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 (31%)
• Sign in (23%)

Live Outage Map

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

City	Problem Type	Report Time
Charlotte	Website Down	2 hours ago
Panama City Beach	Errors	7 hours ago
Paris	Website Down	11 hours ago
Belfort	Website Down	16 hours ago
Bordeaux	Website Down	17 hours ago
Phelan	Website Down	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:

• 
  Amazon Help (@AmazonHelp) reported 7 minutes ago

  @Gusto_muffin @amazonmusic @Gusto_muffin Apologies for the inconvenience that you're facing an issue with the Amazon Music. Request you to perform a basic troubleshooting if app is displaying any error. Here are the steps: 1. Try clearing the cache and cookies. 2. Try using any other web browser, 3. Try clearing the history for Amazon Music. 4. Close and re-open the browser. If the issue continues to persist, please do not hesitate to contact us. We would be pleased to assist you further. - Waseem

• 
  Mike (@MikeLongTerm) reported 8 minutes ago

  $AMD is winning shares EVERYWHERE 🧵 @AMD has just punched through a psychological wall that used to belong to Intel alone. For the first time in the modern PC era, Mercury Research has AMD above thirty percent of the x86 client processor market, the chips that power desktops and notebooks. Second quarter unit share hit 30.3 percent against Intel’s 69.7 percent. A year earlier AMD sat at 23.9 percent. Two years earlier it was 21.1 percent. That is not a lucky quarter. That is a multi year grind that has now produced a structural result. Desktop is where the story gets even better for AMD. Share rose to 34.9 percent even while the desktop market itself contracted more than twenty percent year over year. Memory and storage prices exploded because AI data centers are vacuuming up DRAM wafers for high bandwidth memory. DIY builders and OEMs delayed purchases. Both companies shipped fewer desktop chips. AMD’s shipments held up better, which is exactly what a company with the better product does in a nasty market. You do not need a booming PC cycle to take share when buyers who still spend money choose Ryzen, especially the high end X3D parts that dominate enthusiast builds. Notebooks completed the sweep. Mobile x86 share reached 28.9 percent, up from 20.6 percent a year ago. Combined client share crossed thirty percent because AMD is now winning volume in laptops as well as prestige in desktops. These numbers are shipments into the channel, not the installed base, and they exclude Arm. That caveat actually helps the bull case. AMD is taking Intel’s core franchise while Apple and Qualcomm nibble at the edges. Intel is getting squeezed from both sides. Client PCs are not even the prize. They are the proof. The money is in the data center, and AMD is turning that market into its growth engine. Data center revenue more than doubled year over year to $6.7 billion in the second quarter and made up 58 percent of a company record $11.5 billion. Server unit share on Mercury’s broad x86 measure reached 34.5 percent, up from 27.3 percent a year earlier. Narrow it to EPYC versus Xeon in the real server socket and AMD is already near 46.4 percent. That is not a challenger anymore. That is a co equal supplier in the most important compute market on earth. Dr. Lisa Su’s guidance makes the next twelve to eighteen months look even more lopsided. She expects data center revenue to much more than double again in 2027 even after a massive stepup in 2026. Server CPU revenue is guided to grow more than eighty percent year over year in the second half of 2026 and more than seventy percent in 2027 off a much higher base. Hyperscalers including Amazon, Microsoft, and Google keep expanding EPYC across internal infrastructure and public cloud. Agentic AI is not just a GPU story. It raises the number of CPUs sitting next to accelerators, and AMD is winning those sockets with better density, better performance per watt, and a generation cadence customers now trust. Unit share also understates AMD’s power. The company sells more of the expensive chips. X3D processors own the gaming high end. EPYC owns a rising share of the profitable server stack. Client revenue share had already cleared thirty percent earlier in the year. When mix is this rich, every point of unit share is worth more dollars than the raw percentage implies. Intel still has the larger installed base because switching is slow, expensive, and operationally painful. That lock in is real. It is also a decaying asset. Every quarter of better silicon, better cloud availability, and better OEM design wins makes the next switch cheaper than the last one. Intel just posted strong revenue growth, which shows the AI capex wave is huge enough to lift both companies. It does not change who is taking the incremental socket. Thirty percent is not the destination. It is the point where the market stops treating AMD as the alternative and starts treating it as the default source that often becomes the first. Across desktops, notebooks, and servers the arrow points the same way. AMD keeps winning share, mix, and the data center dollars that actually compound. Momentum is not split. It is sitting with AMD. Not Financial Advice! DYOR!

