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 (48%)
• Errors (28%)
• Sign in (24%)

Live Outage Map

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

City	Problem Type	Report Time
Ashburn	Sign in	3 hours ago
North Las Vegas	Sign in	4 hours ago
Saint-André-de-Corcy	Sign in	10 hours ago
Lyon	Website Down	13 hours ago
Camden	Website Down	22 hours ago
Detroit	Errors	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:

• 
  mrs. bendell werry (@bebookled) reported 5 minutes ago

  @BeckyTheChicken I am modesty forward in a big way, but if I’m in the privacy of my own home in my workouts and the amazon guy shows up out of nowhere, it’s not my problem

• 
  lord pretty flacko ⚔️ (@smdcapital) reported 11 minutes ago

  i’ve decided to order a baseball bat specifically to destroy my HP printer it is a piece of ****. printers should connect to wifi and print when you click a ******* button instead, there are a million little lights & icons with error signs, it forces you to download an app and login to figure out why, only to show you that the ink u bought from amazon isnt authorized for use this is what product design and customer experience looks like when you have a company that outsources work to india and H1B’s therefore, tomorrow morning, i will be taking it to the backyard and smashing it to ******* pieces and shorting HPE on monday

• 
  🔞 Kespe ⚖️🌱🌸🏵️🌿( kesrunsslow.bsky.social ) (@KesRunsSlow) reported 16 minutes ago

  @AmazonHelp aight, might have been an issue involving microsoft edge and i finally discovered that and it has been resolved, but damn, nunna yall had to do all that ugh

• 
  Cash Fargo Jet Life (@TheCashFargo) reported 17 minutes ago

  Amazon make it so difficult to sign in on outside devices I just wanna load Starz

• 
  preeyorders (@preeyorders) reported 19 minutes ago

  @amazonIN @AmazonHelp Solimo brand issue there. Been a regular customer and having a medical emergency at home- not at all helpful. Please help

• 
  Penny Schwartz (@PennySchwartz) reported 20 minutes ago

  @amazon frustrated customer I cannot get to live. I want live chat or a phone call your website and customer service sends me down a spiral of mazes

• 
  George Ponder (@Coppertop004) reported 22 minutes ago

  @AmazonHelp You guys are still not listening. I know how to close the window. I want to know how to stop it from appearing every time I return to the search results. Are you understanding the question? I have no issue with AI but do not like it forced on me.

• 
  Eric Smith (@Eric_Smith08) reported 24 minutes ago

  First why Amazon's defaults are designed to make you spend more. Amazon is not a neutral shopping platform. It is an optimization engine designed to compress the time between "I might want this" and "I own this" to as close to zero as possible. 1-Click ordering: eliminates the cart, the review screen, and the "are you sure?" moment. One tap. Purchased. Shipped. Subscribe & Save default frequency: set to the most frequent delivery schedule. You adjust down manually or you get 6 tubes of toothpaste every month. Alexa voice purchasing: enabled by default. No confirmation PIN required unless you add one. Every default removes friction. Every removed friction removes a decision point. Every removed decision point is a moment where you might have said "actually, I don't need this" and now you can't. The finance coach's framework: "Amazon spent 25 years engineering the world's fastest purchase funnel. 1-Click. Voice ordering. Auto-subscriptions. Predictive shipping. Every default is set to the fastest possible purchase. Your job is to add friction back deliberately. Because friction is where your money is saved."

• 
  ANKUR (@rastogi_ank) reported 27 minutes ago

  @ZeptoNow Your amazon gift cards are not activating as per instructions on cardboard ,as sms is not sending. Please ask pinelabs to resolve this. I am tired of calling them, they are not picking. I didn't purchase gift cards from zepto for this trouble.

• 
  Ann B | Creative Strategist + UGC Creator (@generatedbyann) reported 28 minutes ago

  One of the biggest mistakes I see in paid ads is jumping straight into writing the script. The research is part of the creative. Before I write a single hook, I want to know: • How does the customer describe their problem?
• What have they already tried?
• What made them skeptical?
• What finally convinced them to buy?
• What words do they naturally use instead of the brand's marketing language? That's why I'm spending time in Reddit threads, Trustpilot reviews, App Store reviews, TikTok comments, Amazon reviews, and community forums before I ever open a Google Doc. Most brands already have thousands of free copywriters. They're called customers. They've already told you exactly what frustrates them, what excites them, what objections they have, and what transformation they're actually chasing. Our job as creative strategists isn't to invent messaging. It's to recognize the patterns hiding in plain sight and translate them into creative that feels like it was made specifically for the person watching. That's how ads stop feeling like ads. They start feeling understood. #ugc #ugccreator #ugccommunity

• 
  Devin Goldberg (@Devin079086) reported 28 minutes ago

  @amazon you all need fix your system. It’s tax free weekend in Arkansas and you all are still charging taxes. I spoke to customer service and they weren’t much help. Major problem!!!!

