How Microsoft Vaporized a Trillion Dollars

You’re deep in the cult I can see. I hope you one day get better and see that being an zealot for a shit ideology which killed millions isn’t the way. Good luck friend.

[-]

programming-ModTeam@reddit

Your post or comment was overly uncivil.

[-]

I don't have any specific recommendations since doing so will basically dox me. My suggestion would be to find a company focused on engineering. At the end of any interview, you can ask these kinds of questions

How is quality viewed in engineering, what does success look like, how long do features get worked on after they are "done"
What's the ratio of new development vs bug fixing and firefighting
Is the company engineering, business, or production lead (or other stakeholders that are relevant)
If the company is business development focused, it might be a red flag, it also might not be though
Who has the power, how much autonomy do engineers have
How often do engineers get randomized
How important is process vs just getting work done (I hate burdensome process)
How supportive is leadership to engineering, how involved are they
How often do engineers get randomized

I've learned there are a ton of questions you can ask to see if the people you're interviewing think its a nightmare without asking if its a nightmare to work there.

[-]

RubikTetris@reddit

Hey thanks for the answer. For what it’s worth I already work at a very nice place with a strong focus on engineering. I was just wondering how Msft was on the inside for a dev because I think out of all the bigger companies it would probably be one of the better fits for me.

[-]

torrent7@reddit

oh, totally misread your post.

for msft, its a huge company and your boss/skip can do a lot to insulate you from all of the bullshit

i will say that the attitude in the blog post is pretty prevalent from what i saw though

[-]

BlueGoliath@reddit

The old guard leaves and everything turns to chaos and anarchy because no one plans for the future.

[-]

torrent7@reddit

Honestly, I think a lot of it is dictated by the board/Sataya.

As soon as they fired all of the test engineers and pushed that work onto everyone else without ever changing schedules or budgets meant that this was inevitable. One instance out of many of dissonance.

I was never on the Azure side, but holy hell, its a verbatim copy of what I experienced - you just have to change the org names and product names.

[-]

roodammy44@reddit

Yeah, I was at Microsoft under Ballmer and the engineering was the best I had ever experienced. It’s just management was insane. No fixing bugs. Ignoring fundamental features chasing flashy features that were immediately abandoned. Musical chairs reorgs every 6 months.

When I heard Microsoft dropped its process for agile and then fired all the testers, I think the die was cast. Bad management and bad engineering? That’s quite a combination.

All through this series of articles it made me wonder where the management with actual technical goodness is? Why didn’t one of the old guard step in and shout “enough with this bullshit”.

[-]

torrent7@reddit

From my 8ish years at MSFT, Seniors/Principals spent half their day in meetings and the other half managing people, hardly getting any time to actually do real work.

When I hit that level, you realize you spend most of your time trying to do damage control and keep your organization lead from committing to too much work while trying to calm down your reports when they see how much work your organization lead committed to.

[-]

MiserubleCant@reddit

When I hit that level, you realize you spend most of your energy trying to do damage control and keep your organization lead from committing to too much work while trying to calm down your reports when they see how much work your organization lead committed to.

I swear out of any role at msft, middle management has the least amount of agency.

I think that's the same just about everywhere.

(To be clear, this is a dig at corporate management culture in general, not a defense of msft. I've never come close to working at msft or anywhere even remotely like it, but as soon as I became a middle manager, it was the same shit: most time went to pointless meetings, fighting a losing battle with 'managing upwards', no time for actual work. Bailed on that and demoted myself to a generic dev and carefully avoided promotion ever since, because sanity > salary)

[-]

kani_kani_katoa@reddit

I've experienced the same. Middle management is hell and I'll stay an IC till I retire.

[-]

MiserubleCant@reddit

I know agile (as it is actually practised) isn't necessarily very popular around here, but one thing agile (as it was originally conceived on paper) got very right, imo, is the inversion of control. Project managers don't tell devs what to do, devs tell PMs what to do. Instead of "you need to finish widget X by the 27th, so you need to work on Y", it's "for us to finish widget X by the 27th, you need to get management to sign off decision Y".

Every single arrow in an organisational chart should go backwards. The most important people are at the 'bottom'. The customers tell the customer service people what they want, from that the customer service people tell the devs what they need developing, from that the devs tell the middle management what decisions they need deciding, from that the middle management tell the upper management what they need budget for, etc. Managers shouldn't be managers, they should be something more like 'coordinators'.

And this isn't just self-importance from a dev perspective, this is from a customer perspective, in every sector of business I ever interact with. You see it literally everywhere: C-tier fuckwits pulling strategies from their arse and that shit trickling downwards when the customers literally tell the front-line staff what they want and need, but the front-line staff are not empowered to do it because it contradicts with policies and strategies made from on high that have precious little correlation with reality, because all the arrows are going the wrong fucking way.

Well, a guy can dream.

[-]

kani_kani_katoa@reddit

Oh I remember reading the manifesto not long after it came out and I thought it was spot on. Like everything good, it go co-opted by the system and turned into another way to perpetuate top-down control.

I run a small software dev agency and we run our projects using agile as originally intended. It actually works really well in that environment, but I guess that makes sense given a bunch of the people that wrote the manifesto were in consulting.

[-]

guareber@reddit

You mean you don't think SAFe is a good enterprise framework to work from?

I'm shocked! Shocked, I tell you! No one manages to read through the whole thing.

[-]

Morlark@reddit

Managers shouldn't be managers, they should be something more like 'coordinators'. They should all view themselves as subordinate to the people 'below' them.

There's the old adage that one of the most important skills for any manager to learn is how to manage upwards.

[-]

AlarmedNatural4347@reddit

It’s the MBAs corporate world unfortunately. People who only know how to… well do they know anything? Read a budget and think strategically to maximize their next bonus maybe? Anyway they are running everything while patting themselves on the back while everything turns to shit

[-]

JJJSchmidt_etAl@reddit

This is what the role of staff developer/engineer is supposed to be for in theory right?

[-]

torrent7@reddit

It doesnt have to be that way. The next company i worked for actually had leadership teams with real engineering experience and could digest what they were asking for and listened/implemented feedback from the lower levels.

It makes all the difference in the world when leadership actually knows what the fuck is going on rather than arm chair directing

[-]

Practical-Positive34@reddit

Were you on the Sharepoint Team? That's the team I was on and man I hated that job so damn much. God sharepoint was/is such a piece of absolute trash...

[-]

vky_007@reddit

Yep. I remember Surabhi.

[-]

torrent7@reddit

No, in the Xbox org

[-]

Practical-Positive34@reddit

sounds more fun lol

[-]

torrent7@reddit

Only surface level fun. Sometimes its toughest to see a product you love die a slow death.

Coincidentally, I did work on sharepoint as an intern a wayyyyyyy long time ago.

[-]

OffbeatDrizzle@reddit

surface

ehehehehehehh

[-]

float34@reddit

Hi, may I reach you in DM with a career advice question, given your past experience in Microsoft? Thanks.

[-]

torrent7@reddit

Yeah, that's fine, cant guarantee it will help though

[-]

RabbitLogic@reddit

I heard from some friends that SharePoint CI experiences issues practically daily.

[-]

Outside-Storage-1523@reddit

Thanks for sharing, is kernel team still sound? I think the old guards are still there.

[-]

torrent7@reddit

I'm sure the kernel team is a thing, unsure what it looks like

[-]

Outside-Storage-1523@reddit

Thanks. Always wondering how the team is doing.

[-]

verrius@reddit

You think that MSFT is this panacea when you start there out of college, but in reality, it was horribly run.

I'm really curious when you graduated, because the only stories I've heard coming out of people recruiting for MS is that essentially its a complete clown show, that's only worth it for the money. But maybe that's cause I mostly remember talking with Windows team people.

[-]

torrent7@reddit

2012ish

[-]

blind_ninja_guy@reddit

I'm not sure I will ever take Microsoft seriously again after finding out that they decided that just because one of my fellow students in my graduating class didn't graduate from an Ivy League school, she would get a lower salary, and that if she had graduated from an Ivy League school she would get a higher salary. Nothing about her talent or anything, totally which school. Which I thought was the stupidest thing ever, because I didn't exactly go to a school with what was then known to be a bad computer science tier.

[-]

torrent7@reddit

that doesn't make any sense to me - something else is going on

salaries at MSFT are normalized as part of salary bands. there isn't much room to move around in the band, especially for new hires/level 59s

[-]

Waterwoo@reddit

I was always curious about the Satya success mythology when I was using Microsoft products throughout that time and they were clearly only getting worse and worse. Glad it's catching up to them. This is a perfect encapsulation of everything wrong in software engineering over the past decade and it is only getting worse with LLM coding.

[-]

ysustistixitxtkxkycy@reddit

I left after a VP who took direct leadership of ICs for a very important project excitedly explained that he had got permission to skip all mandated security and privacy reviews - for a Windows feature.

Absolutely bonkers and going against all of the lessons we had learned over decades of engineering.

Management's bet was that this was a new age and it would take bold visionaries to revolutionarize software engineering. Mine was that I was too old for this shit ;)

[-]

gimpwiz@reddit

Bold visionaries aren't needed to ship code with blatant security flaws that lead to enormous real-dollar losses to customers, but I suppose being one really helps it move faster, eh?

[-]

OxfordTheCat@reddit

It's a calculated risk.

MSFT is demonstrably successful and still going strong despite shipping with security (and sometimes stability) taking a backseat.

It's a values misalignment.

MSFT isn't in the business of technical perfection, it's in the business of making money. Those real-dollar losses incurred by clients are just a risk to be managed and accepted.

[-]

heyheyhey27@reddit

Not calculated very well...they ruined the quality of so many of their products. As if they never heard of the Trust Thermocline.

[-]

ysustistixitxtkxkycy@reddit

Arguably it's the right move from a career velocity perspective: first plow down the rollercoaster of fail-fast-with-unrealistic-expectations faster than anyone else ever could, and then reveal a "tada, nobody could have seen this coming!" completely unexpected failure with a slide about lessons learned and get promoted again for discovering and fixing all the issues...

[-]

gimpwiz@reddit

Have you considered plowing down the rollercoaster of very quickly executed poor decisions, and then right before it looks like it's about to crash, simply leaving for a better job title at a different company? This way you never have to admit any sort of fault for anything, and if anyone asks, it was those idiots who took over your work who ruined years of excellent planning and execution, seems like you were the only one even holding the team together.

[-]

JJJSchmidt_etAl@reddit

"Since the brass knows me, I got permission to remove the brakes and safety features on my fork lift. Yipee! Think how fast I'll move this pile of explosives!"

[-]

starsfan18@reddit

Recall, by chance?

