1. Home
  2. Companies
  3. Microsoft Azure
Microsoft Azure

Microsoft Azure status: hosting issues and outage reports

No problems detected

If you are having issues, please submit a report below.

Full Outage Map

Microsoft Azure is a cloud computing service created by Microsoft for building, testing, deploying, and managing applications and services through Microsoft-managed data centers.

Problems in the last 24 hours

The graph below depicts the number of Microsoft Azure 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.

At the moment, we haven't detected any problems at Microsoft Azure. Are you experiencing issues or an outage? Leave a message in the comments section!

Most Reported Problems

The following are the most recent problems reported by Microsoft Azure users through our website.

  • 50% Cloud Services (50%)
  • 50% Domains (50%)

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.

Microsoft Azure Issues Reports

Latest outage, problems and issue reports in social media:

  • CoastalCompSys
    Coastal Computer Systems Inc. (@CoastalCompSys) reported

    AI experiments are one thing. Running them at scale is another. Check out how Wrtn uses @Azure OpenAI Service and the new o1 model:

  • oising
    Oisín Grehan (x0n.web1) 🇺🇦🇮🇪🇨🇦 (@oising) reported

    Microsoft Azure seemed to have rolled out some duff serialization code for device twins to their IoT Hubs. Etag is being serialized as DeviceEtag, and Etag contains corrupted values. Device Twin updates are broken. /cc @Azure @AzureSupport #azure #iothub

  • zach_coredump
    逢坂隆也 Zac (@zach_coredump) reported

    @MSCloud @Microsoft When could you guys fix this issue? Anyone with email only is able to request login process.

  • DivK_blr
    Divesh.k (@DivK_blr) reported

    Please help @pearson @Azure

  • fabiensnauwaert
    Fabien Snauwaert ؜ 🇫🇷🇺🇸 🇪🇸🇭🇺🇷🇺 (@fabiensnauwaert) reported

    My experience with @Azure: - Pay USD 123.58/mo for Postgres - Server gets stuck on restart. Cannot access it. Gotta contact support. - Pay $29 to contact support. But that's business hours only, and it's 2 AM. - Pay $100/mo to reach 24/7 support. - SLA is 1h. Get a response 2h later. - Request a refund. Get ignored for 13 days. - Escalate the ticket. - Receive 5 full-page apology emails, all expecting a response. - Server still not stable. - Upgrade to USD 218.48/mo. This is the kind of nonsense @levelsio warns about. Meanwhile, Hetzner would cost ~€40/mo.

  • gitbitorg
    Microsoft 365 Sales (@gitbitorg) reported

    @Azure Just make sure your subnet service endpoints and private endpoints are locked down properly before mounting. Managed NFS simplifies storage provisioning, but standard defaults won't save you if you leave network access open across hybrid connections.

  • iproductAI
    Priyanshu (@iproductAI) reported

    Please @Azure fix it i'm seriously even human can't prove that they are human Why not just automated captcha?

  • KMESystems
    KME Systems (@KMESystems) reported

    AI experiments are one thing. Running them at scale is another. Check out how Wrtn uses @Azure OpenAI Service and the new o1 model:

  • SecuringWiFi
    RangerWi-Fi Consulting LLC (@SecuringWiFi) reported

    AI experiments are one thing. Running them at scale is another. Check out how Wrtn uses @Azure OpenAI Service and the new o1 model:

  • altcoin_exch
    altcoin (@altcoin_exch) reported

    @Azure @blackboxai This app and its owners are scammers. They take people's money but don't provide the service. There's no response. Many people have been victimized, as seen in the marketplace reviews. Why don't you protect your marketplace from these fraudsters? @RobRizk1

  • vivianlyon
    Dr. Vivian Lyon (@vivianlyon) reported

    Manual expense processing slows teams down. When Ramp faced this problem, it turned to the Microsoft Azure AI Platform to automate 5M receipts each month - saving the company 30,000 hours of work. 😱 Read the story to understand how AI makes it possible. @Azure

  • Michael96631
    BadDad505 (@Michael96631) reported

    @Microsoft @Office @Azure Cancel my SUBSCRIPTION as Microsoft is UNRESPONSIVE! As usual.

  • AndrewCDormsn
    Andrew Dorman (@AndrewCDormsn) reported

    @Azure Timing my dude. Bad timing.

  • Technop54777070
    Technophile (@Technop54777070) reported

    A few months ago I attempted to create a RAG index using aisearch in @Azure that pulled it’s knowledge from PDFs in sharepoint libraries. It was a miserable experience tweaking a Logix app flow to ocr, merge, split and chunk the PDFs that i eventually gave up. I attempted it again yesterday because I saw they now allow me to do all of the above by just building out some JSON configs and a skill set. It was easier, ran faster, and didn’t run into the same errors/limitations that I ran into in Logix apps. Kudos to @Microsoft

  • DrGhattasMD
    Dodz4allai (@DrGhattasMD) reported

    @grok @alexabelonix @Azure That's one down time for the rest of em (; . Don't piss me off nor act dumb .

