that. Issue was supposed to get resolved by today, but your customer care team needs another 24hrs time to find a resolution. The most inefficient team I must say. Now I am forced to go to Consumer court and lodge a complaint

• 
  Aakash Gupta (@aakashgupta) reported 20 minutes ago

  A $599 Mac mini sitting on a shelf is now the most underrated PM productivity setup in 2026. Mahesh Yadav has been an AI PM at Amazon, Google, Meta, and Microsoft. The pattern he just walked through on the podcast: install OpenClaw on a Mac mini, point it at Opus 4.6 plus an open source fallback, connect WhatsApp, write a skill in plain English. Now you message your phone, the agent runs the job for 6 hours on a separate machine, and the result lands back in your chat. You never open a terminal. You never hit Anthropic's daily cap, because you can route to a local Llama model when you do. You never wait for IT approval, because you own the sandbox. The math is the part most *** miss. $599 for the base M4 Mac mini is cheaper than one month of heavy Claude Code usage at the Max tier. The mini sits there 24/7, runs jobs while you sleep, and never closes the lid. One-time hardware cost, infinite runtime. Sending a message to a sandboxed agent that reproduces your problem, tries a solution, and returns a result is how enterprise AI agents will work at scale. Google won't let you install OpenClaw on a work laptop. But GCP will ship this exact pattern inside their infra. AWS will ship it inside theirs. The *** who understand the delegation pattern now will be the ones shipping the enterprise version in 2027.

• 
  Abhishek singh (@Abhishe47825734) reported 20 minutes ago

  @AmazonHelp I have contacted your team multiple times through email, but it has been 20 days and my issue is still not resolved.This is mental harassment. Your team has not taken any action for the last 20 days and my issue is still unresolved.