[-]

ysustistixitxtkxkycy@reddit

Indeed. Now I wonder if we've worked together ;)

[-]

KevinCarbonara@reddit

It sounds like Satya Nadella got his successes through YOLOing unfinished code and it’s now become company culture, but there’s only so many times you can do that before it destroys the business.

I think it's more malicious than that. I think Ballmer pushed forward a bunch of ideas and focused on developers to increase the stock price, then reversed course when those plans failed to deliver, to make way for the next CEO.

Satya also pushed a bunch of ideas and focused on developers, increased the stock price, and now that things are going the other way, he's reversing course to make way for the next CEO.

It's just stock manipulation, really.

[-]

ModernRonin@reddit

"Prepare three envelopes..."

[-]

jacenat@reddit

When I heard Microsoft dropped its process for agile and then fired all the testers, I think the die was cast. Bad management and bad engineering?

Kinda sounds like Boeing after the MD "merger". When the issues publically surface, it's much too late and stuff will be BAD for a while.

[-]

torrent7@reddit

its exactly like that

[-]

Synaps4@reddit

And yet, despite the MD merger being in 1997 and all the well documented shitshow since....Boeing is today the 4th largest defense contractor and the largest exporter company in the united states.

Incompetent management doesnt always lead to failure, and it may not with MSFT

[-]

torrent7@reddit

yeah, i think its important to make the distinction that companies can be successful but be awful to work at

they can also be successful for a time, until they are not

[-]

japanfrog@reddit

Satya is just the pretty face that is allowing the board to hyper focus on stock price as if they were all VC trying to gut it.

Engineering has taken a nose dive and the culture that Satya has (perhaps inadvertently) promoted within is one of executive elitism and budget cuts. Employees keep getting thrown under the bus, particularly around the OS orgs. I've heard from close friends in core positions that they have continuously been losing benefits, morale budget slashed entirely down to a single meal twice a year, and that the extremely few leadership that have technical backgrounds, are now glorified Yes Men, because that is how they get fast tracked to promotion.

They have also frozen hirings while outsourcing and allowing hiring to continue in their fancy new India campus.

[-]

Practical-Positive34@reddit

Yeah everyone I know there still is super demoralized most are no longer passionate about anything Microsoft is doing they are just trying to keep their jobs and don't really care about much else...To me that's the sign of a company heading for a very steep nose dive. If you lose the passion of your employees, your done. Especially all the old school ones that even got tattoos of your logos and shit.

[-]

Aaronontheweb@reddit (OP)

I was in Developer Evangelism (working with U.S. West Coast startups) in 2010-2012 under Ballmer trying to pitch Azure and Windows Phone 10.

Azure was an absolute dumpster fire back then, basically unusable. They didn't have Linux support or any real IaaS; SQL Azure's performance was so terrible that the Azure CAT recommended that one of my startups shard a \~2Gb database into 10 partitions in a federation to handle peak demand (mobile game.)

I had thought the foundations of all that stuff had been repaired with the ARM-generation set of services and infrastructure, but these posts are making think a lot of the cracks got papered over rather than thoroughly addressed.

[-]

arpan3t@reddit

It didn’t have IaaS because it was PaaS back then. They were a dumpster fire, but they pivoted to IaaS, and honestly as someone that uses a lot of Azure services today, I don’t have a lot to complain about.

Sure, they could do better don’t get me wrong. Azure Automation feels all but abandoned when they barely support Python 3.10. Front Door is constantly having issues, etc… but it’s not Google Cloud bad.

[-]

RogueJello@reddit

It’s just management was insane. No fixing bugs. Ignoring fundamental features chasing flashy features that were immediately abandoned. Musical chairs reorgs every 6 months.

Sorry, sounds like you can about the company and product, which is a little nutty. How well did the managers do on their bonuses, promotions, and exits to other better paying opportunities?

[-]

megacewl@reddit

It was still that engineering that grew up with the beginning of the tech industry and were forged in fire. Probably everyone understand things at a relatively lower level

[-]

Sability@reddit

as soon as they fired all the test engineers

OOOOOOOOOOHhhhhh, now it makes sense why win11 task bar gives you slightly more icon space when centre justified vs left justified

[-]

cjarrett@reddit

I converted from sdet to sde during that and spent the next five years in reorganization every 9 months and writing a test suite for each new team because none of the devs wrote or ran tests. Hilarious

[-]

rich1051414@reddit

Instead of moving the engineers upward, it seems like they stretched out the job descriptions of the executives downward and marketing sideways.

[-]

sleepDeprivedSeagull@reddit

I work at a different corporation and this goes hard. The amount of firefighting I have to do is insane with SAP.

Then I’m unable to fully commit to my current 4 to 5 projects I’m on, which then creates new firefighting.

Meanwhile the ceo collects their 20 million salary and asks us to “do better” while letting go thousands of employees, and forcing the remaining talented folks out.

[-]

rndmcmder@reddit

This sounds very much like where I currently work at, but I'm a new hire and I have not been properly introduced to our systems and whenever I complain about insane practices to anybody I either get "we don't have time to do this correctly" or from my boss "I thought you could make a difference in your team".

I think of myself as a good developer, but in my current company I feel like I can barely do anything since I have no access to the knowledge that is essential to do anything but the simplest bug fixes.

Time to go looking for a new company once again.

[-]

sleepDeprivedSeagull@reddit

Yikes. Sounds the same.

“Sorry we don’t have an onboarding process and everyone is so overworked that the employees under the most pressure are the documentation. Anyway, here’s some temu water wings, don’t drown in the ocean. Bye”

In other news I have applied for 5 roles today, worked out for the first time in a year and now I’m chillin. Hope you find something soon my guy.

[-]

Ok-Bill3318@reddit

Sounds familiar.

[-]

justme89@reddit

I work here, I know about some of the things mentioned in the article. I think the writer is right but he is a bit too focused on the engineering side of things and trying to fix everything from the beginning.

Even if you worked there, getting hired and then trying to change the core functionality of an entire business division while you barely have one year there is realistically impossible.

You need to figure out how to tackle the problem and from where to start, ideally small. I tried changing some things from the beginning like he did, and it ended up badly. The hardest part if convincing people to follow you and accept your ideas, which is an art of itself.

But from what I have seen, very few people have what it takes to mobilize everyone into a new direction, or just mobilize people in general, and be also good technically.

[-]

Aaronontheweb@reddit (OP)

> Even if you worked there, getting hired and then trying to change the core functionality of an entire business division while you barely have one year there is realistically impossible.

He'd been there for quite a bit longer, but working in different divisions - he mentions this _numerous times_ throughout the articles.

[-]

justme89@reddit

I know, I wrote it in my comment, "Even if you worked there,". It still seems to me like he didn't try first to gain some trust and people to vouch for him before trying to tackle serious things. He seems like a brilliant engineer but not the kind of person to gain following in such a company. He tried too much from the very beginning and annoyed people a bit which made implementing the things he wanted even harder.

[-]

lost_send_berries@reddit

It works both ways though. If his good idea doesn't get buy in it's not just because he wasn't persuasive enough.

[-]

Aaronontheweb@reddit (OP)

I get it, but he was brought aboard specifically for his expertise in that area:

> I rejoined in 2023 as an Azure expert on day one, having contributed to the development of some of the technologies on which Azure relies and having used the platform for more than a decade, both outside and inside Microsoft at a global scale.

Having to placate dipshit middle managers worried about getting their next bonus is exactly what leads to the slop treadmill he's describing.

[-]

FlyingBishop@reddit

99% of your job will be placating middle managers if you want to effect the kind of change he was trying to do.

[-]

DrunkensteinsMonster@reddit

I’ve been there just as long as the author, and there are things in this article that are straight up wrong both in terms of company policy and just architecture. Not sure what they think they’re doing with this article to be honest.

[-]

BobDope@reddit

Would have hit harder if you were still there I bet

[-]

j_lyf@reddit

now you work at tenstorrent

[-]

chicknfly@reddit

As someone who applied to the Core team recently, fuuuuuuuck that. I’m glad it didn’t work out.

[-]

j_lyf@reddit

where did you end up working

[-]

chicknfly@reddit

I accept a role recently, but at the moment I’m not too keen on sharing where. It definitely wasn’t MS tho.

[-]

j_lyf@reddit

Google?

[-]

chicknfly@reddit

nope, but even so, I'd prefer to keep my new role private for a bit.

[-]

j_lyf@reddit

nope

[-]

pilibitti@reddit

onlyfans?

[-]

chicknfly@reddit

I’m a Latino man with a Latina booty — an OF would certainly pay the bills. But no, that’s not the job either.

[-]

wannaliveonmars@reddit

The current plans are likely to fail — history has proven that hunch correct — so I began creating new ones to rebuild the Azure node stack from first principles.

So our messiah decides Azure is garbage, and it's up to him to plan a total rewrite of Azure "from first principles". I'm sure management will be thrilled to find out what he's been spending his worktime on. Let's find out.

...

This vision was met with disdain among lower-level management in Azure Core, who may not have understood the urgency — or the scale — of the changes needed to make the platform truly scalable while lowering long-term OPEX costs.

Wat? Management rejected the total-rewrite of Azure? I'm shocked! What will our messiah do?

On November 19, 2024, I sent a detailed letter to the Executive Vice President of Cloud + AI.

Ok, good, but can't we go higher?

On January 7, 2025 — still months before any public indication of strain in the OpenAI relationship — I sent a more concise executive summary to the CEO.

Finally, he phoned the CEO.

It also noted that I stood ready to help lead a first-principles reconstruction of the Azure node management layer, if given the opportunity in the right capacity.

Surely the CEO will realize his genius and immediately appoint him Head of Azure, with unlimited mandate to do a total rewrite.

In the months that followed, I received no reply — not a single acknowledgment, question, request for clarification, or confirmation of receipt — from the EVP, the CEO, or the Board.

I'm shocked.

[-]

starsfan18@reddit

While I agree with his analysis as a former Microsoft employee, I also found it laughable that he concludes that Azure agents making 10,000 WMI calls per minute (and other assorted icky issues) is the primary driver behind the stock price being down 30% this year.

[-]

AdPsychological7568@reddit

I did not get that impression. The implication is the loss of OpenAI is the primary driver for the market cap drop, and that OpenAI was lost because of what he saw.

[-]

starsfan18@reddit

Agree, I skipped a few steps for pith in my comment. It's a very myopic conclusion, though.

[-]

AdPsychological7568@reddit

It’s one possible explanation. All eyes are on OpenAI, and Microsoft was their proud backer: a huge flair and love story. Now the worst quarter in 17 years and a $327 bn single-day plunge, what happened? The OP says Azure is dogshit and OpenAI left.

