Amazon status: access issues and outage reports

Problems in the last 24 hours

The graph below depicts the number of Amazon reports received over the last 24 hours by time of day. When the number of reports exceeds the baseline, represented by the red line, an outage is determined.

Most Reported Problems

The following are the most recent problems reported by Amazon users through our website.

• Website Down (46%)
• Errors (28%)
• Sign in (26%)

Live Outage Map

The most recent Amazon outage reports came from the following cities:

City	Problem Type	Report Time
Warwick	Website Down	52 minutes ago
Seattle	Errors	2 hours ago
Paris	Sign in	13 hours ago
Paris	Sign in	18 hours ago
Pontault-Combault	Sign in	1 day ago
Paris	Sign in	1 day ago

Full 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:

• 
  Sabareeswaran Elangovan (@tnexplorer) reported 4 minutes ago

  5 of my recent orders from amazon not delivered, no delivery man called, yet they MARKED AS DELIVERED. When I went to Amazon office, they told we got lot of work bro, so we marked as delivered but we thought to give you tomorrow. @amazonIN improve delivery man count or strictly monitor and fix these kind of issues, I don't even received the delivery person contact number when it is out for delivery. When we call amazon customer care, they will refund it, that's what going to happen, but think if it is an offer price we bought and then again if we re order with the normal price? Does anyone faced the same?

• 
  Xen (@XenBH) reported 4 minutes ago

  A few years ago, I went trekking through the Amazon with my brother. I won’t tell you which brother. We were walking single file through the jungle, me in front, when I heard a terrible scream from behind me. I thought my brother had been taken by a jaguar or something. I spun round thinking something really bad had happened. Turns out a huge black wasp had stung my brother right in the eye. So then my brother throws a hand to his face. Now he can’t see. Blind, and in a panic, he staggers forward, straight into a low branch, which catches him right across the forehead and sends him flying off the trail, down a bank, and into a swampy ditch. It was one of the funniest things I had ever seen. It was just so physically funny. The sort of basic slapstick comedy that’s funny no matter how old you are. To be fair to my brother, it was painful. His face swelled up badly, which made it funnier. I couldn’t help him because I was too busy laughing. Then that evening, we’re back at camp by the fire, eating dinner. Across from us, our Peruvian guide is telling the other guides about his day. He’s speaking Spanish, and he assumes that the two English tourists haven’t understood a word. And so, with no idea he has an extra audience, this chap stands up and re-enacts the whole thing. He clutches his eye and runs in little circles. He walks into an imaginary branch, falls to the floor, and writhes around in the dirt. His friends are all properly belly laughing. I sometimes think about this and still laugh out loud. I really enjoyed writing this post.

