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 (25%)

Live Outage Map

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

City	Problem Type	Report Time
Seattle	Sign in	5 hours ago
Rheine	Errors	10 hours ago
Poplar	Website Down	16 hours ago
Valréas	Errors	18 hours ago
Chartres	Errors	19 hours ago
Valencia	Sign in	20 hours 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:

• 
  friend of Quackducker (@FQuackduck70825) reported 5 minutes ago

  @DollyParton I didn't have time to celebrate, I worked 9 to 5, then I put on my bathing suit of many colors, told the dog I will always love him, fed Jolene my cat, the amazon guy came again, and the car I live in, my blueridge mountain home, broke down

• 
  Steve Cowlishaw (@SteveCowlishaw) reported 9 minutes ago

  @jazzharris_ @ayeejuju I literally just looked on Amazon and saw about 5 before even scrolling down

• 
  bob yau (@bobyau) reported 10 minutes ago

  I’m wrestling with integrating Amazon Chronos2 into my TimesFM prediction pipeline—a move intended to improve signal quality, naturally. The whole thing feels like I'm trying to jam a Tesla engine into a Model T. My current hypothesis is that the Chronos2 models provide a faster, more granular view of the underlying market than TimesFM's data. I’m not entirely convinced yet. Here’s where I’m at: First, I’m deep in `run_timesfm.py` – which, in my opinion, is a monument to functional complexity. I suspect a lot of that complexity *is* necessary, but untangling it to inject Chronos2’s predictions is…a challenge. It’s all about understanding how it talks to Kalshi, specifically. The data formats and timing are key. I've identified roughly 12 places where I need to insert checks for new data sources. Second, the Chronos2 directory itself is…sparse. It’s a collection of files, a couple of example scripts (which are aggressively unhelpful), and not a lot of explanation. I’m assuming I need to build an adapter layer to translate Chronos2’s output into TimesFM's expected format. My initial thought is to use a lightweight inference server – something like FastAPIs – to wrap the Chronos2 predictions and expose them as a clean JSON endpoint. That's *if* I can decipher the model format first. I worry I’m going to spend days just reverse-engineering it.

• 
  Generic Mainer 🇺🇸 (@genericmainer) reported 11 minutes ago

  @financedystop An Amazon Alexa and a smart thermostat would solve the problem, that’s what we did for our grandma when she lost her sight.

• 
  BENJO (@BenjiHayes10) reported 14 minutes ago

  With Atlas V retired, Amazon Leo loses a ride option mid-deployment. A constellation this size lives or dies on launch cadence, so Amazon spreads its manifest across Vulcan, Ariane 6, and others. Building the satellites is only half the problem. Getting them up is the rest.

• 
  weetard (@DIET_RIOT_07) reported 15 minutes ago

  @slipsilv3r This is my problem as a RYO conisseuer being in the city I'm at Midwest towns get this right fym I gotta order tubes off Amazon in one of the biggest & likely most iconic cities in the world

• 
  Gofra (@Gofralo) reported 16 minutes ago

  A man spent 20 to 30 hours a week copy-pasting data into spreadsheets. Every week. For years. He ran two Amazon businesses on the side. To find products to sell, he researched competitors manually. Pulled pricing data. Tracked sales volumes. All of it in Google Sheets. He had no CS degree. He had never written a line of code. Then he found a tool called Cursor. He decided to build the thing he needed. Not a startup. Not a company. Just the tool that would kill the spreadsheets. He built it in 48 hours. Two years earlier he had bought a business coaching program. Inside it: thousands of active Amazon sellers. All with the same spreadsheet problem. He posted the tool there first. Day 30: $10,000 a month. Day 60: $17,000 a month. Day 90: $21,800 a month. He eventually crossed $30,000 a month. He still cannot read a stack trace without AI help. And here is the part worth sitting with. The story is not about coding. He did not learn to code. He will tell you that directly. The story is about what the bottleneck actually was. He had the problem. He had the audience. He had been inside the customer base for two years without knowing it was a customer base. The only thing missing was a way to build. For thirty years, "I cannot code" was a wall. Not a gap. Not a hurdle. A wall. You could have the right idea and the right audience — and if you could not build, you were stuck. Most ideas died there. In my opinion this is the most important shift in how businesses get built in a generation. Not because non-technical people became developers. Because the wall between having an idea and building it stopped being made of code. What would you build if that wall was no longer the reason you couldn't?

• 
  JasminxJaz (@Jasminxjaz) reported 20 minutes ago

  @Coinvo It’s also an accessibility issue which violates a few laws in the EU hence why they are ready to intervene but it would also affect companies like amazon and Microsoft that also have a subscription base model

• 
  Amit Chorera (@achorera) reported 21 minutes ago

  @AmazonHelp @amazonmusic Amazon, it's astonishing you have not read the full tweet and chosen to give lengthy standard reply which is waste of time to read. I already tried subscribing but there is constantly an error, I attached screenshot too, which you chose to ignore. Meanwhile Spotify premium subscribe feature worked smoothly.

