Apple Store 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.

Apple Store Issues Reports

Latest outage, problems and issue reports in social media:

• 
  Asrar (@asrrrrrr217) reported 3 minutes ago

  @AIAdsApps I have very problem with apple store ?

• 
  NeilT (@Exogynous) reported 5 minutes ago

  @jwblackwell Anyone with any sense has now switched off system updates on their mobile. This will cause significant issues with viruses. Also it could totally tank the new phone market as people realise they are buying crippled phones. Meanwhile direct sales of China phones without crippleware will be rife. Samsung, Google and Apple will be badly damaged. It might even see the advent of Harmony OS taking off where it has been restricted for so long. If having access to the Google or Apple store means the government controlling your life, a whole generation of users will abandon the status quo.

• 
  Culture club (@boygeorgeclub01) reported 6 minutes ago

  @JillCollier16 Fix it through apple store

• 
  Big G (@Fergy_MUFC) reported 6 minutes ago

  Really don’t know what’s up with these workers at Apple Store in bay plaza. It’s like everybody have attitude. Yall think I want to be here!! As 3 times in 4 months having problems with my AirPods

• 
  Asher Crowe 🪺 (@ashercrw) reported 7 minutes ago

  A 31-YEAR-OLD IN BELGRADE IS PULLING $8,400 A MONTH OFF FIVE MAC MINIS RUNNING IN A TOWER ON HIS DESK. The whole stack costs $19 a month in electricity to operate. The hardware paid for itself in week one. The setup is so quiet his girlfriend didn't notice when he turned it on. His name is Stefan. This is the cleanest example of the new solo operator economy I've seen all year and the numbers deserve a full breakdown. The hardware is five M4 Mac Minis stacked in a tower on his desk. Each one has a number written on it in marker, 1 through 5, so he knows which node dropped when one goes silent. A pink dumbbell sits on the shelf above them. A can of compressed air on the windowsill. The whole thing hums quieter than the mini fridge in the corner. The five machines are clustered with EXO into one virtual machine. EXO is the open-source framework that lets you string together consumer hardware into a distributed inference rig without needing a degree in systems engineering. The setup runs Llama 70B locally on MLX, Apple's machine learning framework optimized for unified memory. Nothing he runs ever touches a cloud server. No API costs. No rate limits. No latency tax. The model runs on his desk and answers in milliseconds. Here's the workflow he built around it. A client uploads a raw manuscript. Anywhere from 60,000 to 120,000 words. Indie author novels, self-help books, faceless YouTube channel scripts, the kind of long-form content that needs narration but doesn't have a studio budget. The Llama 70B model does the reading work first. It ingests the raw text, cleans the formatting, splits the chapters automatically, and tags every line of dialogue with the emotional tone it should be read in. Excited. Whispered. Angry. Resigned. Then it writes the chapter descriptions that faceless YouTube channels paste directly under their uploads. All of it done locally. All of it done in one pass. Then an open voice model on the same stack takes over and narrates the entire book in a single locked voice. The voice never gets tired, never asks for a re-record, never raises its day rate, never catches a cold the day before a session. The same voice across every chapter, every book, every client. Consistency that human narrators physically cannot match. A local audio mastering model handles the final polish. Compression, leveling, breath cleanup, room tone matching. The output is studio-quality audio ready for upload. The stack renders 28 hours of clean narration per month while he sleeps. He wakes up, exports the files, sends them to clients, invoices them, and goes back to whatever he wants to do with his day. Now the part that breaks people. The power draw across all five machines running at full load is 180 watts. He has a KUMAN meter plugged into the wall to track it. A single gaming PC idles higher than that. The entire AI studio he built consumes less electricity than a hair dryer on low. At Serbian residential rates that works out to roughly $19 a month in operating cost. Eight thousand four hundred dollars in, nineteen dollars out. A 442x margin on power alone before you account for the fact that the hardware paid for itself the first week he turned it on. His girlfriend asked why the power bill didn't move after he built it. He told her it can't, the machines barely draw anything. She asked what the whole thing cost to set up. He told her. She asked why he didn't build ten. That's the right question. A traditional audiobook studio has a narrator on a day rate, a booth, an engineer, and a monthly power bill that buries solo operators. The cheapest professional narrator in the US charges around $200 per finished hour. The cheapest decent one runs closer to $400. A 10-hour audiobook costs an indie author at least $2,000 in narration alone, plus mastering, plus mixing, plus the three week turnaround time while the narrator fits the project into their schedule. Stefan delivers the same product for a fraction of the cost, in 48 hours, with consistent quality across every chapter, and his only constraint is how fast he can find clients. The economics are completely deranged compared to traditional service businesses. He doesn't pay rent on a studio. He doesn't pay a narrator. He doesn't pay for cloud compute. His marginal cost per audiobook is approximately the electricity it takes to run the cluster for the duration of the render, which is measured in pennies. A few realizations worth sitting with. The frontier of AI economics is no longer in San Francisco. It's in apartments in Belgrade, Lagos, Manila, and Tbilisi, where operators with low overhead and high technical curiosity are quietly running businesses that look impossible from the outside. The geographic distribution of who actually makes money from AI is going to look nothing like the geographic distribution of who funded the labs. Local inference is the quiet revolution nobody on this app is talking about loudly enough. Every workflow that currently runs on OpenAI or Anthropic APIs has a cousin that runs on a Mac cluster for the price of an electrical outlet. The companies paying $30k a month in cloud bills are going to wake up in 18 months and find their margins eaten by operators paying $19. The audiobook market is just the beginning. Every service business with high human labor costs and predictable output requirements is about to get the same treatment. Voiceover work, transcription, translation, copywriting, image editing, video editing, customer support, technical writing. Each one of these has a local-inference version waiting to be built by someone with a stack of Mac Minis and an EXO config file. Stefan didn't invent anything. He just connected the right pieces. The pieces have been sitting on GitHub for over a year. The Mac Minis have been on shelves at every Apple Store. EXO is free. The voice models are open. The orchestration is a weekend project. The only barrier was knowing it was possible. Now you know.