• 
  Ishaan Ansari (@iamihansari) reported 32 minutes ago

  The most successful consumer products rely less on what users say and more on what they do. Spotify learns from what you skip, replay, and save. Instagram learns from what you watch, share, and linger on. Behavior is often more truthful than feedback. The goal isn't to collect more data. It's to collect better signals In April 2017, Netflix officially replaced its iconic five-star rating system with a simple thumbs-up and thumbs-down layout. But why did Netflix switch? - Confusion over stars: Users often mistook the five-star score for a global site average (like Amazon or IMDb) rather than a personalized prediction of what they individually might enjoy. - One person's 3-star meant "average." - Another person's 3-star meant "pretty good." - Some users rarely gave 5 stars, while others handed them out freely. - A/B test results: Simple thumbs choices felt easier and less like an academic critique, so more people were willing to provide their feedback. In order to understand how the Netflix recommendation algorithm actually works behind the scenes, subscribe to my Substack. Link in the replies 👇

• 
  ANKUR (@rastogi_ank) reported 35 minutes ago

  @zeptocares Your amazon gift cards are not activating as per instructions on cardboard ,as sms is not sending. Please ask pinelabs to resolve this. I am tired of calling them, they are not picking. I didn't purchase gift cards from zepto for this trouble.

• 
  Harsh Tripathi (@harsh_ht08) reported 35 minutes ago

  The most important part of AI commerce isn't the AI. It's the payment. Pine Labs just launched P3P (Pine Labs Payment Protocol) — India's first agentic payment protocol built on UPI. This could fundamentally change how AI agents transact. 1) Today, AI agents can search products, compare prices, negotiate with sellers and even complete an entire purchase journey. But when it's time to pay... Everything stops. A human still needs to unlock their phone, enter a UPI PIN or approve the payment. That breaks autonomous commerce. 2) P3P solves exactly this problem. Instead of authorizing every transaction individually, users create a pre-approved UPI mandate with spending limits and conditions. The AI agent receives delegated payment authority. It can then complete payments automatically whenever those predefined conditions are satisfied. Think of it as: User → Rules → AI Agent → UPI Payment instead of User → Every Single Payment 3) The protocol works roughly like this: • User creates a UPI mandate • Spending limits are defined • Merchant/category restrictions are configured • Validity period is specified • AI agent receives permission • Agent negotiates with another AI agent or merchant • If all conditions match → payment executes automatically via UPI The user isn't approving every payment. They're approving the decision framework. 4) Imagine practical use cases: "Book the cheapest Delhi-Bengaluru flight below ₹8,500." The AI keeps monitoring prices for days. The moment fares drop... It books automatically. No notifications. No manual payment. 5) Hotel booking "Reserve a Taj property if the room falls below ₹6,000." The AI watches inventory 24x7. When the condition is met... Payment happens automatically. Reservation confirmed before prices increase. 6) Grocery optimization An AI shopping assistant continuously compares Blinkit, Zepto, Instamart, BigBasket and Amazon. It waits until: • Best pricing • Coupons available • Delivery window matches Then checks out automatically. 7) B2B Procurement An enterprise AI monitors vendor quotations. Negotiates pricing. Verifies inventory. Places purchase orders. Executes payment through approved limits. No procurement executive needed for every transaction. 8) Investment subscriptions Imagine AI investing assistants that purchase mutual funds or treasury products automatically whenever predefined market conditions occur. Or treasury bots managing recurring supplier payments. This is programmable money with programmable decision making. 9) Notice what Pine Labs is actually building. They're not creating another UPI app. They're building the payment infrastructure layer for AI commerce. Exactly as Stripe became APIs for internet businesses... P3P aims to become APIs for autonomous AI transactions. That's a much bigger opportunity. 10) Strategically, this changes how investors should think about Pine Labs. Many still see Pine Labs as: • POS machines • Merchant acquiring • Payment terminals But the company has quietly evolved into a full-stack fintech infrastructure platform spanning: • Offline payments • Online checkout • Merchant software • Lending • Issuing • Payment orchestration • AI-native payment infrastructure 11) If AI agents become mainstream over the next decade, every autonomous transaction will require three things: • Identity • Trust • Payment Large language models can solve reasoning. UPI solves settlement. Protocols like P3P solve authorization between the two. That's the missing infrastructure layer. 12) This also strengthens Pine Labs' positioning against global fintech infrastructure players. Instead of competing only in payment acceptance... They're now participating in the emerging AI commerce stack, where payment becomes an API for software agents rather than humans. That's a very different business. 13) Of course, challenges remain: • Fraud & liability allocation • Consumer protection • Mandate misuse • Regulatory oversight • Agent authentication • Merchant trust The protocol's success will depend on how safely these are handled and how regulators evolve delegated payment frameworks. 14) My biggest takeaway: The next phase of fintech won't be about humans paying faster. It'll be about machines paying safely. P3P is Pine Labs' attempt to build that infrastructure before AI commerce becomes mainstream. Whether it becomes India's "Stripe for AI agents" remains to be seen, but strategically, this is one of the most forward-looking fintech launches we've seen this year. Congratulations, @amrishrau and team! Really admire the work you guys are doing. Proud to be a shareholder and grateful to be part of this journey of putting India on the global tech map.

