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 (49%)
• Errors (27%)
• Sign in (24%)

Live Outage Map

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

City	Problem Type	Report Time
Mississauga	Website Down	9 hours ago
Grand Coulee	Errors	11 hours ago
Sanguinet	Website Down	17 hours ago
Bigastro	Sign in	1 day ago
Perth	Website Down	2 days ago
Dallas	Website Down	2 days 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:

• 
  LUMENX (@Lumenix0) reported 6 minutes ago

  CHINA'S TOP AI LAB IS NOW BUILDING ITS OWN CHIPS BECAUSE THE US BAN LEFT THEM NO CHOICE. Zhipu AI, the team behind the GLM models, ran into a wall. Demand exploded, but Nvidia is off the table, they're on the US Entity List, no American silicon allowed. So they're doing the only thing left, talking to Chinese chipmakers about a custom processor built specifically for their own models. Right now they're stitching together Huawei chips, other domestic hardware, and whatever leftover Nvidia stock they still have. Here's the irony. Sanctions meant to slow China down might be the exact thing pushing them to build faster. Cut off from American tech, you stop waiting around and start building your own stack. And it's not even a radical move. Google's had TPUs for years. Microsoft, Meta, Amazon are all doing the same thing. When you're running millions of servers for one job, a custom chip beats a general purpose one every time. Even 1% power savings at that scale is massive money, and that's the real long term threat to Nvidia.

• 
  Glampdring (@FoeHammer519) reported 7 minutes ago

  Day 6 of trying to get @amazon to fix a very basic return issue on a $500+ purchase from a third party seller. Zero reason to keep paying for Prime or ordering from Amazon over Walmart if they can't fix something like this.

• 
  Andrew Wilkinson (@StartupsILike) reported 8 minutes ago

  Amazon has now borrowed more than $100 billion over the past year. It is the largest debt issuer among every hyperscaler funding the AI buildout. The latest deal alone raised $25 billion across eight tranches, with maturities stretching out to 40 years. Peak demand hit $62 billion. Orders got trimmed down to about $41 billion once the banks tightened pricing. Roughly half the demand Amazon's $37 billion deal pulled in back in March. The sale hit more than just Amazon. Investors sold existing hyperscaler bonds to make room for the new debt. Prices fell across the sector, including at Alphabet, Meta, Nvidia, and Oracle. None of this points to weaker credit. Amazon's ratings haven't moved. Investors are starting to ask how much more of this debt the market can keep absorbing before appetite runs out.

• 
  Macro Bombastic (@MacroBombastic) reported 9 minutes ago

  @unusual_whales Mate, talk is cheap when Amazon funds your server bills.

• 
  Brian Scott (@brianscott9582) reported 10 minutes ago

  @jeff_gluck Anyone else's TruTv not working at all? Wouldn't have this issue if it was all streamed in Amazon.....

• 
  The 5-Minute Detective ⏱️🕵️ (@DetFrankFrank) reported 11 minutes ago

  Oh-ho the Amazon Prime van is a-coming down the street Oh please let it be for me Oh-ho the Amazon Prime van is a-coming down the street I wish I wish I knew what it could be Something special (something very very special now) Just for me! 🎶📦 (With apologies to Meredith Wilson)

• 
  Television David (@televisiondavid) reported 12 minutes ago

  @PassagePress @MorlockP @ark_press I ordered from Amazon UK, not directly. I am guessing it is an inventory issue in the UK. I ordered back in 2025. It hasn't dispatched. Just one of those things. Thanks for the response. I am sure it will all sort itself out.

• 
  Skwat (@SkwatSwat) reported 13 minutes ago

  @RickyPeebs They shut down all the Amazon fresh stores where I live 😭 Where is this

• 
  The Figen (@TheFigen_) reported 13 minutes ago

  Hey @nikitabier ?? You keep adding unnecessary features! That's what's important: Amazon links are being added under my paid promos! You need to fix that too!!

• 
  Ted Mosby (@TheICP108) reported 19 minutes ago

  @amazonIN @amazon @AmazonHelp 4. Delivery Issues: Two items from my orders were guaranteed for delivery today. They have not arrived, and I have received no communication regarding the delay. As I pay for Prime for reliable same-day delivery, I expect compensation for this breach of service as well.