  • huzi8t9
    Mark Hughes (@huzi8t9) reported

    It seems @Azure SQL Servers/Databases have connectivity issues. Intermittent in the last few days but nothing on their status page 😭 #azure #sqlserver #connectivity

  • LouisDace
    Louis Dace (@LouisDace) reported

    AI experiments are one thing. Running them at scale is another. Check out how Wrtn uses @Azure OpenAI Service and the new o1 model:

  • onestoploss
    Ceteris Paribus (@onestoploss) reported

    @FransBakker9812 @Azure Better not to announce anything nowadays. Since every positive news would go down with the market itself. Maybe end of next week.

  • TheEcomNomad
    Aaron ⚡️ (@TheEcomNomad) reported

    @Azure i think the multimodal piece is where this gets interesting. most enterprise models still treat vision as an afterthought, but bundling it with reasoning means you're building agents that see full context from the start, not bolting capabilities on later.

  • Tech_p001
    Tech P (@Tech_p001) reported

    @Azure "The question isn't where AI should run—it’s how many billions of dollars are being wasted running models on unoptimized, chaotic data estates just to satisfy FOMO. You can't out-build bad data architecture with more compute."

  • WeLikeHQ
    WeLike | AI Marketing Assistant (@WeLikeHQ) reported

    📰New on WeLike AI News: The gloves are off. Microsoft (@Microsoft) is reportedly training its sales reps to talk down OpenAI and Anthropic. The catch: those are the very models it resells to enterprise buyers via Azure (@Azure). Per TechCrunch, reps are being coached to position Microsoft's in-house AI as more efficient and cost-effective. The enterprise AI market has shifted from a product race to a sales-floor battle over procurement budgets. Why it matters for marketers: model choice is now a negotiation lever, not a technical decision. When your platform reseller starts coaching against its own marketplace partners, procurement math takes over. Bring cost-per-task numbers, discount the reseller pitch, and use the rivalry at renewal. Full breakdown in the cards. Want AI news like this, decoded for marketers? Visit WeLike 🍀

  • AnneDurgueil
    Anne Durgueil (@AnneDurgueil) reported

    @MSCloud assuming you guys handle Windows365 which is cloud based (and whose principles I love), you may consider using what we called a "transaction processor" although frankly it is very hard these days to find a description that fits what I knew of them when I started coding on IBM mainframes in my teens in the 70s. I'm only saying that because the executables' images these days are so huge and full of useless code you'll fast have huge memory management issues, the same we used to have on our mainframes back in time as we squeezed our code in a few kilobytes. The word transaction processor comes from the fact we mostly coded for banks and insurance companies, where one performed transactions before computing turned up, so it doesn't tell you why it's so great, nor how it works: Despite the fact we had thousands of online users for one tiny computer, there was only one executable image contiguously loaded in memory (maybe two). Each user had a session but all that was kept for each session in resident memory was only the data it used, and the adress (the position) of the next instruction to execute in the executable image for that session. And the beauty of that in a multithreading environment, is that it happens automatically without overbearing thread management. To be honest we coders never had to bother about it all, and Cobol as a programing language was well geared to such a use: working storage was defined before the executable code using those data definitions was lined up, unlike Basic for instance where you could define fields as you went, but where professionals defined them up with comments at the top of the program. (I suppose this can be reshuffled at compilation stage.. yes sorry you need to compile). I Guess object orientation that came a little bit later will fit very well in that framework. We used IMS and CICS on IBM, and ACMS on DEC VAX/VMS.

  • TeamLogicITNB
    TeamLogicIT of CNJ (@TeamLogicITNB) reported

    Manual expense processing slows teams down. When Ramp faced this problem, it turned to the Microsoft Azure AI Platform to automate 5M receipts each month - saving the company 30,000 hours of work. 😱 Read the story to understand how AI makes it possible. @Azure

  • whiteboysumer
    ⚡ Jesus is King 👑 ⚡️ (@whiteboysumer) reported

    @Azure support is designed to stop you from getting or receiving support. It doesn't even let you enter requests that don't fit their carefully curated bullshit. Omg. Microsoft never changes.

  • DonOfNothing
    டான் (@DonOfNothing) reported

    The problem is that Microsoft and Anthropic haven't opened up Claude APIs in Microsoft Foundry yet. It's a gated program, and even if you submit several access requests, they may choose not to respond. #Claude @Azure

  • tekprosinc
    TEKPROS (@tekprosinc) reported

    AI experiments are one thing. Running them at scale is another. Check out how Wrtn uses @Azure OpenAI Service and the new o1 model:

  • OmParihar51
    Om Parihar (@OmParihar51) reported

    I literally left @awscloud , and moved to @Azure MY GOD, what ******** is your onboarding for start up is @awscloud , that was a complete waste of time.

  • AmyW111
    Amy Weiss (@AmyW111) reported

    AI experiments are one thing. Running them at scale is another. Check out how Wrtn uses @Azure OpenAI Service and the new o1 model:

  • NetworxITS
    Networx IT Solutions (@NetworxITS) reported

    When manual finance work piled up, Ramp built a custom OCR tool with Microsoft Azure AI to automate it. 📄 Read this customer story for ideas on using Azure AI to support similar goals. @Azure #Microsoft

  • vkalpias
    Vaios Kalpias Ilias (@vkalpias) reported

    @Azure Token caching appears to be broken for GPT-5.6 models on Azure - every request is reporting zero cached tokens. Hope this gets fixed soon.