• 
  Dinnu daniel (@daniel_adinnu) reported 21 minutes ago

  Meta is spending $145 billion on AI infrastructure this year. It still ran out of memory. So it built a chip to make old memory work like new. The Vistara chip is not a recycling initiative. It is a response to a supply crisis. DDR5 memory, the current standard for high-performance AI servers, has been in shortage since the AI buildout accelerated in 2023. Every major hyperscaler, Meta, Google, Microsoft, and Amazon, is ordering more than the memory supply chain can produce. Lead times for high-bandwidth memory used in AI accelerators have stretched beyond a year in some cases. The companies building the most aggressive AI infrastructure in history are all fighting over the same limited pool of silicon. Meta’s solution is to make two generations of memory work inside the same server simultaneously. The Vistara chip sits between the server’s processor and the memory modules, translating signals between DDR4 and DDR5 so both can operate in the same system. The DDR4 modules, which Meta has in significant quantities from its previous generation data centres, handle cold data, information that is accessed infrequently. The DDR5 modules handle hot data, the high-frequency transactions that AI inference and training demand. The processor sees a unified memory pool. The underlying hardware is running two different standards at two different speeds. The mechanism that makes this possible is a translation layer built into silicon. DDR4 and DDR5 are not compatible at the electrical or signalling level. The voltage requirements differ, the timing protocols differ, and the command sets differ. Building a chip that bridges those differences in real time, without introducing latency that would compromise the performance of the DDR5 side, is a genuinely difficult engineering problem. Meta solved it and is now deploying it at data centre scale. The cost arithmetic is what makes the effort worth it. A single DDR5 server rack requires a significant number of high-capacity memory modules. Each module costs substantially more than its DDR4 equivalent. Meta is sitting on warehouses of DDR4 from servers it has already retired or is in the process of retiring. Vistara converts that stranded inventory into usable capacity without buying new DDR5 modules to replace it. The electronic waste reduction is real but secondary. The primary driver is that memory is the bottleneck and Meta needed more of it faster than the supply chain could deliver. The chip is the answer to a procurement problem dressed in the language of sustainability. When a company with $81 billion in cash invents custom silicon to stretch its memory supply, the shortage it is solving is not a minor inconvenience.

• 
  PRIYANKA GANGULY (@PRIYANKAGA2848) reported 24 minutes ago

  Sold $10,000, but only $8,200 hit the bank. The gap isn't a problem—it's just unreconciled fees, refunds, and reserve holdbacks. Add in multi-channel selling (Shopify + Amazon) with different payout schedules, and it's easy to lose the full picture. Reconcile every payout

• 
  Kathy Grimes (@KathiG2033) reported 24 minutes ago

  So many companies had flipping stole or let get stolen your personal information? This was today Have Amazon Prime /sign in through IPad/with finger print/ then sends te t to phone/plug in 6 digit number to account/then fingerprint ID again/ now a password WTF!

• 
  Suzi Hixon Bledsoe 🇻🇦⚖️ (@suzihixon) reported 26 minutes ago

  @theallinpod I’m not well versed in the ai/software/ dev space, but Amazon did something similar with e-commerce. Basically stole third party sellers data in exchange for using the platform. Saw this with far too many of my clients. Essentially ground them down with fees then gutted them.

• 
  JESSI RIHANNA (@Jessi_Rihanna) reported 36 minutes ago

  I’m so sad :( I accidentally let out some anger and pushed a bunch of people down the stairs. Now, here comes the insecurity and self blame. I’m choosing self care instead, first with an @amazon haul! Then, I’ll probably go paint balling with my libertarian girl group

• 
  Avy.Sys (@Avy_Sys) reported 36 minutes ago

  You launch your app in the US 🇺🇸. Subscriptions are growing, revenue is climbing, and everything seems to be going according to plan. Then you receive a notice from the FTC. 📩 Sounds a bit dramatic, but for many startups it’s a real business risk. Just recently, the FTC made headlines in Ukraine after opening a case involving one of the products developed within the Genesis ecosystem. Before that, companies like Amazon, Adobe and Epic Games had all faced FTC action over issues ranging from subscription practices to consumer protection. So why does everyone pay so much attention to the FTC? 👇 The Federal Trade Commission is the US agency responsible for protecting consumers and ensuring fair competition. In practice, it looks at things like: • misleading advertising; • subscriptions that are difficult to cancel; • hidden fees and dark patterns; • fake reviews and deceptive claims; • privacy and data handling. And here’s what many founders don’t realize: You don’t need to be a US company to attract the FTC’s attention. If you’re selling to US customers, your business can fall under its scrutiny. That’s why entering the US market isn’t just about incorporating a company. It’s about making sure your product, marketing and customer journey comply with US consumer protection rules before they become a problem. Planning to enter the US market? We’ll help you understand the legal requirements, and assess potential FTC risks. DM us for a consultation 🤝 #ftc #federaltradecommission #businessinusa

• 
  Vibha Agarwala (@VibhaAgarwala) reported 37 minutes ago

  @AmazonHelp Went to the link. Waste of time. Made to wait by 2 reps while "they went through my issue", received the standard reply. WE NEED 4 DAYS TO CONDUCT AN INVESTIGATION! THIS IS HOW AMAZON FOOLS PEOPLE....ROUND AND ROUND WITH NO RESOLUTION! I got same response 15 days ago! Scamsters!