• 
  Abdul Haqq (@AbdulHaqq123837) reported 22 minutes ago

  @MasterMaliq For a discussion of more scientific errors in the Qur'an, including texts and commentary treatment of them, see chapter four in “The Qur’an’s Difficult Verses Examined – with Commentaries, Hadith, Biographies, and History”, available on Amazon and on other online booksellers.

• 
  The Characterist (@maune99) reported 23 minutes ago

  @FLOTUS @FLOTUS Please read my book. Stop throwing money at the problem and use my solution to save all children from neglect and abuse due mainly to welfare. 80% of Foster kids come from welfare and subsidized homes. Read WALLFARE on Amazon books.

• 
  munnu π² ❖,❖ base.eth (@munnumkumar) reported 26 minutes ago

  @AmazonHelp Link not working

• 
  हम भारत के लोग (@nitishtripath) reported 28 minutes ago

  purchase your RO before some days ago form Amazon , your givan TDS Meeter not working , i believe you give me used TDS meeter please have a looking @Livpure_India

• 
  Alex Veremeyenko (@alex_verem) reported 35 minutes ago

  i love how we talk about ai automating our lives and businesses, when trillion dollar companies' user experience is just atrocious - meta is notorious for being unnavigatable - openai sucks with email change/logins - amazon website looks like it's still in 2015 - apple's siri is mentally challenged - google's dashboards are overwhelming if AI is so incredible.. fix your goddamn user experience first?

• 
  Keith Hollis (@keithhollis60) reported 36 minutes ago

  @RAIRFoundation 🤔 There are alternatives to Walmart. And cheap Chinese goods from Amazon. Sounds like a ripe opportunity for someone to compete. Knock them down a peg or two.

• 
  Leah Jane (@Redaxoid) reported 40 minutes ago

  @HunterBiden Companies compete now by stealing ideas from other companies (Amazon basics) then selling at a loss for years to drive down the competition that created the profits. Billionaires should not exist. They are cells that have become a cancer, always growing too large for the system.

• 
  Xah Lee (@xah_lee) reported 43 minutes ago

  the ugly web dev history, frontend. I was ignorant of JavaScript and frontend tech, from 1999 to 2010. am a web backend dev in 1998. back then, professional programers sneer JavaScript (and php). except, those big corp actually need to use it sans choice, such as gmail, google map, amazon, ebay. (the need to update page sans reloading the whole page.) Only when Node.js in 2009 By Ryan Dahl, changed things. It introduced the async, which puzzled and derided by most programers. (async, a pain in the ***. One of the worst thing in programing language history, still so today, even with latest await etc. Reason is because js is single threaded. Async is a pretend fix on this concurrency issue.) In 2010 js did not have a library system. (nor did php, scheme lisp, for the first 10 years of their life. emacs lisp has no module/namespace to this day.) Commonjs and amdjs module system mess got introduced in the wild. A very complicated mess, which created concepts like bundlers (browserify, webpack), code repo and package manager (bower, npm), task runners (gulp). npm is a corp greed, designed so that 1 line code is considered a lib, to promote its popularity. Then in 2015, js finally had some basic features, such as real local scoped variable (let), real dictionary/hashtable/map data structure. but for reason, the JavaScript committee refused a “use new js”. (google chrome dev got stun by a “use strict”.) So, one thousands warts of js, stay in js. (such as var, use strict, string is utf16 and all methods cant deal with emoji, no real array masking as magical object with magical length. etc.) JavaScript landscape, is really 30 years of **** pile due to a in-10-days mistake. But cpp is worse.

• 
  Calypso 📺🦈 (@bestdemonshark) reported 44 minutes ago

  @vampnoctra No geniunely is there like issues with Amazon when it comes to non- canon ship merch??? Cuz it's not like the crew are unaware of RS being popular, The Offer being by far the most sought after trading card is proof of that so what gives??? Do they not want money???

• 
  deion (@Mr_too_soon) reported 45 minutes ago

  @Majindraye Crazy I never had issues with USPS. It usually onetrac snd sometimes Amazon. I was shocked. The thing that make me mad is they said they needed an access code. Usps know there no access code.

• 
  The Graphic Master (@RealMasterMrx) reported 47 minutes ago

  @AmazonHelp @ComoAsiPty I gotta say that Amazon has been dropping the ball lately. I have no problem buying an item on Amazon that says same as new and get a discount but when you order the item and it arrives, it’s some garbage that somebody pulled off their front porch. And Amazon is reselling it.