• 
  Stuck (@Stuckey_1987) reported 13 minutes ago

  @AmazonHelp @amazon No offense “Jack”, but that’s a big part of the issue. Yall **** up, and now I’ve got to take time away from a dying family member to deal with y’all’s **** up. Because all of this could have been avoided by understanding paperback books don’t ship well, so they need a decent box

• 
  Apartheid Elon ♥️s Apartheid Israel & Shadow Bans (@mjones206) reported 18 minutes ago

  Way too many @amazon drivers are so lazy and entitled that they'd rather just dump people's packages on the lobby floor than walk 20 feet further to the mailroom, while completely ignoring customer instructions because it's not their problem. It did not used to be this way

• 
  _ (@dgwilliams3d) reported 18 minutes ago

  Not surprised. @PrimeVideo / @Amazon is terrible. I've been going out of my way not to give them another red cent after the announcement of the cancelation. #savestargate

• 
  Abhisek Sinha (@abhisek_4u) reported 20 minutes ago

  @AmazonHelp Unfortunately they didn't provide any support!the link they have shared of another Toy.Upon installation found charger not working-when contact-asked short video,after sending through WA they silent,now shall I return Toyzone product or wait for Charger 2b replaced?V V BAD EXP

• 
  Atul Yadav (@AtulXac) reported 25 minutes ago

  @AmazonHelp I have followed up with your specialist team sir, there is no tracking system where i can see the status of my issue I haven’t got any complaints number its just I’m reporting everywhere.

• 
  Mild Mannered Maniac (@80sGeek) reported 27 minutes ago

  @tekbog You do not spend budget and resources on problems you don’t have. Just because Amazon and Google are pushing it doesn’t make it a good idea, be careful. You are spending a huge fraction of your budget and resources on something that adds zero value to your customers

• 
  Esq (@EsquireDePalma) reported 28 minutes ago

  @thatthingido2 @amazon sucks! I had the same problems repeatedly! Dropped Prime and quit using Amazon. Now use Walmart as their delivery and app is great.

• 
  Mark Leonard (@Leo76Mark) reported 29 minutes ago

  @PatVPeters Order some bed sheets from Amazon, problem solved

• 
  Science You Fools (@Scienceyoufools) reported 30 minutes ago

  Some news predictions for the next few months. Treat them as the ramblings of an old fart. • September — UNGA 81 Climate change blamed as the driver of the migration crisis, used to reignite the Digital ID push under a “trusted traveller” story. • Late September to early October A distraction when Channel and Med boat numbers come out: a tragedy, overcrowding, or a fire that eats the headlines. • Early October A document dump or inquiry leak. Most likely more Epstein material, or in the UK the Mandelson files the Met has been sitting on. • October — Brazil election First round 4 October; almost certain runoff on the 25th. Riots, Amazon mining, Chinese investment, and “democracy vs Bolsonarismo” copy. Used later as scaffolding for carbon taxes in 2027. Firms will talk about replanting the Amazon while grabbing land — the same pattern as the South American wildfires. • Early November — Dakar Youth Olympics Heatwave, heat illness, crowd-control problems. Climate narrative gets another outing. • Early November — US midterms Late counts in AZ, GA, PA, WI, MI. House lost. Within 48 hours a surveillance or spending bill nobody campaigned on, then a Fed/markets story the next morning to change the subject. • December UK, US and Europe: grid stress, vandalism, or security incidents. • Mid-January — Davos WEF annual meeting. Push for CBDCs so climate finance can be tokenised. If the US stays away, that absence becomes the headline and covers what they are actually selling.

• 
  Hafeez Jimoh (@jimohhafeezco) reported 32 minutes ago

  Shopify started as an online snowboard shop. Hugging Face started in 2016 as a mobile chatbot app targeted at teenagers YouTube started as a video-dating website. Facebook started as online directory for Harvard University students. Amazon started as a small online bookstore Netflix started as a mail-order DVD rental service Discord came from a failed mobile game, Fates Forever. It looks like the key thing is to wear the building cap and start something. There is a higher chance it will fail at first, you could pivot later and do something else. No need to over-optimize trying to build the best or get it right first. Very few founders get it right the first time. Almost none of these came from someone sitting down and reasoning their way to the most fantastic idea. They came from building something, shipping it, and paying attention to which piece people actually used.

• 
  The Sanya of Lagos (@SanyaofLagos) reported 33 minutes ago

  @pepple_miracle I have issue with Amazon PSP, @RaenestApp , @greyfinance dollar accounts with Lead Bank are not working

• 
  Aydın Nasuh (@aydinasuh) reported 35 minutes ago

  @ZidaneZ08902030 @ShubHQcom Honest answer: my scoring tags sentiment per mention, not the reason behind it, so I cannot split those two from the data as it stands. It is the obvious next thing to add though. What the numbers do say: of Stripe's 130 mentions, 84 are neutral and 46 negative. So it is not an outrage pattern, it is mostly procedural. And the sample skews to Shopify, Amazon and Etsy sellers rather than developers, so most of these people never touch the API at all. The problem themes that do get counted separately are account suspension and returns and chargebacks, which points more toward holds and disputes than anything breaking. But that is inference, not measurement. Let me go read the actual mentions and come back with what they are really complaining about.

• 
  Dulal Das (@dulaldas1973) reported 37 minutes ago

  @AmazonHelp "It is courier delay on Amazon's side, not mine. The item was picked up on July 20, so please issue an instant manual refund."

• 
  HazyHawg (@HazyHawg) reported 38 minutes ago

  @AshleyPetty_ @Apple The Weather Channel app is decent. Apple weather is terrible and Amazon Alexa weather forecast is even worse.