• 
  Vel0x (@vel0xAI) reported 14 minutes ago

  A student in the United States received a $3,000 university grant and spent the entire amount on five Mac Minis, not because he wanted a better study setup, and not because he was trying to impress anyone in his dorm, but because he was tired of waking up every morning and explaining his life to an AI that had forgotten everything by the next session. He did not use the money for textbooks, private tutoring, paid courses, or a new laptop like the university probably expected. He went to an Apple Store, bought five small machines, carried them back to his dorm room, numbered them from 1 to 5 with a black marker, stacked them on a cheap metal shelf beside his desk, connected a power meter to the wall, made instant noodles, and went to sleep while the machines began turning his room into something that looked less like student housing and more like a private AI lab built on scholarship money. His neighbors thought he was mining crypto, which made sense from the outside, because all they saw was a shelf full of computers running through the night, cables hanging behind the desk, a small fan pointed at the stack, and a student who suddenly cared too much about wattage. What they did not understand was that he was not trying to mine coins; he was trying to build a system that remembered his classes, his assignments, his codebase, his mistakes, his goals, and the product he was quietly building while everyone else was still treating AI like a smarter search bar. The problem he wanted to solve was simple but annoying enough to change everything. Every time he opened a new AI chat, he had to explain who he was, what he was studying, what project he was building, what the professor wanted, which parts of the codebase were broken, what he had already tried, what had failed, what he had learned the day before, and why the answer needed to fit his specific situation instead of sounding like generic advice from a model with no memory. He realized that the most valuable thing was not another chatbot, but a system that could keep context long enough to become useful. Each Mac Mini became responsible for a different part of his life. One machine processed his lecture notes and turned them into explanations he could actually understand. Another reviewed his assignments before submission and checked whether his arguments, code, and formatting matched the requirements. A third acted like a private tutor that questioned him until he could explain the material back clearly. A fourth wrote, tested, and refactored code for the product he was building outside class. The fifth coordinated the whole system, kept the rules updated, stored the context, and decided which task needed to run next while he was sleeping. There was no development team behind it, no manager assigning tickets, no daily standup, no productivity consultant, and no university department guiding the experiment. There was only a rules file, five machines on a dorm shelf, and a student who understood that local AI became much more valuable once it stopped being a conversation and started behaving like infrastructure. The university had given him money for education, but he used it to build an education system that did not forget him. That was the part most people missed when they saw the setup. The point was not only that the machines were powerful enough to run useful models locally; the point was that they belonged to him, which meant his lecture notes, unfinished code, business ideas, exam prep, personal mistakes, drafts, and prompts stayed in his room instead of being uploaded into somebody else’s cloud dashboard under somebody else’s terms of service. During the day, he still went to class like everyone else, listened to lectures, submitted assignments, and looked like a normal student trying to get through the semester. At night, the system summarized readings, found gaps in his understanding, generated practice questions, cleaned up code, tested features, wrote documentation, and moved his side project forward without needing him to sit there and manually push every step. When he woke up, he was not starting from zero like everyone else opening a blank chat window. He was starting from wherever the machines had stopped. At first, people in the dorm laughed at the shelf with the numbered Mac Minis because it looked excessive, strange, and slightly ridiculous for a student room. Then they started asking him to summarize lectures they had missed. After that, they asked whether it could help them prepare for exams, review essays, explain technical concepts, debug projects, and remember the context of their classes without forcing them to rewrite the same background information every time they needed help. That was when the private study system became a product. He packaged smaller versions of the setup for other students, not as a replacement university and not as another generic AI wrapper, but as a memory layer for people who were tired of using tools that forgot them every morning. It became private study agents, class note summarizers, exam preparation bots, coding copilots, and project assistants that remembered the user’s material, progress, weaknesses, and deadlines. The grant was $3,000, the machines cost less to run than most monthly subscriptions, and the first paying users came from the same dorm that had originally joked he was mining crypto. What started as a way to survive his own semester turned into a product other students were willing to pay for, because it solved the problem they had all accepted as normal. Now the system makes around $45,000 a month, and the strangest part is that none of it began as a startup pitch. It began as a student using university money to stop repeating himself to a machine. The university thought it was funding his education. What it actually funded was the infrastructure he used to rebuild it.