• 
  Gunnar (@JGunnarGrey) reported 48 minutes ago

  @ChuckThePhoenix As an author and publisher, I DO use AI. It's amazing at the marketing stuff we creative types don't do all that well. Amazon keywords, categories, blurbs… Grok beats me hands down.

• 
  Order 66 (@roundedcubes) reported 49 minutes ago

  @knoyhead Thought I still had 10 days, but without notice, Amazon shipped out my ETBs. Would've canceled had I known they were shipping soon. Preordered at 103, shipped at 100. Also, right after it was marked as shipped.... the price went down a bit more to 93.

• 
  haitham (@theonlyhaitham) reported 53 minutes ago

  My Amazon rating dipped from 4.1 to 4.0 this week. Three new reviews landed and at least one was low. First instinct is to shrug it off, averages move. Wrong move. I read every new review word by word, because a negative review is the only market research that walks in the door for free. A one star that says "too spicy" is a variation idea. One that says "arrived melted" is a packaging fix. One that says "not worth $25" is a pricing test. The rating is the score. The words are the game.

• 
  Ribdro (@Ribdr0) reported 56 minutes ago

  @slowmouomo I reject your premise. The lawsuit is in regards to those who sell on Amazon are required to charge the same or lower as they have on other online stores. Sellers can sell for cheaper they just can't undercut Amazon prices of they volunteer to list on Amazon. I see no issue.

• 
  Trevor Long (@TrevorL415) reported 56 minutes ago

  Go on Reddit lists and purchase some items on their Amazon wish lists. So many are in need as this is not about to get better. Shop local. Avoid Big chains. Avoid Target 🎯. Definitely avoid Tyson Foods. I'm definitely at the point where I don't even mind if people steal from Cronie Capitalism stores, do you? Prices will not come down any ways. Even cronie motels and hotels would rather have empty parking lots then lower their room rates to get more customers/profits as the bottom line is to redistribute wealth.

• 
  Eric Sheffield 🎙️ Professional VO 🎙️ (@EricSheffieldVO) reported 57 minutes ago

  Bro why is twitters video uploading so terrible. I shot this on a 4K lens with a 55mm 1.4 aperture camera. Literally ruins all the work I put into specific lighting and angling. $2-$3k worth of equipment to look like it was shot on a $10 amazon basic webcam.

• 
  P Equity Research 📰 (@pequityresearch) reported 58 minutes ago

  Mizuho Securities: CPUs & GPUs Market Forecasts & Growth > Shipment Growth: Industry server CPU shipments are forecasted to reach 35 million units in 2026 and grow to 50 million units by 2027, representing a 40% year-over-year increase. > Long-Term TAM: The long-term Total Addressable Market (TAM) estimate for 2030 has been raised to $170 billion (up from the previous $107 billion forecast), driven by higher CPU-to-GPU ratio assumptions for AI inference servers. > CPU-to-GPU Ratios: The CPU-to-GPU ratio on AI servers is accelerating and is expected to approach 1:1 by the end of 2027 or 2028. Supply Chain & Technical Bottlenecks > DRAM Constraints: A critical bottleneck exists in DDR5/LPDDR5 supply, with a projected fulfillment ratio of only 70% over the next 12–18 months. > Demand vs. Supply Gap: Based on current models, the 2027 demand for DDR5/LPDDR5X (over 300 billion 1Gb equivalents) significantly exceeds the projected supply (220–250 billion 1Gb equivalents). > Potential Risks: The shortage of key materials—DRAM, substrates, and passives—is expected to persist through 2027 and could pose downside risks to downstream server assemblers, potentially leading to lower server rack output. Key Player Insights (2027 Forecasts) > Nvidia: Expected to reach 5.0–6.0 million units for the Vera CPU, including 2.0–3.0 million units specifically for agentic AI stack racks. > Google: Axion CPU production is projected to increase more than 2x year-over-year, aligning with the growth trajectory of TPU units. > AMD: The N2 Venice CPU is forecasted to exceed 6.0 million units. GPUs/ASICs Market Growth Projections > Rapid Expansion: The total AI ASIC market is projected to grow from 4.1 million units in 2025 to 24.0 million units by 2028. > Volume Drivers: The growth is driven by substantial increases in deployment by major hyperscalers including Google, Amazon (Annapurna), Meta, Microsoft, and OpenAI. > External Demand: The market for external (non-Google) AI ASIC units is expected to surge from 0.6 million in 2025 to 7.2 million by 2028. Key Hyperscaler Activity > Google (TPU): Continues to be a dominant player, with total shipment units increasing from 2.5 million in 2025 to 7.1 million by 2028. > Anthropic: Significant ramp-up is forecasted for their "TPU Ironwood/Sunfish" chips, moving from 0.6 million units in 2026 to 6.2 million units by 2028. > Amazon/Annapurna: Shipments for the Trainium line are projected to double from 1.5 million in 2025 to 3.6 million by 2028. > Meta: Rapid scaling of MTIA chips is expected, growing from 0.1 million units in 2025 to 2.7 million units by 2028. Technical Trends > Advanced Packaging & Nodes: There is a heavy reliance on sophisticated packaging technologies like CoWoS-L and CoWoS-S, and advanced foundry nodes including N2, N3, N4, N5, and A16. > HBM Integration: Nearly all listed high-performance ASICs utilize High Bandwidth Memory (HBM), with a transition toward newer generations such as HBM3E and HBM4/4E to meet performance demands. > ASP Variance: Average Selling Prices (ASP) range significantly, from approximately $2,000 for entry-level models to as high as $40,000 for top-tier specialized chips like the TPUv10. $DRAM $EWY $MU $GOOGL $AMKR $TSM $ASE $NVDA $AMD $AVGO $MRVL $INTC $MSFT $META