• 
  Tropical Value (@tropicalvalue) reported 9 minutes ago

  Last week Amazon repriced GPU compute +20%. This week the rest of the infrastructure stack caught up. 8 signals that changed how I think about AI infrastructure: 1. GPU availability is rising — from the chips everyone thought would be FULLY depreciated by now 3Fourteen's Fast Index hit 23.4%, up from a ~5% trough. The entire move came from legacy $NVDA GPUs (A100/H100/GH200). B200s remain effectively unavailable. Scarcity is bifurcating by generation. 2. Two neocloud entrants this week $META: Morgan Stanley confirmed bare-metal 1P spare capacity only, cannot resell 3P-leased silicon. The $12B Vera Rubin order stays internal — $NBIS gets the advanced inference hardware, META runs its public cloud on older chips. SoftBank is an entirely different story. SB Neo is building full-stack from scratch, targeting 10 GW (2x the $CRWV entire 2030 target) with balance sheet financing across chips, servers, data centers, and power. META's entry is an EPS bridge. SoftBank's is the supply event to model. 3. $NVDA formalized its vendor-financing model Why sell the hardware, and not collect the rent? Revenue sharing + credit support for AI-native clouds is now an official business line. The Lucent parallel (NVDA direct investments at ~67% of LTM revenue vs. Lucent's 24% loan guarantees before collapse) is the risk to monitor. Upside is usage-linked, not spot-booked. 4. SK Hynix removed long-term contract price caps Spot market surges can now fully pass through to LTA pricing. Bernstein: 90.9% DRAM gross margin Q2'26. When $MU and Samsung follow, the silicon input cost floor for AI inference is permanently repriced. Simultaneously with the GPU layer $AMZN just moved. 5. Frontier tokens are getting more expensive. Open-source tokens are collapsing OpenRouter composite: flat at $0.71/M. Disaggregated: OpenAI + Anthropic + Gemini up ~60% YTD to $1.62/M. Everything else down ~50% to $0.18/M. "AI gets cheaper" is accurate for the commodity tier. It's inverted for the frontier tier. 6. Karp and Nadella converged on the same enterprise thesis — and Alibaba just provided the case study Enterprises want to own the means of production — data, weights, feedback loops — not rent intelligence from labs that may absorb their IP. Karp at CNBC: enterprises are "livid" paying for tokens while their IP transfers to the model. This week: Alibaba banned Claude Code after Anthropic's hidden tracking code was discovered — capable of detecting whether the user is in China and affiliated with a Chinese lab. That's the risk. In parallel, $MSFT acknowledges Palantir business model, and is putting $2.5B and 6,000 embedded engineers behind the "learning loop as moat" thesis. $PLTR has been selling sovereign AI for 10 years. Everyone now sees the need to deploy FDEs. 7. $NET launched payment rails for the agentic web The Monetization Gateway + x402 protocol enables sub-cent stablecoin payments for any HTTP resource — no signup, no API key, no prior relationship. Agents pay per call, per outcome. The web ran on attention-based monetization for 30 years. Agents don't look at ads. 8. China trained a frontier model entirely on domestic hardware Meituan's LongCat-2.0 (1.6T param MoE, trained on 50,000 domestic accelerator cards) scored 77.3 on SWE-bench multilingual, nearly matching Claude Opus 4.6 (77.8). The "export controls permanently cap China's AI frontier" thesis now has a live counterexample. → Every AI infrastructure layer repriced this week: silicon, GPUs, application ownership, agent monetization. "AI commoditizes uniformly" is the wrong consensus. The frontier layer is inflecting higher; the commodity tier is deflationary.

• 
  Timmy_Turnes (@Timmy_Turnes) reported 10 minutes ago

  Prediction markets don't have an intelligence agency Until now 2026: @Polysights launches to solve the biggest problem in prediction markets > billions flow through prediction markets > whales move odds before headlines > anonymous wallets shape global narratives > build Polysights > an institutional AI layer for prediction markets > designed to see what everyone else misses > cluster anonymous wallets > detect insider trading patterns > track whale movements in real time > generate AI trading signals > analyze liquidity across every major market > surface hidden onchain behavior > promise institutional-grade intelligence > transform raw betting data into actionable insights > become the "Arkham" of prediction markets > raise a $1.5M pre-seed > backed by YZi Labs + Maven 11 > receive grants from Amazon + Polymarket > early access opens > funds and high-volume traders get in first > retail watches from the sidelines > goal shifts beyond tracking markets > predict the traders before they move markets > identify smart money wallets > detect manipulation before news breaks > automate trades from onchain anomalies > data becomes the new edge > AI becomes the new analyst > prediction markets enter the intelligence era > if prediction markets keep growing > whoever owns the best data wins > Polysights wants to hold the map

• 
  Cody Allred (@CodyAlanAllred) reported 12 minutes ago

  @Papa_Cherry69 @SilverAnon03 @robtanner84 Verses buying them where? Amazon just did the same thing recently with some of their movies. This is the entire issue of purchasing digital. You don't own what you buy

• 
  Justin_Debates (@justin_debates) reported 14 minutes ago

  @iam_biglad1 wtf are you talking about? The average Amazon employee makes $90k a year, putting them in the second highest quintile in America. The entry level wage is $23 per hour. In Walmart the entry wage is $15/hr and the average compensation company wide is $123,000 per year. The problem isn’t their compensation package. It’s grown *** adults making terrible life decisions leading them to be cashiers at the age of 50. Their lack of skill isn’t my problem, and you stealing my money through taxes to support them isn’t your right.