[-]

PancAshAsh@reddit

The 10,000 WMI calls per second were specifically referenced as a problem to the nominal reason he was hired, which was to port the management layer to a low power SOC. The example provided was an example of the whole project being a fool's errand being undertaken by people who did not understand the basics of their own architecture and who did not do the requisite research before committing a lot of resources to it. That is the real problem, more than any technical debt.

[-]

Pyryara@reddit

But the issues are just emblematic of the core structural problems at Microsoft. I think it's absolutely difficult to run a company the size of MSFT but if deep technical issues are just ignored to the point where you routinely copypaste commands to manually manage that huge fleet, or routinely give people access to protected data that they should rarely need, that is a huge huge issue. It's not the issue for the loss directly, but it shows there is a very deep management problem.

[-]

starsfan18@reddit

Fully agree.

[-]

dezsiszabi@reddit

Per second.

And I don't think he said that's the sole reason for the drop in stock price.

It was just an example of poor design/architecture.

[-]

AdPsychological7568@reddit

I actually looked and missed the part about “total rewrite of azure”—where does it say that?

[-]

AdPsychological7568@reddit

That is a strong opinion 😅

Did you miss that: Microsoft’s Cloud is a Pile of Shit”—The US Government

https://www.propublica.org/article/microsoft-cloud-fedramp-cybersecurity-government

[-]

AdPsychological7568@reddit

I don’t think this was about rewriting Azure from scratch. The software running on the nodes is critical but only a tiny part of the system.

[-]

SmellSmellsSmelly@reddit

Are you some kind of paid shill?

[-]

AyeMatey@reddit

No. The person you are replying to sounds like a realistic , experienced person. One does not “phone the CEO” (or email them) with a problem, except in fake TV reality shows.

People work in an organization. If you have ideas you need to persuade people around you. There are senior people who make decisions- you need to influence THEM. You don’t call the CEO. If the people around you don’t like your ideas then… you need to go elsewhere. Either you are not right; or they are not right. Escalating 6 levels is not going to change that. And writing the board ? Cmon. Delusional.

[-]

GetSecure@reddit

You are 100% correct.

I see this same delusion everywhere, not just with engineers:

"If only our great leader knew of the problems, they'd do something to fix it."

You look around and all you see is shit, but the leaders are different... It's a mental coping strategy, to believe otherwise would leave you with no hope.

[-]

rmrf_slash_dot@reddit

Or also “If everyone would just recognize how amazing I am and that I’m always right, everything would just suddenly work right!”

The writer of the article definitely identified some serious issues but his ego features so prominently in the series that it’s hard to read seriously.

[-]

GetSecure@reddit

I don't see it as ego. He's clearly thought hard about it, and like all good engineers, worked out what the bug is. He's looking at it logically. He was there at the start, he knows how it used to be, saw it go to shit and investigated why. Most other people don't have that much luxury, it was always shit for them.

His solution is lacking the human element, the executive and managerial experience.

[-]

Outside-Storage-1523@reddit

You can't save things at this stage using "normal" politics. OP's method is the only one that might work, but yeah as you said it won't work. Reform rarely worked in history for that reason.

Same that talking about getting a union for software developers is almost a joke.

[-]

barsoap@reddit

So our messiah decides Azure is garbage, and it's up to him to plan a total rewrite of Azure "from first principles". I'm sure management will be thrilled to find out what he's been spending his worktime on.

Nope, you got that wrong: The Messiah (co-)wrote Azure core infrastructure, saw that it was good, left, returned, and could not believe what kind of nonsense had been duct-taped to it in the meantime.

So, as a senior and core engineer, he went ahead and hashed out a plan how to remove that duct-tape and make everything reliable again. Literally job description. Or do you think software architecture just happens out of nowhere?

[-]

OdderG@reddit

I kinda take it the same way. He has the creds to make the claim for Azure, because he was there when it was created. He laid out his reasoning clearly that the development is drag down by an insane amout of live fixing that will keep growing if the core is still a mess, so he intended to fix the mess incrementally so that it will work better in the long run

[-]

wannaliveonmars@reddit

About his general bitching about code quality, I trust him. But a wholesale rewrite of production code like that is a great way to bankrupt a company. Any manager worth his salt would have rejected that proposal, it's bonkers. Rewrites like that never work, when you have a massive codebase like Azure you just work with what you have, and try to improve it incrementally, which is what they're probably doing anyway.

[-]

OdderG@reddit

His proposed approach seems reasonable, isn't? He suggest that, gradually, a small component will be chosen for isolation and replaced, easily tested and integrated hack to the core one by one. It's a huge task, but already broken down into multiple small tasks. His goal is to reduce firefighting and his plan seems to be reasonable to me.

[-]

wannaliveonmars@reddit

The Messiah (co-)wrote Azure core infrastructure

In his own words. I know a guy who likes tooting their own horn like that, and is what you'd call an "unreliable narrator". He routinely boasts about how he's "saving the company" by fixing this or that bug, when in reality he's just doing his job.

If he's anything like my guy I know, "co-wrote" could mean just "added a few lines of code here and there", or "submitted a few bugfixes". It could mean anything, or nothing at all. I would be more interested in getting an opinion from his colleagues about his skill and importance than from him.

Job-hoppers are good at resumes but I've seldom seen them do a lot of actual useful work - that's usually for the less ambitious "fixtures" who have stayed for 10 years in the same company.

[-]

GimmickNG@reddit

This sentiment seems to be rooted in a lot of biases. Don't get me wrong I have an equal disdain for pompous braggarts but the logical conclusion of what you're implying is that you should just keep your head down and say nothing. Which is kinda how they ended up in this mess to begin with.

[-]

seven_seacat@reddit

This was the distinct impression I got from the series too, lol. The hero coming in to save the day to rewrite everything from scratch - without getting everyone else on board first.

[-]

gjosifov@reddit

How dare they!

Those same C-level or VP level when people are switching to Linux

[-]

DrunkensteinsMonster@reddit

Glad I’m not the only one who had this take away

[-]

usernamedottxt@reddit

While I believe a lot of he wrote, I picked up on the same. The actions were so self centered even if the problems were systemic.

[-]

RandomNumsandLetters@reddit

As an ex, super smart talented people working there spending half their time being poorly micromanaged by upper leadership

[-]

OriginalTangle@reddit

Ok but why doesn't this lead to a catastrophic failure for the entire enterprise? What would it take for something like MS to go bust?

[-]

maikuxblade@reddit

Linux is gaining steam the last few years but largely M$ has a near-monopoly on the gaming space and corporate spaces (any computer touching job that isn’t actually IT) and it’s been that way for decades. They could probably fail for quite a while and be fine

[-]

PancAshAsh@reddit

Unironically no, they do not hold anywhere close to a monopoly on the gaming space if you look at gaming as a whole. The vast majority of games work fine using Proton/SteamOS and they have 0 relevance in the (very large) mobile space. The only place where Windows is required is AAA multiplayer games, and even then not all of them.

[-]

RGBrewskies@reddit

ibm is still around, shit

[-]

barsoap@reddit

IBM retreated to its core business, mainframes, and abandoned the PC market which was always too nimble for its taste. IBM might not be as big any more, but they know how to steer a tanker, the exact opposite of move fast and break things. MS' core business at first was operating systems, then shifted to office, then, critically, shifted to cloud-based office.

Which means that their current core business is exposed to that burning house of cards that is Azure. Valve is attacking Windows from the gaming side while shoehorned AI nonsense is actively driving end-consumers away from it, all kinds of places running managed environments, especially governments, especially ones that aren't the US government, are moving away from the OS because if you just need something to run a custom business logic interface, a web browser, and an email/calendar program then guess what, Linux can do that.

Which means that their past core business is also in heavy shit.

What's left? Support contracts for legacy Excel sheets running just as tech debt ridden companies? At that point IBM might just as well buy up what's left of MS they're good at that kind of thing.

[-]

JJJSchmidt_etAl@reddit

Even though I do not use linux right now, I would be so overjoyed if Valve could get some motion on steamOS. The world would be a far better place with another good option for PC gaming.

[-]

barsoap@reddit

SteamOS isn't meant for general PCs but handhelds and consoles and will likely not even boot on your PC. Valve created it as something to ship with their devices, not as an OS for general use.

Literally any Linux distribution can game just as well. In case you have an nvidia card make sure you have proper GPU drivers (nvidia's drivers aren't open source so they usually need an extra hoop to jump through), then install steam, log in, done. Not really up to date on the current meta but I've heard good things about Bazzite when it comes to being beginner-friendly and gaming-focussed (but not overcommitted).

[-]

EveryQuantityEver@reddit

Inertia. They have many large customers on multi-year contracts.

[-]

Nowhere_Man_Forever@reddit

I legitimately think that the cruz of the issue is that the American public (and to a lesser extent, everyone else) are completely unwilling to accept even slight inconvenience in order to hold anyone to account. Windows gets worse and worse every year while stealing more and more data from you and putting fucking ads in the start menu? "Oh well, what can you do?" Instead of learning how to install and use Linux or even Mac. Artemis II has to call home to try to troubleshoot two different versions of Outlook being installed that both don't work? "Oh well, Outlook is too deeply entrenched to replace." It's a culture of laziness and indifference on a massive scale. Until people decide to start giving a shιt, nothing will be fixed

[-]

CreationBlues@reddit

If only personal solutions to systemic problems worked!

[-]

RandomNumsandLetters@reddit

It's a giant company, they can't win them all haha. I'm sure they'll be fine for a long ass time on the back of azure + legacy windows momentum. Would still love to see them change course a bit though

[-]

Simple_Woodpecker751@reddit

they wont

[-]

Kendos-Kenlen@reddit

Because migrating is costly and it needs a lot of shit for a company to seriously consider migrating all their infrastructure to another provider.

Microsoft’ firefighting and approach to issues may mitigate the disappointment of average customers, who do not feel it is the shit it actually is under the hood.

[-]

Korona123@reddit

I think it's one of those situations where it's just too big to suddenly die. It takes a thousand cuts and slowly bleeds out over decades.

[-]

Practical-Positive34@reddit

As an ex-microsoftie this isn't suprising. They make some of the dumbest freaking decisions ever. Was one of the reasons I left. Just absolutely braindead detached from reality decisions. I was there during the whole Silverlight Azure era, so was a while back that many don't even remember lol...yeah Azure started off with a Silverlight UI believe it or not. It was worse than the current UI which is pretty hard to do really.

[-]

localhost8100@reddit

I am a mobile Dev. The Xamarin / Maui framework is such a cluster fuck. Microsoft people believe it, they build a product, they make money. Eventually Microsoft drops support or its good enough for basic operation (can't scale the apps). I get hired migrate to Native. My whole career is this lol.

[-]

MrDrummer25@reddit

Really is telling when the Teams rewrite isn't using it.

[-]

ososalsosal@reddit

Fuck man.

.net android is pretty decent I must say. Pretty much just straight android dev without the need for java.