• 
  Zachary F. (@ZacharyF1) reported 1 hour ago

  @Killerewok76 @ChrstnAntwan To be fair to retailers, if they were shown these concept photos, they'd have a reason to be doubtful and not wanting to do it Being turned down by one or two of the major retailers is one thing but Walmart, Target and Amazon all turning you down simultaneously?

• 
  Joshua Levy (@ojoshe) reported 1 hour ago

  @valkenburgh @deanwball It seems like both acceleration and deceleration advocates tend to talk about AI as if it's a technology exception, a singularity. But the balance of evidence doesn't seem to support that as the clearest way to look at it. Some parts are happening extremely quickly, that's true. Current agents and models are stunningly powerful. So of course there are both foreseen and unforeseen risks. And we should have very focused conversations on those risks. If you truly believe that large language models are technologically akin to nuclear weapons, then we should talk directly about those very specific risks. But I think with the exception of a vocal minority of the most pessimistic doomers, most people recognize they're not like that in key ways. There is not a thin bright line like you have with a nuclear weapon. And the technical components for models are now widespread. So the best mechanisms for safety are more likely to involve a complex blend of new technologies, transparency, secrecy, process control, and in some cases regulation, just as it is currently with software security. The frontiers are both fast-moving and extremely jagged. The practice of software engineering has changed more in two years than it did in the last twenty. But simultaneously, many things at the layers of model capabilities, enterprise adoption, regulations, economic impact outside knowledge work, and human factors likely will not change nearly as fast as AI hype leads some people to believe. In short, patterns and incentives we see with AI are now incredibly fast changing, but they rhyme with technology shifts we've seen before. Decades ago, credit card payment networks had a lot of features of the internet, but they would have fought tooth and nail if told that their networks had to be open. Had you asked key players in that industry at the time if there should be an open network, I’m sure some would have told you it's madness and would lead to fraud. And in a way they’d be right: the internet did lead to fraud. But their argument would still have been wrong. We really could have gone down a path of having a bunch of complex, proprietary networks and no internet. Likely no Amazon. No Stripe. We could talk about lots of other examples too, like GPS or Linux servers or streaming media. Also, many times in the 20th century, American business interests have branded business threats as communism. (In fact, this is so common and extreme it’s almost comical. See historic footage like the film Atomic Cafe for examples.) This is not to advocate for blindly utopian open-source source software, let alone to advocate for actual communism! Rather, framing discussions around AI as capitalism versus communism in this regard is fuzzy thinking and doesn't help us understand the real issues. It doesn't highlight the actual strengths we have in a capitalist system to build viable businesses that do technologically powerful things. As Jim Barksdale famously said, there are two ways to make money: bundling and unbundling. Right now, it sure seems like selling APIs to foundation models is a great business. But if it does turn out that models are not as high margin as some people thought they were, then that's a sign that we need some free market innovation. Bundle models in new ways where there can be defensible businesses with high enough margins to fund real work. That's the creative innovation that we want to see in a capitalist society. Calling a belief in open tools and software communism is a disservice to those who strongly disagree with authoritarian government tendencies and believe technology developed in a free market society is in fact one of the best ways to have a positive impact on humanity.