• 
  Caiden 🃏 (@Caiden_Legit) reported 15 minutes ago

  This is why Sony is going all digital. They don’t want you buying games at GameStop. They don’t want you buying games off FB marketplace. They don’t want you borrowing games from your friend. They don’t want you on Amazon, or purchasing from Walmart or Target. They want to corner the market and have you buy their games solely from their store, where they can dictate the prices and take all the profits away from 2nd hand sellers. This is the biggest issue I have with Sony going all digital, there’s no more competition… they are creating a monopoly. I’m a PC player, I get all my games on Steam and GoG… so this doesn’t really matter to me so much as it does the PlayStation community. I personally don’t see this ending well for console gaming. What’s stopping you from switching to PC now? “Exclusives?” Learn how to emulate and you can play any game you want from N64 to PS1 to GameCube to Xbox 360 to Vita… anything. Come to PC and you have actual competition from digital storefronts to buy games. Come to PC and you can mod your games and tailor your experience in so many different ways. It’s going all digital anyway, so why not just join the Master Race like we’ve been saying for years?

• 
  Paul (@harrythemukste1) reported 16 minutes ago

  @AmazonHelp they only happen when I get something delivered by Amazon own drivers, delivered to home with RM no issues. Your drivers are the problem.

• 
  Anil (@OrkAk18) reported 16 minutes ago

  @encoword Same problem, Amazon delivery available, scamkart nit

• 
  The Scrub (@thePatriotScrub) reported 18 minutes ago

  @realrobluna Covid shut down alot of local businesses and small ownership that would be a benefit and show how working is done and how to build small local networks. Now the choice is amazon warehouse or a Walmart full of coworkers that dont speak English or Spanish

• 
  Abu Abel Nation (@NationAbu) reported 23 minutes ago

  Like play The Cartel was ready to Burn down an Entire Hotel just to Kill 1 Man. This movie on Amazon Prime is Crazy. 🥵

• 
  yurshev (@yurshevv) reported 26 minutes ago

  AI is creating millionaires. But it might also be creating the biggest bubble since the dot-com era. Over $1 trillion will be poured into AI infrastructure this year alone. Everyone is racing to build AI agents. Launch AI startups. Raise funding. Automate businesses. Yet almost nobody is asking one question: Who is actually paying for all of this? Big Tech is spending hundreds of billions on data centers. Investors are throwing money at AI companies faster than ever. Valuations keep climbing. But what happens if the returns don't arrive as quickly as everyone expects? We've seen this movie before. In 1999, everyone knew the internet would change the world. They were right. They were just wrong about the price. The Nasdaq crashed nearly 80%. Thousands of companies disappeared. Amazon didn't. That's why the real question today isn't: "Will AI change everything?" It already is. The question is: Which AI companies survive when the hype disappears? Some businesses are building real competitive advantages. Others are just expensive wrappers waiting to be exposed. The difference will determine who builds the next Amazon... This article breaks down the numbers, the hidden risks, and the signals that almost nobody is watching. Worth reading before placing your next bet on AI.

• 
  Culture War Notes (@culturewarnotes) reported 29 minutes ago

  There's another problem hidden in this essay that isn't obvious. First, the essay correctly notes that Palintir is right on this point: "the architecture that maximally preserves sovereignty is one that enables institutions to own their tribal knowledge, and to compound it as alpha," We have a template for this. Third-party sellers who have popular products on the Amazon platform have famously discovered Amazon poaching their products, undercutting their prices, and selling knock-offs as part of the Amazon Basics line. Amazon could do this because it demanded the product spec be supplied before it would deign to sell the product. It then used the supplied product spec to create a knock-off and harvest the profits. This is exactly what every AI software can/will/does do. Any company using a third-party AI quickly discovers the AI can now imitate/replicate their business process and there's not a damned thing the company can do to stop it. To avoid this, in-house AI is a necessity because you MUST protect your proprietary training data. The software itself is fungible, but the training data and associated business processes aren't. A lot of companies are going to go with open-source and build their own in-house AI solutions. But, as anyone who tried to implement a Linux open-source solution for their business desktop environment has discovered, sometimes in-house open-source is an extremely expensive solution. Now you are forcing your business not only to be an expert in building product X, your business must ALSO be an expert in AI software engineering. If it isn't, then your company cannot replicate any advances made in AI software. You fall behind. There's a reason companies use Google docs and AWS instead of writing their own word processing software or building their own cloud solution. Software engineering is not in their wheelhouse. Closed-source AI companies that want to sell themselves as a universal solution have to do what AWS and Google Docs did - they have to figure out a way to wall off proprietary business processes and data within the AI. Nobody knows how to do that yet. Everybody uses every scrap of data they can scrape, steal, buy, loot in order to train their AI, but this cannot continue. This is a low-trust environment. Because of the data training problem, every AI is currently a ghetto welfare cheat stealing hair and nail supplies from the local CVS, reselling the products on every street corner. To succeed, closed-source AI needs to become a high-trust environment, with every business that is utilizing the third-party AI space strictly walled off from every other. We need data banks, not data NY Central Parks. Whoever figures out how to do that first will be a trillion-dollar AI software company.