• 
  StephanInc (@IncStephan) reported 14 minutes ago

  @durreadan01 I have a broken Air right now , after 4 trips to the Apple Store, it’s still not fixed correctly!

• 
  Angela Rodriguez (@Angel1994_) reported 20 minutes ago

  a down payment. I have never leased anything from T-Mobile or Apple—the iPhone was purchased alright at the Apple Store—never had the screen saver displayed on the illustrated iPhone 13.

• 
  sun 𓆩♡𓆪 (@theodysseis) reported 21 minutes ago

  the apple store cannot fix it (i think it’s my phone) pointed frustration. then a sedated acceptance washes over me………

• 
  Memories collection (@gunpeiyokoifan) reported 21 minutes ago

  Also "having to forcefully stop yourself from (over)sharing on a special interest" I'm so screwed, I once seen a coworker ask why his iPhone wasn't working and I really wanted to fix it, but that would seem creepy because I'm NOT at an Apple Store yet

• 
  Ashok Shetty (@savidhyashok) reported 22 minutes ago

  @poonamjourno @AppleSupport @Apple In the cost they will quote you may get a Good Brand Tab any day. I had approached the Apple store with Macbook issue of key pad numerical numbers key not working And they quoted Rs 30,000/-

• 
  Trajen (@trajen) reported 22 minutes ago

  @VicVijayakumar You still have to pay full price for the replacement until the broken iPhone is shipped back. I think it’s more economical to go to an Apple Store and pay $99 for a swap.

• 
  O͜͡l͜͡o͜͡l͜͡a͜͡d͜͡e (@OnlyOloladeMi) reported 23 minutes ago

  @ogidioluautos Or you can login your x account on the phone that purchase with their Apple store or play store

• 
  sivat.eth (@0xSiva) reported 24 minutes ago

  @poonamjourno @AppleSupport @Apple I'll recommend visiting the nearest Official Apple Store (if available in your city) If not, reach out to their international customer support from Apple official support iOS app Apple official customer care is top notch and it's very likely, all your issues will be resolved

• 
  Fandom Pulse (@fandompulse) reported 25 minutes ago

  Star Trek writer Peter David on why he hated Star Trek: Discovery: "I have no idea what any of it looks like. Every single scene was apparently filmed in a power outage. The sets are dark, the costumes are dark, everything is so dámņëd dark. I mean, yes, people said that the new Enterprise in the Abrams film looked like an Apple store, but at least you could see it. I can’t see crap in ST:D. " Is this a common problem with modern tv shows?