Unexpected €15,000 Azure token bill after a bug, can someone from billing help with my full-waiver request?

Ghaith Almasri 60 Reputation points
2026-09-05T14:25:28.9066667+00:00

Hi everyone,

I’m posting because I honestly don’t know what else to do. A bug caused unintended token usage on my Azure subscription, and I ended up with a bill of around €15,000.

I already have a support case open and asked for a full waiver. Microsoft offered to waive around €4,700, but I haven’t accepted that offer. I’ve escalated again to ask a supervisor or manager to review a full waiver of the accidental charges.

I’ve never received a goodwill billing waiver before. This isn’t money I have, and even with the €4,700 reduction, I still couldn’t afford the remaining bill. It’s been a really stressful few days, and I’m genuinely overwhelmed by the situation.

Is there someone from Microsoft’s billing or finance team who could look at my existing case and speak with me directly about my full-waiver request? I’ve already tried the support route, but so far it hasn’t helped me resolve this, and I’m not sure who else to contact.

I can explain what happened and provide my case number and supporting details through a verified private Microsoft support channel. I’d also like to know whether payment deadlines and collection activity can be put on hold while the case is being reviewed.

If anyone has been through something similar and managed to get a full waiver after an initial partial offer, I’d really appreciate hearing what helped. I’m hoping someone can help me reach the right person to review this.

Thank you for reading.

Cost Management
Cost Management

A Microsoft offering that enables tracking of cloud usage and expenditures for Azure and other cloud providers.

0 comments No comments

1 answer

Sort by: Most helpful
  1. Nathan Roberts (SN) 13,536 Reputation points Student Ambassador Volunteer Moderator
    2026-09-05T19:23:02.47+00:00

    Hey there, Ghaith Almasri

    Sorry to read about your experience with Azure and the token usage causing a high bill on Azure. I can not imagine the amount of stress this has caused.

    Sadly, I have seen people post on the community who have fallen foul to unexpected costs incurred from Azure. It is mainly university students and people who are experimenting with Azure that run into unexpected costs because they either forget about the resource and accidently leave it running or chose a resource that has a high cost % for usage they didn't realise.

    I have found the best way to get a resolution about this is to reach out to the billing team on Azure Support, which from reading your question, is what you have done. Although, the costs they had ran up was ~£1,000 or less.

    Asking to speak to the supervisor is a good step, as this allows you to speak directly to them and see if they are able to waiver the cost for you. They may have more permissions then the agent (But I am unsure for sure if that is the case).

    Unfortunately, we are unable to do this and I don't think the Microsoft external staff are able to communicate to the team who can issue a refund directly.

    In terms of the bug that caused the high token usage, have you explained this to the billing team?

    As you mention token usage, I am guessing this is a project in Microsoft Foundry. However, it if isn't, I am sorry for mistaking this for Foundry. As you have identified a bug, please report this bug using the built-in feedback method in the Foundry portal by signing into your account here: https://ai.azure.com/

    Once you have signed into your account, please click onto the face icon on the left-hand side of the face

    User's image

    This will bring up a feedback form where you can enter information about the bug that caused a high token usage. Please make sure you tick the box to allow Microsoft to contact you. This will allow the product group to see this feedback. This should be an addition to contacting the supervisor.

    Thanks,
    Nathan

    Was this answer helpful?


Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.