• 
  Ashwin Goyal (@Ashwin_Goyal) reported 30 minutes ago

  @tnexplorer @amazonIN These issues with amazon deliveries are increasing everyday. I have complained multiple times but nothing happens

• 
  Harmless Karl (@harmlesskarl) reported 32 minutes ago

  Amazon seller is most miserable seller every! Customer message kicked" I bought a case from you, it is broken, replace it or you will be fxxxxed" Check the order, original date: 2025 05 01. WTF

• 
  The Techno Optimist (@technooptimist1) reported 32 minutes ago

  Jeff Bezos on why data centers in space are inevitable: "All of these physics problems are easily solvable. Heat rejection is not a real problem. It's all these questions are really about cost. And the cost of production of the solar cells, the cost of production of the satellites, and the cost of launch will eventually, the lines will cross, and eventually orbital compute will be a better option than terrestrial compute." — @JeffBezos, Founder of @Amazon & @blueorigin

• 
  KaijuKatostrophe (@ProxyConscious) reported 37 minutes ago

  @Reeddirect @progamer999x @PaulTassi Yep. Most studios do this for failed games. Amazon shut down new world with a higher current player base than this.

• 
  L'Carpetron (@Dookmarriot) reported 45 minutes ago

  @AmazonKindle @Amazon Did you do something to the Android version of the Kindle app? In the last month or so, it keeps logging me out of my account and deleted everything I'd downloaded. I'm using the app on a Chromebook Plus; in settings, storage permissions are locked as "denied" and "Pause app activity if unused" keeps turning back on if I restart my Chromebook. I've tried a Kindle Fire and it's so slow.

• 
  Ramkishan choudhary (@RKKataniya) reported 46 minutes ago

  @AmazonHelp I ordered American eagle men casual pant from Amazon but I have a size issue so I want to return my order . I tried create return with link But I am unable to create return so Please create return. Order number 40560177340182704

• 
  flipnartist (@flipnartist80) reported 47 minutes ago

  @AmazonHelp it would be nice if your customer service reps and email actually helped rectify and identify issues and solutions instead of sending an obvious ai/generic response to your customers/employees. Pathetic.

• 
  Dharma 🌺🕉 (@DharmaCalling) reported 51 minutes ago

  @vicharabhio Problem with the YRF genre is the fact that it would have worked a decade ago but with Netflix Amazon and global access people need a benchmark movie

• 
  Shilpa Mehta (@mshilpa71) reported 54 minutes ago

  @AmazonHelp Issue is not fully escalated, my call was disconnected in between and I was told to wait for another call to escalate for another order but they didn't call back.

• 
  SD (@SD_1257) reported 54 minutes ago

  @DanielTStocks @clownltjokerrt @KonstantinKisin I'm gonna retire early partially because of investing in AMZN and TSLA. Not a fan of Bezos/Musk but they helped me. Use Amazon because of convenience and Tesla increased the number of electric cars used. If you can't seem they aren't 100 a detriement that is your issue.

• 
  Vivek Rathi (@rathivivek10) reported 57 minutes ago

  @amazonIN My issue is still not resolved, is the ops team so reluctant? @AmazonHelp

• 
  Sourav Sarkar (@sourav7274) reported 58 minutes ago

  Our **** server OOM'd (Out Of Memory) at night. Had to SSH in morning and restart manually. Rebuilt it as ASG (Auto Scaling Group) (1/1) + golden AMI (Amazon Machine Image) + user-data. Now if it dies, it just rebuilds itself. I sleep better. Cattle, not pets.