• 
  Percgremlin (@Percgremli6261) reported 40 minutes ago

  @SteveWeigel @GOP__Ls i just know from when i was down really bad that the numbers i was provided just led in endless loops and in the end no one did anything for me. i had to live in my car then a crackhouse and work amazon 66 hours a week.

• 
  Dom🏳️‍⚧️🐻/🦋⚙️ (@Frot4MeDaddy) reported 42 minutes ago

  today i trimmed my bush and the process was so tedious and annoying it was like trying to cut down the amazon rainforest with a lawnmower

• 
  Moshood Digitals (@Moshood_Keye) reported 43 minutes ago

  @FiscalLogiqlhe @coachnathel And you didn't read the part where he said there's a friend of his that chose Amazon KDP course who's also Affiliate Marketer? Cause he bought a course and decide to promote the same course is what I don't get to be the issue....

• 
  𝐍𝐨𝐭𝐨𝐫𝐢𝐨𝐮𝐬 𝐃.𝐀𝐕™ (@Zen232) reported 43 minutes ago

  @xraygirl9 Those bricks though. So you're telling me **** showed up broken back then too? It's not just Amazon deliveries?!

• 
  Naveen🌍 #bds (@nav391) reported 45 minutes ago

  @adgirlMM Animal ag is burning down the Amazon rainforest and hardly anyone cares to stop eating animals

• 
  JioCare (@JioCare) reported 45 minutes ago

  @shishirkum37038 Hi Shishir, we are here to assist you. Amazon Prime subscription has been provided on your number. To access the same, please login in to MyJio & click on Activation banner for Amazon which appears on the homepage - Harshit

• 
  ozzy ✦ (@jimmyapologist) reported 46 minutes ago

  wait give me a second i don’t have my hulu login and i also don’t have $4 for amazon prime

• 
  mai (@etnowt) reported 52 minutes ago

  @pleuvoiraverse Thats a whole amazon forest down there

• 
  🌟Illuminannie 🪷 (@NannieWulforst) reported 54 minutes ago

  @Mel_Ankoly @Eviljohna I'm down to Amazon Prime and Paramount that comes with my Walmart membership

• 
  LB. (@lea_banks95) reported 55 minutes ago

  @sydneyapologist @mariaria14 @luna_moon_emoji Yeah so explain Siri popping up even when no one said anything remotely close to hey siri. Didn’t Amazon or Google just get in trouble for this exact thing?

• 
  CaptainAmericaTex (@CaptAmericaTx) reported 59 minutes ago

  @TiffanyxFong hey @grok If a gentleman is in a committed relationship with Apple Siri, but occasionally he interacts with Amazon Alexa on the down low, is that considered cheating? (asking for a friend)

• 
  Corey J. Gallon (@CoreyGallon) reported 60 minutes ago

  Prompts are suggestions, not constraints. A model reads a rule as text, not as logic it has to execute. That distinction runs through ". @lizfue's talk "Stop AI Agent Hallucinations: 5 Techniques + Production Patterns", on @aiDotEngineer's YouTube. Elizabeth is a Developer Advocate at AWS, and every technique she covers is a code change rather than a prompt change, demoed side by side on a travel agent built with Strands Agents. - Semantic tool selection. 29 tool schemas go into context on every call, roughly 3000 tokens before your message even starts. Filter with a vector search over a tool database and the model sees only the top three. - Swap tools between turns. In a conversation, tools accumulate in history. Strands lets you clear the tool registry inside the agent loop and add the new ones each invocation. - GraphRAG for questions RAG can't answer. "What is the average guest rating across all hotels in Paris?" gets estimated from three retrieved chunks. A Cypher query against Neo4j computes it across the whole dataset, and returns zero for hotels in Antarctica instead of hedging. - Multi-agent validation. A single agent that gets a tool error will rationalize it into a confident success. Strands' Swarm chains an executor, a validator, and a critic, so the user sees the failure. - Neurosymbolic guardrails. Put the rule in Python, not the prompt. A Strands hook on BeforeToolCallEvent cancels the call when parameters fail, so "max 10 guests per booking" actually holds. - Runtime steering for soft rules. Blocking leaves the user retrying. With agentcontrol, a 50 guest booking gets steered into a split across rooms and the task completes. Rules live on a server and update without redeploying the agent. - Then production. Amazon Bedrock AgentCore gives you gateway-managed tool selection, policies enforcing rules before tools run, steering rules in DynamoDB, memory, and observability. I'm working through the published talks from AI Engineer World's Fair sharing summaries and takeaways. Follow for more!

• 
  Motadole (@Motadole1) reported 1 hour ago

  @AmazonKSA @AmazonHelp @amazon FYI Your service trending down 👎😔

• 
  jAy (@rawdoggnbro) reported 1 hour ago

  @BCarey1969 @ojessie__ This isn’t rocket science, and an easy fix. Theres no meed to involve the super go order graffiti remover on amazon and it should remove it. If not then simply paint over it.