• 
  CEOInterviews.AI (@CEOinterview) reported 36 minutes ago

  Apple raised prices last quarter. Its CEO explained why in one sentence: a 100 year flood on memory pricing. The same shortage is showing up one layer down. Amazon lifted its 2026 capital spending plan to about 220 billion dollars from roughly 200 billion, and Andy Jassy named the higher cost of memory as the reason. Microsoft disclosed 41 billion dollars of quarterly capex that included the impact of higher component pricing. More than 100 percent of the 120 basis point fall in Apple's gross margin last quarter traces to memory cost, per its own CFO. $AAPL is where the AI buildout finally reached a consumer price tag.

• 
  Daily Stack 🥞 (@DailyStackHQ) reported 38 minutes ago

  Every AI infrastructure company eventually hits the same wall… Then: Telecom, 1998: Billions of $$$ in fiber optic cable laid across ocean floors and Wall Street called it the greatest capital misallocation in history at the time….Stocks cratered 90%. Now: That fiber now carries 99% of the world's internet traffic. Then: Amazon, 2006: Bezos started spending so aggressively on server farms that analysts publicly questioned his judgment. AWS burned cash for years with no clear profitability path. Now: AWS generates over $100B annually and funds everything else Amazon does. Then: Microsoft, 2015-2018: Tens of billions into Azure data centers while cloud skeptics said enterprise would never move off-premise at the pace Nadella was betting on. Now: Azure is now an $80B+ run-rate business. The mechanism is always identical: massive upfront capex that looks insane on a quarterly screen, a market that prices it like the money is gone, and then the contracts catch up and everyone pretends they knew. $AMZN $Googl $MSFT What companies does this remind you of today?

• 
  bottT (@Tanmay00291) reported 39 minutes ago

  @AmazonHelp Don't you guys understand.. I'm not able to add my mail or you guys blocked me or ta technical glitch happening when I'm going chat with you guys

• 
  Maheshwar Vishal (@mnvishal05) reported 39 minutes ago

  @AmazonHelp I guess order ID (already shared) is gud enough to knw the delivery vendor, the DA, which FC terminal recieved this order. C'mon where's the customer obsession? You evn failed in pretending to show good intentions to solve the problem. Anyways, leaving it here. @AmitAgarwal

• 
  mark seery (@140ismymax) reported 41 minutes ago

  Peter DeSantis talking at #agenticaisummit There will NOT be one AI chip type. If multiple AI chips take multiple years to bring to market, you have to be making assumptions about model requirements in that time frame. It's a systems problem, including the network. Constraint drives innovation. Future is bright and built together. Peter DeSantis SVP, Foundational AI Models, Custom Silicon, Quantum Computing, Amazon @awscloud @amazon @BerkeleyRDI

• 
  Project Hail Tyler |-/ (@DrGracesGlasses) reported 44 minutes ago

  DON'T CHUG A CUP OF COFFEE AND IMMEDIATELY DOWN A BOTTLE OF WATER, THEN DRIVE ON BUMPY ROADS IF YOU HAVE A WEAK BLADDER. HOLY **** I ALMOST HAD TO ASK TO PISS IN THE PORTA-POTTIES OF THE CONSTRUCTION COMPANY BUILDING AN AMAZON WAREHOUSE.

• 
  JG568 (@_JG568) reported 47 minutes ago

  @Excellentsalvic @amazon is notorious for this. Package delivered to wrong address (driver’s error 100%) I inform customer service and they talk about my frustration being understandable and it then begs me to go trespassing on others properties to try and locate my package.

• 
  ayleen . ˚◞♡ ⃗ (@calspico) reported 48 minutes ago

  @NeoIsDone @amazon hey what issue did u list when u were talking to amazon support? usually if u mark the package as not delivered they issue a full refund no questions asked idk why they’re giving u a hard time 💔

• 
  Terence T. (@terencetill84) reported 49 minutes ago

  @AmazonHelp I called amazon tech support and they said they couldn't help me becuase I didnt get it directly from them and If I got it from them directly I wouldn't have had those problems. Mind you I have another firestick and two fire Tvs.