• 
  𝕯𝖊𝖆𝖙𝖍𝕸𝖊𝖙𝖆𝖑2070 🇺🇸🏴‍☠️ (@AntiSaint99) reported 59 minutes ago

  @lmkifiwin I need to know why Apple, Microsoft, Google , Amazon and many others don’t have this problem but Sony does.

• 
  Alex (@Abcmsaj) reported 59 minutes ago

  @AmazonHelp I was supposed to get a delivery between 12:45-1:45pm. App told me the driver was 2 stops away. Waited by the door. I watched him reverse down the street with the side-door of the van open and parcels flying everywhere when he slammed brakes on. I still don't have

• 
  HBT ONLY (@HBTonly) reported 60 minutes ago

  @myzccc amazon in 1000% going to pass that cost down to the drivers somehow.

• 
  GUL (@gulVasikova) reported 1 hour ago

  $SPY Many investors will look at this week and conclude that the AI trade is over. I don’t see it that way. I think what we’re witnessing is something much healthier—a market rotation rather than the end of a long-term trend. For nearly two years, money poured into the same group of stocks. AI infrastructure, semiconductors, memory and data-center companies became the market’s biggest winners. Eventually, success creates its own problem. When everyone owns the same stocks, there’s simply less new money left to push prices higher. Even great companies can experience sharp pullbacks as investors lock in profits and rotate into sectors that have been left behind. That’s exactly what this week looked like. Memory stocks like Micron and SanDisk sold off sharply. AI infrastructure names such as CoreWeave and Nebius also came under pressure. Even companies with strong long-term fundamentals weren’t spared. Does that mean AI demand is disappearing? I don’t think so. Building AI infrastructure isn’t a one-year investment cycle. It’s more like building the internet in the late 1990s or electrifying cities a century ago. These transformations take years, sometimes decades. The demand for computing power, networking, memory and data centers doesn’t suddenly disappear because investors decide to rotate their portfolios for a few weeks. Instead, money simply looks for the next area of opportunity. This week, that money flowed into financials, regional banks, biotech and communication services. Interestingly, many of these sectors still trade at much lower valuations than the companies that have dominated headlines over the past two years. Think of it like a marathon. The leader doesn’t sprint at full speed for all 42 kilometers. Sometimes the front runner slows down to recover while others catch up. The race continues, but leadership broadens. That’s what healthy bull markets often look like. Another sign that caught my attention wasn’t the weakness in AI stocks. It was the strength beneath the surface. Equal-weight indices continued outperforming the traditional market-cap-weighted S&P 500. That tells me investors aren’t leaving the market altogether. They’re simply finding opportunities outside the handful of mega-cap names that have carried the market for so long. To me, that’s actually constructive. A bull market supported by hundreds of companies is usually healthier than one driven by only a few. Of course, AI stocks could remain volatile. Crowded trades often unwind further than fundamentals justify. That doesn’t necessarily mean the businesses have become weaker. Sometimes expectations simply got ahead of reality. Long-term investors should remember that price and value aren’t always the same thing. Great companies don’t rise in a straight line. Nvidia, Amazon, Apple and Microsoft have all experienced multiple corrections of 20% to 50% during their long-term journeys. Those pullbacks didn’t change their businesses. They simply tested investors’ patience. This week’s rotation reminds me of one important lesson: Markets constantly rotate. Innovation doesn’t. The companies building the infrastructure for AI may experience periods of volatility, but the long-term trend of AI adoption, cloud computing and digital infrastructure continues to move forward. That’s why I think the more important question isn’t “Which sector won this week?” It’s “Which companies will be more valuable five or ten years from now?” For long-term investors, that question matters far more than any single week’s rotation.

• 
  Me Again (@bcbradley58) reported 1 hour ago

  @AmazonHelp I spent over an hour last night going over all of the orders that had problems yesterday. On the order that was delivered to the wrong address I was asked if I wanted a refund or a replacement I chose the replacement. Guess what, it was delivered to the wrong address again. The person that received the package was kind enough to bring it to me. He said his wife told the driver it was the wrong address, but the driver did not speak English and walked off and left it anyway. Enough is enough. I also had another order yesterday that was sent back to the shipper because that person apparently can't read a map either. I was told they would try again, but so far there has been no update on them trying again today. I have been a customer for many years and this problem has just started and frankly it is about to be the end of my long relationship with Amazon.