It doesn't look to be getting enough love in the jetpack era though so I can imagine Google deprecating the XML views any day now and leaving the csharp android world in the lurch

[-]

mordack550@reddit

And there’s me, working on NET8+ WinForms in 2026, i’m happy and my customers are happy, and I don’t see why I should switch since the frameworks is still being maintained

[-]

UndeadMurky@reddit

WPF is a decent upgrade from 2001 to 2006 technology! Just don't touch anything after that

[-]

meischtero@reddit

Sales are not happy because it is not new, hot stuff.

[-]

Practical-Positive34@reddit

I worked on Xamarin back when it first started before it was acquired by Microsoft!!

[-]

metahivemind@reddit

I knew Miguel before it became Xamarin. He thought he was the shit and surrounded himself with people who kept jerking him off. He was genuinely good, but also, he was replicating what had already been done rather than being innovative.

The only other person I've seen like that was Sebastian McKenzie from Babel, who also crashed and burned when the runway ran out on their self perceived genius.

[-]

danstermeister@reddit

Sounds like xobni

[-]

jeremiahgavin@reddit

I worked on a xamarin/dotnet native app for a bit, and while it was honestly cool how much you could do with c# on an Android or iOS device, we saw serious performance issues with decently complex UIs. Seemed like the bindings to native UI code had a lot of overhead. AOT support made iOS pretty dang good though. Android still doesn't have it as far as I know.

[-]

the_tip@reddit

Oh I remember, having to use IE instead of edge/chrome for the XDS web interface, ridiculous.

[-]

dvidsilva@reddit

Azure for startups would give away like tens of thousands of credits and we wouldn’t use it

Good way to get free Microsoft keys for a time tho

[-]

Eric848448@reddit

Silverlight! There’s a name I haven’t heard in a long time.

[-]

tekchic@reddit

Wow, yeah. I'd blocked that one out of my developer brain.

[-]

NenAlienGeenKonijn@reddit

I absolutely loved working with silverlight. WPF in the browser? Yes please!

[-]

z960849@reddit

It makes me feel like everything else is just barbaric

[-]

austinturner01@reddit

Around that time I wanted their slogan to be "Build your new app on Azure, we can be in Beta together"

[-]

Practical-Positive34@reddit

As a massive fan of Mark, and Mark is such an incredibly good engineer it really bums me out that Azure didn't turn out all that great. I mean it makes them money so I guess it's a success but it's just not very good.

[-]

gnuban@reddit

I'm currently in a company with close relationship with Microsoft, and I facepalm every time there's a town hall. It's getting unbearable.

[-]

Worth_Trust_3825@reddit

ugh. They were adamant against terminal interfaces weren't they?

[-]

ineffablol@reddit

I work with one of their cutting edge products and their API is filled with bugs. In some cases, their documentation has code samples that don't even run. I'm pretty sure no one knows because no one is actually running the code before they publish the docs.

I think the expectation is supposed to be that eventually someone will come along, run the code, notice that it doesn't work, and report the bug(s) so they can fix it. So there's no problem with them shipping buggy code or code that doesn't work. But I'm not doing that.

[-]

spareminuteforworms@reddit

Its agile, its jugaad, I'm not sure I see the problem.

[-]

vitek6@reddit

Yes. If you make shit ton of decisions some will be bad. That’s how it works.

[-]

SquishTheProgrammer@reddit

I haven’t heard of silverlight in a long time. I still use WPF daily though.

[-]

Paradroid888@reddit

I'm sure most of my imposter syndrome comes from it being so difficult to get data out of the Azure UI.

[-]

Practical-Positive34@reddit

Nobody knows everything, even they try to make you think they do...

[-]

Paradroid888@reddit

Thanks. It's just the only system I've used where I've found an error in a log, posted it to a chat thinking it's what someone needs to look at, only to realise it's from 3 days ago. It's just so upside down to navigate. Good enough for sales pitches though!

[-]

FlyingBishop@reddit

Agree. Every time I log into the Azure portal I want to burn Microsoft to the ground. I think if I worked at an azure shop I might quietly migrate things to AWS or GCP without even asking.

[-]

prescorn@reddit

I think I had suppressed that memory. Early azure services were a mess as a customer. Worker and web roles, appfabric, etc.

[-]

Practical-Positive34@reddit

I've tried to suppress a lot of my Microsoft experience tbh...

[-]

keyboardhack@reddit

Last year i started writing down all issues i have encountered with azure and the duration of those issues. It is actually insane how many issues you encounter in azure when you start to use it just a little bit. A few issues we've encoubtered over the past year * zombie resoures that can not be deleted. Regularly happens with key vaukt and storage accounts. * vm provisioning not working at all in aks(azure kubernetes service) that was a fun one. * azure restoring our aks from a backup but without any of our services running except stateful sets..... * connection in aks between pls(private link service) and ilb(internal load balancer) just dying out of the blue. * azure deleting some of our production resources. No incident report from them on this. They only told us they were sorry after we created a service request about it.

I have a lot more and it is just sad. Actual time for us without any azure caused issues is <90%.

[-]

gimpwiz@reddit

Damn, imagine not even getting a single 9 of reliable uptime on your service.

[-]

renatoram@reddit

Github sure can imagine that. Heck, they don't have to imagine (they're down to 89.91%).

[-]

gimpwiz@reddit

Amazing. That's over a month per year of downtime.

[-]

OffbeatDrizzle@reddit

they want you to spin up 2-3 instances for resilience / high availability, that way they can charge you triple.... ayyyyy

[-]

Ileana_llama@reddit

i cant believe that small niche cloud providers are more reliable

[-]

keyboardhack@reddit

That's to be expected. You generally use a large cloud provider a lot more then you would a small cloud provider simply because the large cloud provider is capable of much more.

There is probably also a lot more competition between small cloud providers. Small bad cloud providers goes out of business leaving the good ones. That unfortunately doesn't happen with large cloud providers.

[-]

FullPoet@reddit

connection in aks between pls(private link service)

Haha is this still happening? I remember this was happening in the preview.

[-]

Martin8412@reddit

That you need private link service to begin with is insane. How did they decide that having everything on public IPs by default was the way to go? It seems like a way to make customers spend more money. Why can’t I deploy all core services into virtual networks with all private IPs?

A lot of their services require you to buy the more expensive tiers to make your security team happy.

Not to mention that their VM performance is garbage. I don’t know if I’m just unlucky with my neighbors on the hosts I have my VMs on, but the performance is atrocious.

[-]

Empanatacion@reddit

Interesting insights, but if I'm looking to hire a principal engineer and read this airing of dirty laundry, it's giving me doubts about wanting to work with the guy that wrote it.

It probably feels really good to write it, but it also seems petty.

[-]

echoAnother@reddit

Counterintuitively, someone with such ownership and work ethics, such to "air" the dirty laundry, would make me want this person as a coworker or as an employee.

[-]

HotlLava@reddit

To a point sure, but I think once you start writing emails to the board of investors because the CEO didn't respond to your emails (see part 6), that goes a bit too far.

[-]

grauenwolf@reddit

Would you prefer that he allowed incompetent middlemen to hide serious issues from you?

[-]

Kered13@reddit

Agreed. The article was all very good except that part. No reasonable person would expect a response from the CEO at a company like Microsoft. And then escalating to the board? How does he think these companies operate? I understand that he had good and justified intentions, not that's just not how large corporations operate in the real world.

[-]

planetworthofbugs@reddit

He lost me when he started rewriting Azure by himself…

[-]

teknikly-correct@reddit

As it turns out, boot licking actually isn't a great attribute in an engineer. I'd love to work with this guy because I appreciate someone who is willing to break up groupthink.

Software benefits from dirty laundry being exposed early and often, otherwise you get a bunch of bootlickers congratulating the cross-functional team on delivering "the best software ever" on schedule and under budget. Meanwhile the customers just quietly walk away.

[-]

planetworthofbugs@reddit

Yep, good on him for calling it out. But when he started rewriting Azure by himself, that’s when he became that guy I don’t want to work with anymore. I’ve been in the same situation he was in, but we got the devs in agreement, approached management with a plan, and then designed and implemented the replacement system together. If that hadn’t of worked, I would have left long before I started spending someone else’s money rewriting it myself.

[-]

zappini@reddit

I've rewritten plenty of stuff. Most times as a sanity check. Even just a proof of concept can help calibrate my own position.

[-]

gjosifov@reddit

it's giving me doubts about wanting to work with the guy that wrote it.

Microsoft is a key company for the world economy
and I'm sure you want a dirty laundry to be address ASAP, because you like or not, your bank account maybe compromised or your basic utilities or your medical record or ...

Microsoft isn't some small shop selling tourist souvenirs, so airing a dirty laundry is such a drama
we are talking about major economical problem if the culture at Microsoft doesn't change in the next 5 years

[-]

Outside-Storage-1523@reddit

That's the why the world is worse than it should be -- engineers are the ones who have the real power here, but they chicken out and handle the power to the wrong guys.

Yes, I chicken out, too. But I would applaud others who can stand up.

[-]

RagingCalmness@reddit

I read through the whole thing, I'm in the org he was in, and while he sheds light on real issues, his recommendations are a bit unrealistic. He wants to not do billion dollar business deals and focus only on improving quality. When faced with situations like this, good leaders try to balance both through hiring more etc.

The complexity of this can't be overstated, he himself says it's like changing an airplane engine mid flight. There are some 20-year old legacy code still running in Azure. The direction to use Rust was an attempt at modernizing those code. People are trying but the market forces, pressure from investors etc are giving tough deliverables and we end up in the mess he's talking about.

[-]

torrent7@reddit

Nah, that's a false choice. MSFT has/had the money and can acquire the talent to do both. It is a matter of the willpower to hire/pay talented people, and conviction. Microsoft has always underpaid compared to the other major tech companies (sans Amazon) and under staffing is a chronic problem from basically everyone I ever talked to.

I mean, its fine to knowingly make the choice that you're going to go fast and break stuff, but you need to know the trade off that this method will come around to bite you in the long run. I don't think anyone in leadership knows the true cost of these problems.

I don't read it as him being like "we have to stop all new azure deals". I read it more like "if we don't invest in quality, this will significantly hurt us long term".

[-]

anticipat3@reddit

I still look at the Windows Phone debacle as the point where anyone with eyes could see that the emperor had no clothes. No other company in the world was better positioned to create an iPhone competitor, no other projects could possibly be as important to the long term success of the company, and they completely dropped the ball. When all the money, personnel, and expertise in the world can't deliver a competitive product, who else is to blame if not corporate leadership?