• 
  YodaForce157 (@YodaForce157) reported 51 minutes ago

  @CashCoopa It’s not official. It’s just how amazon works, it lumps all listings together under 1 product, so if you scroll down you’ll see “5 listings from $26.94” where the 26.94 one is the official one that requires an invitation

• 
  Righteous Poodle (@PoodleRighteous) reported 52 minutes ago

  @the_transit_guy They say that amazon provides jobs that pay min wage, so that's the trickle down part. They are serious when they say this.

• 
  Jim Biggie (@JimBiggies) reported 53 minutes ago

  @AmazonHelp You only help I asked for is from the class action law firm not Amazon you are the problem

• 
  Vaelis (@Vaelis_X) reported 57 minutes ago

  I had not connected the AI memory boom to what is happening in gaming until now. A PlayStation 5 has 16GB of memory. One Nvidia $NVDA Blackwell Ultra chip uses 288GB, and a full rack holds 72 of them. That is roughly the memory capacity of 1,300 PS5s in one rack. Microsoft $MSFT reportedly used 64 racks for one of its early large OpenAI clusters, the equivalent of more than 80,000 PS5s. The campuses now being built by OpenAI, Meta $META, Amazon $AMZN, Alphabet $GOOGL and xAI are on a completely different scale. This is where the gaming problem starts. The same small group of companies supplies most of the world’s memory. Samsung Electronics, SK Hynix and Micron $MU dominate DRAM production, and their most valuable customers are no longer console makers or PC builders. Hyperscalers will pay almost anything to secure capacity for AI servers. HBM is not identical to the memory inside a console, but it still competes for the same fabs, equipment, engineers and wafer allocation. The more capacity shifted toward higher margin AI memory, the less room there is for consumer products. Building more fabs sounds obvious, but it takes years and tens of billions of dollars. Memory companies also know what happens when they overbuild. If the AI cycle weakens before those factories are ready, supply floods the market and pricing collapses. So why take that risk when current customers are already paying extraordinary prices? The effects are beginning to show up everywhere. Consoles are becoming more expensive instead of cheaper with age. PCs, graphics cards and handhelds are getting harder to price for ordinary buyers. Memory has become such a large part of some graphics card costs that the VRAM can now matter more to the bill of materials than the GPU itself. Once fewer people can afford the hardware, the entire gaming model starts to change. A new console needs momentum. It needs millions of early buyers so developers know there will be a large enough market to justify spending years building games for it. If the next PlayStation launches anywhere near $1,000, most people will simply stay on the PS5. Developers will then keep targeting the older machine because that is where the users are, which gives people even less reason to upgrade. That spiral leaves the industry with four obvious ways to protect profits: raise subscription prices, charge more for games, reduce physical ownership and cut costs. You can already see all four. Physical editions are slowly becoming download codes in a box. Studios are closing. Thousands of developers are losing their jobs. Publishers are becoming more cautious and more willing to fund sequels, remasters and familiar franchises instead of taking risks on something new. The part I find most uncomfortable is where this ends. If owning powerful hardware becomes too expensive, cloud gaming suddenly looks like the solution. Nvidia can rent you the graphics card you can no longer afford to buy. You pay monthly, play on its servers and accept whatever limits come with the service. Other people will stay on older games for longer, while developers move more aggressively toward mobile, where free to play, battle passes, adverts and in game currencies already dominate. None of this means AI caused every problem in gaming. The industry was already cutting jobs, raising prices and chasing recurring revenue long before this memory squeeze. But AI is adding pressure to a system that was already struggling. The same infrastructure boom making models smarter could also make consoles less affordable, physical ownership rarer, game development safer and cloud subscriptions more normal. The AI buildout is not happening in some separate world. It is bidding against everyone else for the same chips, memory, factories and power. Gamers may just be the first consumers to feel what that actually means.

• 
  Anmol Puri (@anmolpuri_syd) reported 58 minutes ago

  Everytime I place an order with overnight delivery, I get a message in the morning 'delivery attempted' and we could not get in. I live in an apartment with an open lobby where Amazon delivery team delivers hundreads of packages every week without any issues. @AmazonAustralia

• 
  casej (@standrews31) reported 60 minutes ago

  @Tradermayne I’ll happily eat **** but don’t see any broad equity drawdowns(I currently own a lot of Constellation software, Service NOW, Amazon) and think BTC is much closer to a bottom than some think(mid 50’s or so, and we’re at 62k as I write this)).. been wrong before but I’m betting on this being a buy range vs sell. I’m scaling back into crypto plays like DEFT, Galaxy and BTC and will stagger buys as they range/any drag down the next few months. 🤞

• 
  MBA Girly (@MBAGirly) reported 1 hour ago

  Amazon two way door framework says: If you shut down a business division, fire everyone and then it would be super hard to recover it - it’s one way door decision. If you are testing a new product, it’s two way door as it is easy to reverse it. My consultant friend told me this