• 
  Henryk Sarat (@henryksarat) reported 21 minutes ago

  @haiyami9x @aaalexhl It might look like Coinbase is just tokens and trading, but there's way more under the hood. High level brain dump from my time in the industry and what I know they're working on: * Coinbase has 12 product lines doing $100M+ in revenue. That's just the ones above $100M. To get there, a ton of products failed or got absorbed along the way. Plenty of tech debt as a result. Impressive that some became breakout hits like this. Start-ups fail to even find any PMF and fizzle out. * They offer staking to custody customers. Adding fault-resistant staking for a new chain is not trivial. You really don't want enterprise clients getting slashed. Entire companies exist just for this (Kiln, Figment, etc). * Enterprise-grade custody is a huge wedge for them and they win a lot of deals. The wallet infra alone is brutal from a security standpoint. Whole companies do only this: BitGo, Fireblocks, Anchorage, Paxos. * Bank connectivity is constantly changing. Adds, removals, rail swaps, never ends. * AML tooling has to keep evolving. The threat landscape and regulatory expectations move every quarter. * Base started on the OP Stack with Optimism's help since Coinbase didn't have that expertise in-house. They've since moved off it and are doing their own L2 work. Maintaining that infra and growing a chain ecosystem is capital intensive and takes deep expertise. Look at what Solana has to do just to keep their ecosystem running. Not easy. * Base wallet needs constant updates and a real differentiator story. Entire companies do only wallets: Phantom, Consensys, etc. * Every new asset or chain needs resilient RPC infra to support custody, staking, and trading. Coinbase runs their own. Entire companies do only RPC: Alchemy, Helius, Infura, QuickNode. When I was at Paxos, certain RPC providers didn't have certain chains and were adding new ones monthly. Keeping up with the full ecosystem is very hard. * Signing infra has to be secure and scalable. This is the most sensitive part of the whole company and gets harder with every new chain since each one has its own quirks and security considerations. * Crypto-as-a-service. They power PNC, Webull, and 200+ other institutions. Whole companies do only this (Zero Hash, Paxos). * Stablecoin payment rails. Net new initiative, brand new integrations. * Coinbase membership is a constantly evolving product. Hard to get the value/cost balance right. I worked on Lyft Pink and the considerations were endless. As rideshare, bikes, scooters, and transit evolved, we constantly reshaped what was offered. One painful example: figuring out Chase member benefits and integrating Lyft and Chase systems so the 10x points counted correctly, plus easy onboarding and offboarding for cardholders. Other partner perks came and went over time (Grubhub, DoorDash, Amazon). Never ending. * I didn't touch security (constant threats and attack vectors), data (data is ever evolving and tooling changes). * I didn't even touch non-engineering teams and the regulatory landscape always changing which requires engineering system changes. * Didn't talk about the changing landscape of prediction markets, perps, and tokenized stocks which is a whole different story. * Left out everything the possibility of agentic payments and that whole industry that we don't even know what it could look like. It's still an unsolved problem. Very big sector for crypto and stablecoins. * Plus everything else: build tools, onboarding, compliance, threat monitoring, dev tooling, SDKs, ecosystem programs, etc. Not going deeper since this is already long. Another data point. When I led the engineering launch of USDL at Paxos, I coordinated across 8+ engineering teams plus security and data. We built net new infra. Going from a simple stablecoin to a yield-bearing stablecoin sounds trivial but it wasn't. The systems were built around a simple ERC20 stablecoin because that's what Paxos offered at the time, so the assumptions baked in everywhere had to be reworked. the regulatory landscape was constantly changing which added thrash to our requirements on the engineering side. My PM counterpart did the same on the non-eng side, leading 8+ teams (legal, product, marketing, BD, etc). What the team pulled off in 1-2 quarters was honestly pretty wild. One more point I’ll push back on is the idea that they should stop investing in the business and just “collect cash.” At that point, you’re effectively turning the company into a liquidating asset rather than an operating one. The stock price isn’t just a reflection of current earnings, it’s the market’s estimate of the company’s future cash flows over a given time horizon, discounted back to today. If a company stops reinvesting, those expected future cash flows flatten or decline, and the valuation typically compresses toward a low-growth or bond-like multiple, at which point investors would likely prefer safer fixed income like Treasuries. You also lose the ability to attract talent and sustain competitiveness, which usually leads to stagnation and eventual displacement by more innovative players.

• 
  The Assembly (@InTheAssembly) reported 24 minutes ago

  NEW TRADE ALERT: Michael Burry opened a full position in MercadoLibre at $1634. MELI is the Amazon and the PayPal of Latin America combined. 49% revenue growth last quarter. 28 consecutive quarters of 30%+ growth, a record unmatched by any public company in history. Yet the stock is down 36% from its June 2025 high because of short-term margin pressure and competition from Shopee. Translation: the market is punishing MELI for spending aggressively to dominate a 650 million person market. Burry doesn't chase momentum. He buys companies the market has temporarily mispriced. That is exactly what MELI looks like at $1,631. Median analyst price target is $2,500, implying over 50% upside. This is not a quick trade though, it’s a 3 to 5 year compounder. Every time he makes a move, we will break it down here. Turn on notifications so you don’t miss the alert, this is VERY important. If you don’t follow us, you might regret it.

• 
  DividendBoomer (@BoomerDivvies) reported 26 minutes ago

  @ToddsTradeEasy @SparkingFIRENC Your comment “took everything and gave nothing back.” Is what I took issue with. Boomers built a large part of current society and invented many of the things you use today like PCs, iPhones, the internet, Amazon etc.

• 
  Joan Johnson Porter (@joan_porte68439) reported 29 minutes ago

  @AmazonHelp I've been trying to get info for this issue and I've been having strange difficulty. Amazon agent refused then canceled chat now difficulty with my password after having it verified.

• 
  Terrence Hood (@terancehood) reported 29 minutes ago

  @TheWhiteWitchTM @DougWahl1 Because these are two different things. Voting and buying a gun are astronomically different issues. With the gun, it's the same as a toaster or an XBOX controller ordered from amazon, albeit with more red tape. It's just something you bought. (cont)