Sloppy Nutella will get the boot soon enough, but I think the damage is irreparable. Microsoft is clearly on the path of IBM -- still around in the corporate world, but a relic of the past to consumers and perpetually shrinking. The AI bubble is the only thing propping up the share price, every other segment they're in -- not just Windows, which is the most visible -- has become a dumpster fire of AI-generated proportions.

[-]

gimpwiz@reddit

It's a funny story, the windows phone, isn't it?

They were sort of the original smartphone OS via CE, apart from the likes of Palm, before smartphones were really even a thing. I mean CE was dogshit but it was everywhere.

Then they make windows phone... 7... 7.5... 8... in pretty stupid ways. Holding a funeral for the iphone on release? Embarrassing as hell. But ignoring the stupidity of a bunch of their choices, by the time it got to 8, it was a totally usable product. Adequate hardware, adequate software, adequate ergonomics, camera, wireless performance, battery life, size. Low cost, honestly, versus the competition. People who used it were pretty happy... well, they told us they were happy and then six months later chucked it and got an iphone or an android but you know how it is.

And then rather than addressing the shortfalls, like lack of a legitimate app store filled with programs that are legitimate (vs scam knock-offs) and work properly, which would have taken them like three or five years but god knows they had the resources for it, they just folded like a wet paper bag. Why? Ran out of patience, like two years in?

Then the whole fiasco with nokia really didn't help.

Remember how much everyone was worried about a Windows monopoly in the mid to late 90s through the early to mid 2000s? Smartphones came out and microsoft somehow wasn't able to even be a player. Incredible.

[-]

grauenwolf@reddit

WP8 was a great product. Far better than modern Android in many ways.

Windows Phone 10 was a buggy piece of shit.

I wonder if that coincides with them firing the QA department.

[-]

LowerEntropy@reddit

I remember going to a workshop/presentation by Microsoft. I was blown away by how nicely VS was integrated with CE, and how easy debugging and running an application was. And, MS was still run over by Apple, who released a phone that didn't even have apps or a way to make custom apps when it was released.

[-]

MrSunshineAndPuppies@reddit

> I was blown away by how nicely VS was integrated with CE, and how easy debugging and running an application was.

The compilers, debuggers/VS integration, etc, was done by an external company named BSQUARE.

[-]

float34@reddit

It is Microsoft as usual - good ideas, sometimes (frequently?) ruined by bad management decisions.

[-]

GimmickNG@reddit

Even before that there was the Zune.

[-]

heroyoudontdeserve@reddit

I don't think anyone in leadership knows the true cost of these problems.

This is the part I struggle to wrap my head around — surely at least some of them have come up through the ranks and should get it?

[-]

torrent7@reddit

If your time horizon is only until next bonus season, you are actively penalized by bringing up these types of systemic problems

[-]

b0w3n@reddit

Microsoft has always underpaid compared to the other major tech companies (sans Amazon) and under staffing is a chronic problem from basically everyone I ever talked to.

They built a whole fucking diploma mill over in Hyderabad to fast track H1Bs for christ's sake. "We can't find local software devs, so we're going to bring over 'dotnet experts' and pay them 1/3 of what we should" sort of shit.

Once the old guard like Greenburg, McDonald, and even Gabe Newell were out, everything slowly went to shit. Sataya is just the diarrhea on top of the giant pile of shit that's been building up for the past quarter century at MS.

[-]

sabchint@reddit

Everything you said is blatantly incorrect.

Microsoft hasn't 'fast tracked H1Bs' or even sponsored them for their employees in the Hyd office for several years now

Newell left Microsoft in 1996, so why that has any bearing on the current state beats me. Greenberg otoh left in 2021, so what are you really trying to say?

Yes, Microsoft has absolutely lost their way in the last couple of years - but what's with the misplaced rage?

[-]

phillipcarter2@reddit

Yeah, assuming it’s even real, this guy sounds like the kind of engineer who only sees technical systems doesn’t grasp nor care about business realities.

Insane stuff happens at big tech. I was witness to some of it myself! But also a ton of extremely good work, at a scale nobody else can accomplish.

[-]

sionescu@reddit

It seems you're the one who doesn't grasp that if the technical problems aren't solved, the business reality is that big deals aren't going to happen.

[-]

Sea-Specific-6890@reddit

So... unfortunately this isn't true. I once thought this way, time in the industry, including at the company in question, has taught me otherwise. You can demoware and "that'll be fixed in the next release" your way to quite a successful business. Enterprise customers are resistant to change. As long as it kind of works, even in a hacky, less than ideal way, they will stick around as the cost of leaving you for someone else is greater, both financially and organizational politics wise. Satya has shoved Microsoft down the throats of the Fortune 500 and saved the company from the death it faced under Ballmer. How that was done has everything to do with business and marketing rather than engineering. Things working "just barely good enough" is actually sufficient to make billions. Engineers often struggle accepting this reality. I know I did. But here we are. If we are objective rational persons, we must admit Microsoft since Satya's take over is objectively successful by the standards at which for profit corporations are measured. Stock price, valuation, revenue. By engineering standards, Gods no. But that's not how business works.

[-]

sionescu@reddit

You're underestimating how much the reputation for unreliability is harming Azure right now, and will continue in the long run. I agree that there are lots of Fortune 500 companies with a legacy Microsoft-based infrastructure that will keep using it, but there are many more companies, both within and outside the US, that won't touch Azure with a 1-mile pole.

[-]

Sea-Specific-6890@reddit

People have been saying this for over a decade and it's time and again not reflected in the measures by which publicly traded companies are evaluated. Oracle is another one where we keep saying this same thing. Business, marketing, sales, they do not logically tie to our understanding of good and bad as software engineers. Microsoft is entrenched. I don't have the vocabulary to even articulate the depth to which their products are rooted in the operations of the Fortune 500. And for all their flaws, the stuff mostly works for companies that aren't tech companies and just want to run their business. The drop off in reliability you're forecasting is a loss of one of the 5 9's, and I'm saying that won't be enough to undo the hooks Microsoft has in companies and governments worldwide.

We want to believe "Microsoft sucks just wait everyone is going to leave them and I'll be proven right" but I can't emphasize enough how far that view is from reality. Enterprise licensing, government contracts, sales, compliance, that stuff is just as complicated as software engineering in its own way. Maybe not in a technical skill way, but it's complicated. Microsoft has mastery over it, they're not going away. IN many cases customers literally cannot leave them on account of the contracts they've signed.

[-]

sionescu@reddit

it's time and again not reflected in the measures by which publicly traded companies are evaluated

OpenAI dumped Azure because they couldn't deliver and since then MSFT is down 30%.

[-]

EveryQuantityEver@reddit

Did they? Or did Microsoft decide they didn't want to keep catering to OpenAI because it was a losing proposition?

[-]

Sea-Specific-6890@reddit

Microsoft has a deal with OpenAI by which they literally, legally, cannot use any other cloud provider except that Microsoft approves it. Microsoft let them leave Azure because of the many issues they were having, as the original article highlights, supporting their capacity.

OpenAI can now jointly develop some products with third parties. API products developed with third parties will be exclusive to Azure. Non-API products may be served on any cloud provider.

Read this https://blogs.microsoft.com/blog/2025/10/28/the-next-chapter-of-the-microsoft-openai-partnership

Fluctuations in the stock market month to month don't disprove anything I said in my previous message. The way Microsoft has ensured its own success is not through software engineering. They have OpenAI locked to them, they have a 30 billion dollar deal with Anthropic and use their models in Microsoft Copilot now.

Anthropic has committed to purchase $30 billion of Azure compute capacity and to contract additional compute capacity up to one gigawatt.

https://blogs.microsoft.com/blog/2025/11/18/microsoft-nvidia-and-anthropic-announce-strategic-partnerships

They're not going anywhere. Seriously. We are not talking about software engineering here. We're talking about all that stuff they teach in MBA courses. It can infuriate us all day long but if we are rational and logical we must admit they excel (no pun intended) at making a lot of money with subpar software.

[-]

phillipcarter2@reddit

Incorrect.

[-]

OxfordTheCat@reddit

Open AI's most optimistic forecasts have them posting more than $100 billion in losses in the next five years.

Azure will still be plugging away two decades from now in some form.

[-]

phillipcarter2@reddit

Just to add to this, the software is like an iceberg. The things most people in forums like this complain about are the things above the surface. All the things that actually matter and why the software gets picked? Below the surface.

[-]

OxfordTheCat@reddit

Azure has had issues with reliability from day one, and all that has happened in the nigh on two decades is widespread adoption and everyone falling all over themselves to dump Server.

Anyone that can feasibly be on Linux is already there, and was there ten years ago.

[-]

phillipcarter2@reddit

Incorrect.

[-]

phillipcarter2@reddit

Incorrect.

[-]

Outside-Storage-1523@reddit

Yeah, assuming it’s even real, this guy sounds like the kind of engineer who only sees technical systems doesn’t grasp nor care about business realities.

Engineers-turned-Business-Yes-man are the worst kind of engineers.

[-]

yojimbo_beta@reddit

The business reality is that MSFT have lost two Golden Geese here: OpenAI and the general stewardship of their cloud platform

I don't want to aim this at you specifically, but there's a certain kind of developer, who thinks seeing the technology as a means not an ends is a big insight, and that all of us talking about architecture, quality, system coherence are just artists and philosophers detached from reality. Actually, we already understand that the company needs to make money - we are not five years old - we are screaming at management (and those "delivery-oriented" tech leads) about the icebergs about to sink the ship

I have now been at three orgs that have lost their market dominance (and in one case, a near monopoly) due to simple technical debt. So I take a dim view when someone tells me tech debt "is a nice problem to have".

[-]

ungoogleable@reddit

If you're in a senior technical position, it's not enough to come up with the right technical solution, you also have to communicate the business value of that solution to management in a way they can understand. If they're not convinced, you can call it organizational failure, but you are part of the organization too.

[-]

phillipcarter2@reddit

I'm not framing this in any way, but I am claiming an attribute of the author. He is unfortunately extrapolating his experience witnessing some shameful engineering managers and directors making some poor decisions and thinking that this is how it all works. It's not.

Microsoft trapped OpenAI into a deal that OpenAI never liked -- 20% revenue sharing -- in addition to an attempt to lock them into datacenter and Azure services use beyond this share. That they struggled with their biggest customer ever, something all major clouds do -- my previous company was the #2 customer of Lambda and you can bet your ass AWS had problems with us -- is not particularly unique. The problem always has been, and always will be, that OpenAI needs to keep 20% of its revenue, not hand it over to Microsoft in perpetuity.

[-]

Intelligent_Thing_32@reddit

It mostly always seems to be an excuse of ‘business realities’…. it begins to sound more & more like most c-suites are completely unsuited for their roles.

[-]

phillipcarter2@reddit

Incorrect. Although I have certainly come across some c-suite types who have made what I felt were questionable decisions as well.