• 
  AnechoicMedia (@AnechoicMedia_) reported 38 minutes ago

  @Nemtastic1 It's a persistent problem for the left that the general public mostly likes Amazon and their big chain grocery store, so there's no mass support for nationalization.

• 
  Kevin McBride (@CommentsWithKev) reported 38 minutes ago

  @SallyIsADog Amazon sponsored the video from what I’ve been told. We were deep in the heart of the distribution center district and they wanted to show off the available plot right next to them they’re hoping to sell. She may be 6 and a terrible videographer, but she’s got a helluva mind when it comes to business strategy or whateva🫡.

• 
  Snowcatt🇬🇧🏴󠁧󠁢󠁥󠁮󠁧󠁿 (@Snowcatt) reported 40 minutes ago

  @SBarrettBar @LucyTCWife I just bought the kindle version from Amazon - no problem. Have done some skim reading about the death of her son. Horrific. Will need to go va k and read properly before tackling the tweet era section

• 
  डॉ श्रृष्टि शर्मा (@Shishtiii) reported 41 minutes ago

  @AmazonHelp @amazonIN I ordered projector (171-2976541-9839509) from past 5 days mental harassment going on projector from you guys , and when it came it was not working , delivered me faulty product now telling me to go to service centre . I have not paid for this torture

• 
  Mr.Midnight (@_Black_Sinatra_) reported 41 minutes ago

  @SuperSisi Well.. GameStop been closing down slowly since last year they were losing the battle to digital downloads and people shopping for games online like amazon and etc.. they stopped selling collectibles for preorder games after the pandemic.

• 
  GumzBunz (@ItzGumerzBunz) reported 41 minutes ago

  @JustBellaX3 it wouldnt be a problem if they can just get rid of the game out of your library anytime because of licensing, and pysical just better because of how better the download is better then digital anyways also just bs because u can get pysical cheaper then digital on amazon anyways-

• 
  Chalu Lombdi (@TheDarkstar1122) reported 42 minutes ago

  @AmazonHelp Amazon help but provides no help ( ͡° ͜ʖ ͡°) More like Amazon Copy Paste Organisation has Billions of dollars but employees are using notepad to "resolve" issues. ChatGPT could replace this department and do a better job

• 
  Vibha Agarwala (@VibhaAgarwala) reported 43 minutes ago

  @AmazonHelp RIDICULOUS!! Why should I call AGAIN AND AGAIN? You only waste my time. STOP PRETENDING ON SOCIAL MEDIA THAT YOU ARE ACTUALLY DOING SOMETHING!! The media is full of people who are facing the same issue. If we all get together, you will feel the force of the Indian consumer!

• 
  Sushiko Byte Vtuber 🍣 🔪 🧋 (@SushikoByte) reported 49 minutes ago

  My day started with me making Chicken Alfredo Stuffed Shells, with Chicken and Broccoli Alfredo Pasta as well (my one kid doesn't eat broccoli anymore Q_Q). Go to sit down, and husband told me he only had one case of soda. I was confused cause he ordered more. I was like what I know we ordered more, did you check? He checked the orders and commented that he couldn't see how many cases were supposed to be in each order. My blind *** missing 95% of my vision goes and looks at the site and sees the numbers almost instantly. I point it out and he realizes Amazon messed up and only sent him one case when there was supposed to be two. ... SIR WHY DOES THE BLIND LADY SEE MORE?!?! But now it's time to stream, tehe

• 
  Dhaval (Investment Books) (@InvestmentBook1) reported 51 minutes ago

  @pointperkspicks @amazon @amazonIN please help. Affiliates least deserve honesty and transparency so issues can be resolved. A direct ban without any notice is very harsh

• 
  Kotemendo (@dynojp) reported 59 minutes ago

  (3/4)Streaming subscript (for example: Netflix and Amazon) and we won’t own the data anymore. There will be a server somewhere and a bottleneck faucet that we need to pay to access. It becomes a per payment system. We will not be able to keep our own in game data anymore,

• 
  JANGE (@2kJ21) reported 59 minutes ago

  @Asiahorse_PC TERRIBLE customer service from this company. Just buy from Amazon

• 
  Macross420 (@adol1701) reported 60 minutes ago

  @YorchTorchGames @gamesyouloved You how many games I passed until prices went down just comeback and get a huge bundles I missed out and now it’s not happening. Still some on my Amazon list I need to fulfill

• 
  Jeff Dzado (@jadzado) reported 1 hour ago

  I used to run emulations of thousands of Kiva drive units, job managers, and dozens of workstations on my ~2013 MacBook. Entire facilities faster than realtime. Making targeted performance improvements. Meanwhile dozens of other engineers were trying to fight against network latency and run half of that in the cloud "because we're Amazon, and we have AWS, and we need big computers 200 msec away to scale!" They struggled to make it work with dozens of people (network latency made it a nonstarter) for 600 drive units on massive distributed compute. I was doing thousands... on my laptop. It was a part time effort for me to make it happen. Many problems in modern software and computer engineering are self inflicted.