[-]

Intelligent_Thing_32@reddit

“Some”…. lol

[-]

phillipcarter2@reddit

Incorrect.

[-]

gimpwiz@reddit

I get your point because we all have limited hours in the day and we need to pick whether we're working on or .

The flaw usually is in presentation. If you go to your boss's boss's boss's boss and say I don't wanna ship features to acquire large contracts, I just wanna make code better, they're gonna laugh at you.

In this long-form 6-page essay, I think the author wrote essentially that he spent a lot of energy telling everyone that if they don't make code better, they won't be able to develop the features that the marketing/business/contract folks are selling to customers in order to get them to commit to a large contract. As supporting evidence he presents a huge customer divesting out of their cloud, and he presents an impact on stock price due to that happening.

[-]

Globbi@reddit

This is something that doesn't have a clear answer was is, and even in hindsight what was, a correct decision.

There is a point where an engineer is right that things are breaking and need fixing/rewriting. And if your boss is wrong, going to boss's boss's boss might be last chance (very small one) at fixing things.
We we will never know alternative timeline — not promising new features, not lying about reliability, working on solid foundations. If this happened, would Azure be making more money? If people agreeing with just making robust Azure services were in charge, would they even support OpenAI? Maybe Sam Altman would be out of OpenAI when he was being fired for constant lies 2 years ago, as support from Satya was an important factor (OpenAI could not just go through with firing SamA, as it would lead to OpenAI folding and being re-created as division of MSFT. This would be even worse for all purposes that the board had.) I'm not saying it would be a bad world for MSFT or otherwise, just that a lot more could change.

[-]

gimpwiz@reddit

For sure, once you've got yourself in such a pickle, there's no obvious answer. Maybe chasing stability and bug reduction means your competition surges ahead and you're left holding the bag. Maybe chasing new contracts gives you the revenue to hire people to fix things. Or maybe the exact opposite.

[-]

josluivivgar@reddit

right but then shouldn't microsoft be doing both the slow improvements while still trying to make the deals with another team?

instead they laid off a bunch of people which makes it functionally impossible to make improvements without hurting your bottom line, but also not making improvements will hurt your bottom line (just not at that instant)

[-]

premnirmal88@reddit

I work at Microsoft and I agree with this, hits hard

[-]

Aaronontheweb@reddit (OP)

> He wants to not do billion dollar business deals and focus only on improving quality

I don't know how you can read the posts and walk away with that interpretation - it sounds like his entire point was not addressing quality lead to the loss of \~1T worth of deals, which is what he wanted to do.

[-]

1esproc@reddit

yeah lmao - what? Guy was literally grounding this in losing OpenAI's business to Core Weave directly due to Azure instability, and the issues he knew needed to be addressed to stabilize it.

[-]

usernamedottxt@reddit

A trillion in market cap, not deals.

[-]

DrunkensteinsMonster@reddit

Right there with you, there’s also a lot of stuff in here that’s wrong or at least pretty misleading when considering it’s addressed to an audience that does not have any context except the context they gave. I’m not even sure what the point of this article was outside of axe grinding or showing people how very important and proactive he was.

[-]

Izikiel23@reddit

Better to lose 1B than 1T I think.

[-]

KevinCarbonara@reddit

He wants to not do billion dollar business deals and focus only on improving quality.

Did the billion dollar deals succeed?

It sounds to me like he doesn't want to fail billion dollar deals, and that he wants to fix the underlying tech issues so that they can later succeed at those billion dollar deals.

[-]

ChadtheWad@reddit

I don't see him saying that at all -- each of their stories talk more about how teams were totally out of their depth, domain experts were ignored and eventually pushed out by middle management, and tight deadlines forced compromises on reliability and security that compounded into deeper issues later on.

Incidentally, as someone who worked at AWS for ~5 years, I can see a lot of the same corporate culture there. Biggest difference I'd say between the two is that the fundamental AWS services are all very reliable, and that makes it way easier for subsequent projects to make stupid decisions without as many overwhelming issues.

Agree that perhaps the author may not understand Microsoft's goals here. Having used all three cloud providers, it's fairly clear to me that Azure isn't a production-ready service and if I'm forced to use it, I'm going to have to double or triple timelines. Ultimately Microsoft makes money because they controlled the IT space for nearly two decades and thrive on their vendor lock-in... as well as the massive number of credits they're handing out just to get people to try Azure. If the solution to keeping those customers comfortable is to hire a bunch of warm bodies to generate a constant supply of buzzwords that VP's can proudly repeat at Ignite, I don't really see an issue with that. They're sacrificing their long-term relevance for short-term gains, but honestly that doesn't matter too much since even the biggest companies don't last forever.

[-]

ChickenOfTheFuture@reddit

That's not at all what was said in the article.

[-]

BlueGoliath@reddit

and they really contemplated porting Windows to Linux to support their current software.

Nice April fools joke?

[-]

rep_movsd@reddit

That may be the only way windows survives

Build a UI and emulation layer on top of Linux.

Enterprise windows is a dead horse. Desktop windows doesn't have any use for the fancy kernel stuff

[-]

OffbeatDrizzle@reddit

wine already exists

[-]

FauxLearningMachine@reddit

Ouch, I have been through "Emperor's New Clothes" problems like this before so this resonates with a deep cringe inside of me

... But never anything like on this scale. This was actually so painful to read. Sorry to OP for having to have lived through this.

[-]

LaconicLacedaemonian@reddit

I'm deep in this right now with a team seeking to rush out a change to put out a fire that exists because of their failure to deliver.

because they have too much operatioal burdon they are putting out a risky change to avoid 2 eng of ops burdon. The change risks tens of millions. It's an uphill battle to block them.

[-]

jfp1992@reddit

Mate just open the flood gates and let ms vibe code everything without QA. Eventually they'll have to get their shit together

[-]

Aaronontheweb@reddit (OP)

there's several more parts to this OP published later too

[-]

Ytrog@reddit

Damn. I'm now reading part 4, but I'm seeing why I had a hard time keeping up with the platform. The moment I thought I understood something it was already deprecated and a new thing was the way to do it. Rinse and repeat. 😟

[-]

GregBahm@reddit

I read to part 3 and so far it's just been a bunch of tedious bitching. I'm successfully baited by the click bait title, and the subject is germane to my interests. But so far all the writer has only communicated to me that he's losing a fight against his own ego.

[-]

FullPoet@reddit

Yeah its really long winded and I only go to part 3 before giving up.

Writing does not seem to their strong point.

[-]

sumwheresumtime@reddit

so your complaint here is actually a lack of attention span on your part made the article unappealing?

[-]

FullPoet@reddit

Ironic your reading comprehension is this poor.

I love drama - OPs writing is so poor and unnecesarily long winded maby of us struggled and some gave up.

[-]

Sea-Specific-6890@reddit

Disagree. Not sure how much you understand the technical details behind what he's sharing but as someone that does, this is not just "bitching".

[-]

GregBahm@reddit

Sam Altman, OpenAI’s CEO, declared that “Advanced AI systems require reliable compute, and we’re excited to continue scaling with CoreWeave so we can train even more powerful models and offer great services to even more users” — words that landed as a pointed comment on Azure’s reliability and scalability.

If my stated objective was to come up with tedious bitching, and I would be up all night to come up with whining this lame.

I get that the author of this post was fired from Microsoft and so he's all mad about it. But I can't imagine wanting to work with such a drama queen either.

[-]

Tiagoxdxf@reddit

Any tldr?

[-]

cogeng@reddit

Same old story. Large complicated system (Azure) with a million crusty layers of technical debt is wildly insecure and barely functional. Management completely clueless. Eventually it starts visibly falling apart and big customers (OpenAI) leave, leading to layoffs.

A lot of the article is detailed accounts of technical and organizational disasters. Pretty damning stuff.

[-]

Ileana_llama@reddit

dont forget teams made of mostly jr engineers working in a very soecialized field.

[-]

OffbeatDrizzle@reddit

it sounds like your typical corporate greed / incompetence / code rot story - it is not unique to microsoft, it's just that they're bigger and more important

[-]

ldrx90@reddit

They had so many issues with their 'autonomous' systems that they had to pay people to be on call and manually go in and fix issues. They called it "Digital escort strategy".

That should give you an idea of how fucked things have gotten. Basically, they cut corners, rushed things, people who originally built it all are gone and now there is a culture of firefighting and over promising to customers (OpenAI) on fragile systems that need "Digital escorts" to keep working.

[-]

larrymcp@reddit

Yea, I had ChatGPT summarize it from all 7 parts: here

For some reason it positions itself at the bottom when you visit the link. You have to scroll to the top

[-]

goomyman@reddit

Eh. I worked on azure stack. The original versions failed so bad we issued refunds. The shipped version missed dates by 2 years and delivered half the promise.

The entire ecosystem is a mess… but that’s ok. Every ecosystem is, it’s an extremely large company with infinite moving parts.

And Microsoft is willing to throw time and money at problems, but they do eventually steer the ship.

Azure works, it has a very good uptime. Could it be more efficient, yes. But so can everything once you look at the backend, but coordinating those large architectural changes is complex and you have to start somewhere.

This is not why Microsoft lost a trillion dollars in market cap. It lost it because the market cap of big tech is tied to AI succeeding and Microsoft true or not is perceived tied to Open AI who right now is deeply unpopular and broke. And Google has beaten them, there won’t be multiple winners.

Also Microsoft is associated with AI slop - this one is actually Microsoft’s fault. I worked there when they mandated all Microsoft products no matter what integrate AI or you’ll get a bad perf review. This is how you end up with AI in notepad. I had a product where AI made no sense - was told to integrate anyway and had several brain storming sessions on what to add. We added AI slop - or more like pointless AI.

[-]

Takeoded@reddit

Could it be more efficient, yes

8 VMS per physical node is atrocious

[-]

goomyman@reddit

When I started at azure they would deploy 3 VMs when you requested one because so may would fail. The first one to succeed would be given to you.

[-]

StrawberryCoup@reddit

I kind of got the impression that what you write would be closer to the truth.

I work in Enterprise as well, with similar problems though maybe not as bad. But I have never had significant work I've done rejected, as the OP described happening many times. IMO that signals a failure in the "political" aspect of the job. And poor politics would explain a lot else in the article, like not getting replies or being heard when voicing concerns.

[-]

goomyman@reddit

Half of OPs article was how amazing his resume was. I mean I don’t doubt that Microsoft has impossible goals. That’s common honestly. You have to have “big ideas” to get leadership roles. You don’t get these roles promising minor changes - you get it with a big idea and the confidence to convince others of it. It’s all software at the end of the day - most of this stuff is possible with enough money - which MS has and is willing to spend.

It usually fails on time or just chasing users that everyone knows will never come - like Microsoft’s late attempts at social media, and phones. Everyone knew those things would never get the adoption rates needed to match spend but when you have 100,000 employees you have to have big projects.

Is it a giant waste of money - yes, but stock price isn’t based on saving money at these companies - it’s based on spending massive amounts of money chasing the next trillion dollar thing.

[-]

twnznz@reddit

This reads to me as an indictment of the architecture team (I'm adjacent; a network architect).

[-]

teknikly-correct@reddit

Due to Washington state WARN rules, Microsoft could not eliminate every tester position; hundreds remained.

Fast forward, and large parts of Azure operations were being run by these former testers. Many were dedicated colleagues, but the shift left gaps in architectural depth for mission-critical systems.

Wow, I didn't realize how many SDETs got converted - I thought they were straight up let go. Instead MS "retrained" them to become full engineers, essentially skipping the natural career path into those roles.

[-]

OffbeatDrizzle@reddit

I hope they were engineers that got forced into tester roles, and not the other way around....

[-]

TripleFreeErr@reddit

this guy is a quack.

not a single soul, could articulate why up to 173 agents were needed to manage an Azure node

there are hundreds of “agents” which run on a one time basis to install systems as part of deployment architecture. These agents often amount to pretty simple scripts or programs. They most often run one time per update deployment, or if nodes are repaved. Some install small daemons. It’s called micro service architecture. Guy claims to be some cloud wiz but doesn’t get these basics.

I originally was going to cry fowl at the clear appeal to authority this article starts off with, but decided it would be hypocritical in case I wanted to explain how I’ve been on the Sov sister team of Host OS since 2020, and took ownership of Sov Overlake in 2023. Frankly, i’m not a huge fan of folks who job hop ever 2-3 year that ALSO try to pull rank. I understand the economics, but i’m rarely impressed by their technical skills in practice.

So far reading this article only deepens this prejudice.

[-]

validelad@reddit

I had doubts after reading when he joined right at the beginning of the blog.

He comes off as a ivory tower, knows a lot in theory, type to me. That may not be fair or accurate, but there were a few red flags pointing to that for me. When he joined. Clearly bouncing around between jobs frequently. Talking about things such as 40% test coverage, and 50% of the engineers being junior, as if that isnt just par for the course in major systems like that.

[-]

Kered13@reddit

Talking about things such as 40% test coverage...as if that isnt just par for the course in major systems like that.

I can only provide a single anecdote, but at Google I don't think any systems have such low test coverage. I think 90% test coverage is the standard enforced across Google. And from my experience, the quality of the treats is generally pretty high. You can make major refactors at Google with a high degree of confidence that nothing will break in production is all the unit and integration tests pass.

[-]

wannaliveonmars@reddit

That's just an indicator of a legacy system. Microsoft's codebase is significantly older, and was developed at a time when testing wasn't a practice. This means that the oldest parts were not made to be testable, and usually older means closer to the core/higher importance. Retrofitting unit tests to code that was not made to be unit-testable in the first place is hard, doubly so for kernel code - this is JavaScript we're talking here.

[-]

OffbeatDrizzle@reddit

This means that the oldest parts were not made to be testable, and usually older means closer to the core/higher importance

this is not an argument... if they are that important, retrofitting is not impossible, and at the end of the day you can always start spinning entire components up to test that they work properly, instead of having to test every line of code

[-]

TripleFreeErr@reddit

Yes exactly, But when I got to the end of part 2 and 3 he’s talking about some specific incidents and problems I had personally also experienced as the sov sister team for core os/overlake. This at least tells me there’s some substance to his experience.

[-]

amestrianphilosopher@reddit

Did we read the same article? You lost billions in contracts due to your inability to provide compute to one of the largest actors in the biggest emerging market, wiping out a trillion in market cap. This is not normal.

He outlined exactly what you needed to do. Focus on reducing system complexity and operational costs to create a truly automated and resilient cloud system.

I work on a similar product at my company and I wake up once a year and the issue is resolved in 15 minutes. No zombie VMs, randomly deleted resources, frequent escort requirements. If you think this it’s normal to brush this off at any level of management, you’re frankly deluded.

[-]

TripleFreeErr@reddit

pointing out what’s wrong is easy. Pinging out what’s wrong is not a solution. “Fix it” is not a solution.

[-]

wannaliveonmars@reddit

tbh I feel like the title is clickbait, and should have been "My general bitching about bad code quality I saw in Azure". I doubt any of what he saw actually translates to financial losses, esp. not 1 trillion. And I doubt OpenAI's stocks would have been that much better if Azure's codebase was more stable.

[-]

TripleFreeErr@reddit

Right, microsoft’s isn’t solely responsible for the AI bubble, and the recent stock plunge amounts to a partial price correction.

[-]

planetworthofbugs@reddit

Bro, did you miss the bit where he started rewriting Azure from the ground up? What a solution! 😂

[-]

Outside-Storage-1523@reddit

Looks like you have worked on similar teams, this is great. I have always wanted to read technical refutes like this.

[-]

cogeng@reddit

I work in the space and he's got me convinced. It all seems very credible.

[-]

dxk3355@reddit

For a guy so smart I don’t know why he’s waiting months for a response. He didn’t get a response in the first day or two it’s gone.

[-]

OffbeatDrizzle@reddit

I mean... an e-mail might be unread for a day or 2 any way or a week or 2 if somebody is on leave. It's also not unreasonable to think that things were happening behind the scenes / above his head until they decided to do something / inform him. This is a pretty L take

[-]

LaconicLacedaemonian@reddit

i would expect either a "let's chat" 1:1 within an hour or complete radio silence

[-]

setheliot@reddit

I have never heard this hot take for the downfall of Microsoft

Just months after Satya Nadella became CEO in February 2014, he canceled the dedicated SDET (Software Development Engineer in Test) role, triggering significant layoffs.

Due to Washington state WARN rules, Microsoft could not eliminate every tester position; hundreds remained.

Many of these testers, strong at execution but with limited experience in system design or deep software engineering, were retrained.

Some became data engineers focused on Windows 10 telemetry; others moved into software engineering roles (often down-leveled); and still others landed in lower-impact areas, including Azure OPEX, where they helped keep the lights on through on-call rotations and incident mitigation.

Fast forward, and large parts of Azure operations were being run by these former testers. Many were dedicated colleagues, but the shift left gaps in architectural depth for mission-critical systems.

[-]

albert_head@reddit

Apparently the Road Ahead was a gate heading over a huge cliff. Who knew?

[-]

dukey@reddit

>I submitted several bug fixes and refactoring, notably using smart pointers, but they were rejected for fear of breaking something.

LOL wow .. Anyone that writes any serious modern C++ knows you shouldn't be having naked pointers anywhere in the code base, at least for heap allocated memory.

[-]

FunCoolMatt@reddit

The article really pinpoints the underlying problems involved in a collapse such as this one.

Junior developers with great ambitions, few senior devs because of layoffs, managers lacking visibility, and out-of-touch executives promising unrealistic objectives, leaving the dev team scrambling to complete them.

[-]

wavefunctionp@reddit

There’s a lot of problems, but really get any org of reasonable size and you start getting the leeches trying to work their ways through the ranks by gaming impact. It’s inevitable.

[-]

OffbeatDrizzle@reddit

why do you think there's a massive culture of working somewhere for 2-3 years and then jumping ship? they get to CV build, throw their piece of shit over the wall and then don't have to worry about maintaining it. they walk away with a promotion / salary increase whilst you're stuck trying to explain to management how it's going to take 5 years to fix and migrate the live application, and somehow it ends up being your fault

[-]

B-i-s-m-a-r-k@reddit

Just gonna get worse with how much execs are expecting those ambitious juniors to render seniors obsolete with LLMs

[-]

OffbeatDrizzle@reddit

the suits fail to understand that 2-3 junior devs is not better than or even equal to 1 senior - some things just require knowledge and experience. rote memorisation of data structures gets you nowhere if you can't architect your way out of a paper bag

[-]

amroamroamro@reddit

yea this part to me is quite revealing:

Just months after Satya Nadella became CEO in February 2014, he canceled the dedicated SDET (Software Development Engineer in Test) role, triggering significant layoffs.

Due to Washington state WARN rules, Microsoft could not eliminate every tester position; hundreds remained.

Many of these testers, strong at execution but with limited experience in system design or deep software engineering, were retrained. [...]

Fast forward, and large parts of Azure operations were being run by these former testers. Many were dedicated colleagues, but the shift left gaps in architectural depth for mission-critical systems.

[-]

validelad@reddit

While I somewhat agree. If I recall he critized some of the work as having > 50% junior engineers. Isn't that just normal?

[-]

pdabaker@reddit

If you consider a junior to be less than 3 years of experience, and people in general having 25 year careers, then having mostly juniors sounds pretty ridiculous

[-]

validelad@reddit

I was lumping all engineers into two camps in my head. Senior or Junior. Which may not have been what he meant. If juniors only include people with than 3 years of experience, than yeah, 50% is high

[-]

pdabaker@reddit

Yeah it’s pretty ambiguous and really depends what you mean by Junior. I don’t know the microsoft definition

[-]

Dreadsin@reddit

Teams I’ve been on have had 1-2 juniors, some mid levels, and then a couple seniors and a lead. So pretty consistently, I’d say juniors are like 10%-20% of the team

[-]

validelad@reddit

Thats fair. I guess I was lumping mid levels in with juniors. Depends on his definition of juniors

[-]

zukos_destiny@reddit

Got to part 6 but it was way too much yappin.

[-]

wannaliveonmars@reddit

I asked Gemini to summarize the facts in the story and strip the editorializing. Here is what it came up with:

The "How Microsoft Vaporized a Trillion Dollars" series is a deep dive into Microsoft’s strategic pivots during the late 2010s and early 2020s, specifically focusing on the shift from a "Windows-first" company to a "Cloud-first" entity under Satya Nadella.

Stripping away the author's editorial tone, here are the core facts and events detailed in the story:

The Internal Reorganization (2018) In March 2018, Microsoft underwent a massive structural change. The Windows and Devices Group (WDG) was dissolved. Its engineers and resources were split between two new divisions: Cloud + AI and Experiences + Devices. This effectively ended the era where Windows was the central "hub" of the company's strategy.
The Shift to "Azure-First" The primary factual thrust of the story is the prioritization of Azure (cloud computing) over the local Windows operating system.

The Goal: Move customers from "one-time" license purchases to recurring "per-user" subscriptions (SaaS).

The Result: Internal development shifted focus toward features that drive Azure consumption, such as Windows 365 (Cloud PC) and Azure Virtual Desktop, rather than optimizing the local desktop experience.

The Deprioritization of the "Shell" The series details how the Windows "Shell" (the user interface, Start menu, and Taskbar) lost its dedicated engineering team.

Much of the Windows UI was rewritten using web-based technologies (WebView2 and React) instead of native C++ code.

The author notes that this change led to the "clunky" feel of the Windows 11 Taskbar and Start menu, as they are essentially web-wrappers rather than deeply integrated system components.

Hardware and the Surface Stagnation The story outlines the decline of the Surface line from a "category-defining" product (like the original Surface Pro) to a repetitive, iterative hardware cycle.

Fact: Key leadership and designers (like Panos Panay) eventually left the company.

Fact: Recent Surface releases have focused more on "AI PCs" and NPU (Neural Processing Unit) integration rather than breakthrough industrial design.

The "Trillion Dollar" Argument The "vaporized" claim in the title refers to the author's calculation of opportunity cost.

While Microsoft's market cap has soared due to Azure and AI, the author argues that by neglecting the Windows ecosystem, Microsoft allowed competitors (Apple and Google) to dominate the high-end consumer and education markets.

The story suggests that if Microsoft had maintained Windows' quality while building Azure, the company’s valuation and market dominance would be significantly higher than they are today.

The Rise of "Substrate" and AI The final parts of the story focus on how Microsoft is now funneling all Windows data into the Microsoft 365 Substrate—a massive data layer that feeds their AI (Copilot).

Windows is no longer viewed as a product to be sold, but as a telemetry and distribution node for AI services.

This explains the integration of "Recall" and other AI features that require constant data indexing and cloud connectivity.

Summary of the "Vaporization" Thesis If you ignore the snark, the factual timeline describes a company that successfully transitioned to the cloud at the expense of its legacy software quality. Microsoft chose to trade "Desktop Dominance" for "Cloud/AI Dominance," a move that was financially successful but resulted in a fragmented, web-heavy, and advertisement-filled operating system.

Not sure if everything is correct, but at least I could read it, while I gave up around the "I wasn’t new to Azure, having run..." and "I wasn’t new to Microsoft either, having been..." paragraphs (hats off to the author for not starting a third paragraph with "I wasn't new").

P.S. I probably should have run it through Github Copilot instead, haha

[-]

Sea-Specific-6890@reddit

Dude please don't do this. Let our comments on Reddit be written by humans. It is one of our last stands against the clankers. Read it, summarize it yourself, talk about it as a human. Let's not have AI summarize content for us. Let's read. Let's talk, human to human.

[-]

wannaliveonmars@reddit

Sorry mate, I just didn't have the patience. What can I say, I'm lazy sometimes...

[-]

dezsiszabi@reddit

Then just don't comment? Novel idea, I know...

[-]

RealAmaranth@reddit

Did I skip a section or did the AI just make up like 3/4 of that summary? I don't remember it mentioning anything about Windows 11, Shell, Surface, etc at all.

[-]

wannaliveonmars@reddit

You're right, there is no Win11 now that I looked at it. Well, I guess AI still has some way to go. Maybe I should try reading the whole thing. It seems to get more interesting from chapter 3.

[-]

drislands@reddit

Just finished reading the full series. Damn.

I used to work for a company that does DR for virtualized environments, and I remember supporting Azure being a huge pain....this seems to explain why, at least in part.

[-]

zeno@reddit

Seems like a tale as old as time: Business priorities trump technical excellence, time to market beats stability, and a technically-focused engineer complains to decision makers and gets shut down. I can only guess but it's possible that spending more time solving engineering problems would have made Microsoft lose contracts. Of course a poor product will also lose customers too but Microsoft's history shows me that time-to-market is more important than quality.

[-]

1RedOne@reddit

I don’t know why someone would publish a thing like this , it’s pretty scorched earth imho and I don’t think it would help in their future job hunt

[-]

Sea-Specific-6890@reddit

Some people care more about the truth and sharing their experiences than money. Also this guy worked on original Azure, if he didn't sell his stock in Microsoft he's rich. despite everything he said Microsoft has been insanely successful (revenue wise, pure corporate accounting) since Ballmer left.

[-]

zeno@reddit

I see less truth telling and more resentment for not being given more responsibilities to "right the ship". Despite the chaos described, it seems to be in Microsoft's DNA to ship half-baked products, prioritizing delivery speed over quality. This strategy has gotten them to be a trillion-dollar company, despite the obvious quality issues we encounter every day with their products and services.

[-]

letmewriteyouup@reddit

Looking at his own career descriptions in the articles he might be nearing retirement age. Nevertheless he may not get upper management roles but with that kind of sheer experience he certainly ain't going to be unemployable

[-]

Outside-Storage-1523@reddit

They probably already earned the FU money.

[-]

Klanowicz@reddit

Do that's why GitHub sucks now...

[-]

idiotiesystemique@reddit

So it's not even vibe coding that killed them

[-]

letmewriteyouup@reddit

I have no idea how much truth this author's specific tale holds, but it is an undeniable fact that Microsoft fumbled OpenAI despite having every deal in its favor, just out of sheer incapability. It is hilarious.

[-]

_verel_@reddit

I'm amazed daily how bad azure dev ops. It's missing absolutely basic ass features while I'm wondering why Microsoft even bought GitHub just to maintain TWO horrible platforms for git

[-]

agumonkey@reddit

Are good engineering shops still around these days ? beside pornhub backend

[-]

float34@reddit

Apple maybe?

[-]

pakoito@reddit

haha no. Never have been.

[-]

_dude_404@reddit

Microslop

[-]

idebugthusiexist@reddit

And the exec who is now running all of Microsoft comes from… being an exec at Azure. Yeh… great company. What’s the saying about Microsoft? They will always do the wrong thing until all options are exhausted and then they will continue doing the wrong thing?

[-]

Desth-Metal@reddit

I tried blazor and then never used MS again

[-]

Party-Stormer@reddit

Wow

The manual part of the whole azure maintenance was unexpected

[-]

surrealerthansurreal@reddit

So I read all 6 parts and idk what to feel. OP kinda comes across as like “I told you so and my stuff was good it was management who failed me” but if this is real then like yeah I totally get that lmao I work in a parallel enough domain that like is this really what’s up at Microsoft? Like I’ve felt the effects in the product suite, but to truly be so out of touch and failing as an org?

[-]

cr1mzen@reddit

Heh, i just finished pushing myAzure repo to Github (over frustration with Azure pipelines). Then i sit down, open Reddit,and this comes up. Shocking.

[-]

amestrianphilosopher@reddit

This gave me flashbacks to my old organization, and filled me with a sense of dread I have not felt in a while. Not quite the same scale, only $1 billion vaporized. I voluntarily left when all my concerns fell on deaf ears. Very similar pattern though. Even attempting to show management that we could reduce OPEX by 99% with a clear plan of execution was met with insane defensiveness, and zero willingness to negotiate.

At some point you have to vote with your feet.

[-]

cosmicr@reddit

I thought we were banning articles related to AI?

[-]

ComfortableTackle479@reddit

that’s not just MS, that’s cultural decay and intellectual decline you can observe everywhere

driven by unlimited greed and lust for power of late stage capitalism

i imagine eye rolling management reaction to this whistleblower escalation attempts, like who gives a fuck when moneys keep flowing

[-]

recycled_ideas@reddit

I'm sure that there's some interesting content here, but I can't find it within the author blowing smoke up their own asshole.

I have no doubt that Microsoft has messed up policies and processes, but this guy is literally so full of themselves that I'm not remotely surprised no one listens to them.

[-]

Humdaak_9000@reddit

"Digital escort strategy".

Oh, my. I wouldn't have been able to continue working after hearing that. I'd be giggling too hard as security rolled me out the door.

[-]

rk06@reddit

this is actually true. we have digital escorts . and they are used more frequently than they ought to be

[-]

Outside-Storage-1523@reddit

I actually read it out in public because I have the habit of reading when...reading. It was a bit embarrassing.

[-]

bzbub2@reddit

interesting read. a little cruel to just air all this dirty laundry...this is not like old history it's like last couple years. but in any case, i can relate having worked on many semi-failing things for a long time lol

[-]

remoteDev1@reddit

the part about firing all the testers and pushing that work onto devs without changing timelines or budgets hit close to home. watched the exact same playbook at my last company - "everyone owns quality now" which actually meant "nobody owns quality now." two years later half the senior engineers were burned out from doing three jobs and management was shocked when things started breaking. then they laid us off because the product wasn't shipping fast enough. can't make this up.

[-]

monsters_from_the_id@reddit

fucking cursed lol

[-]

bluewhackadoo@reddit

Minimum viable product culture killed it. Product managers that dual as people managers has always been the kryptonite that keeps the ass kissing eating away at the best talent from the inside out. So sad

[-]

hearwa@reddit

You can skip about the first 8 paragraphs as it's just the author autofellating himself.

[-]

jacenat@reddit

Fuck. Imma need some popcorn.

[-]

Socrathustra@reddit

I worked on a collection of services predicting hardware demands for Azure (nominally, anyway - I was having a lot of problems because of the pandemic). What I found tracks with this closely: tests were all over the place. Everything was supposed to be "integration tests", a term I use loosely because the integrations were in fact just badly emulated services. On call was a constant flood of issues.

Even in my pandemic related struggles, it was apparent how badly things were going.

[-]

levodelellis@reddit

So it isn't just windows they dropped the ball on. I'm not sure if excel can keep M$ afloat

[-]

o5mfiHTNsH748KVq@reddit

Damn, publicly burning your previous employer is a decision.

That said

and I learned that they had identified 173 agents

What does this even mean. What is an agent in this context? The org created 173 discrete agents to operate as a swarm before swarms really became viable? Or did they have an MCP that's being mischaracterized as agents? Microsoft is known for not understanding how to make an MCP that works - they love overloading them with hundreds of tool calls such that an agent will never actually discover the right tool.

I 100% a believe every word out of this post. However, I think there's some context that will make the situation worse and I crave the tea.

[-]

player2@reddit

This post has nothing to do with LLMs. The word “agent” doesn’t just mean “thing that Claude Code calls”.

[-]

o5mfiHTNsH748KVq@reddit

Great, so the post does not define what an agent is in this context at all, but does talk at length about OpenAI, Anthropic, and Microsoft's misuse of LLMs.

[-]

o5mfiHTNsH748KVq@reddit

I was lost away from home in a bizarre territory where people made plans that didn’t make sense with the aplomb of a drunk LLM.

This made me think they were referring to agents, along with their talk of OpenAI and Claude.

Given that Microsoft is an AI-first company now and is, in fact, on a death march to implement LLM agents top to bottom in every workflow of their business - I think my interpretation was sensible.

[-]

and the delayed features publicly implied as shipping since 2023, before the work even began

What the fuck?

[-]

PepperLuigi@reddit

Microslop

[-]

MrChocodemon@reddit

Is it focus on short term gains? Investor driven decision making?