AWS's inevitable destiny: becoming the next Lumen
(2026/01/27)
- Reference: 1769468407
- News link: https://www.theregister.co.uk/2026/01/26/aws_destiny_lumen_corey_quinn/
- Source link:
For a decade, AWS's position on multi-cloud was clear: don't.
Multi-cloud meant a lowest-common-denominator architecture. Multi-cloud meant giving up the managed services, the security integrations, the features that justified the lock-in. Multi-cloud was for that indecisive tranche of companies that couldn't commit. And until 2019, [1]mentioning other clouds at all was forbidden at AWS conferences.
Then in November 2025, AWS launched AWS Interconnect with Google Cloud as a launch partner, with Microsoft Azure joining in 2026. Just a bit of jiggery-pokery in the console to connect your VPC to a competitor's network, and boom: the thing they spent years telling you was a mistake is now productized and fully managed. AWS didn't change its mind, so much as get told by [2]about 84 percent of its customers to shut up and get with the times.
The pattern
AWS has a tell: it markets hardest wherever it's losing.
2017: Machine learning. SageMaker launched with promises to "democratize ML" through "single-click model training." The keynotes were as breathless as they were incomprehensible. Meanwhile, Google's TensorFlow had already become the industry standard, and AWS was playing catch-up with a product most data scientists ignored. AWS trumpeted "250% user growth" the following year — impressive until you realize they were growing from approximately nobody.
2024-2025: Artificial intelligence. At re:Invent 2024, Bedrock would lead the generative AI revolution, touting a marketplace of one hundred models, multi-agent collaboration capabilities, and Nova foundation models with names that sound like car trim levels. The messaging was unambiguous: the AI future runs on AWS. But by mid-2025, the market had rendered its verdict. Internal documents obtained by [3]Business Insider showed Bedrock hit "critical capacity constraints," pushing customers to competitors. Epic Games took a $10 million Fortnite project to Google Cloud because AWS couldn't provide sufficient quota. Thomson Reuters chose Google for its CoCounsel AI after finding Bedrock 15-30 percent slower. When you're losing eight-figure deals because you can't keep the lights on for your flagship AI product, the marketing isn't the problem.
2025: Multi-cloud. After years of resistance, AWS launched interconnect products with competitors, published guidance on "succeeding with a multicloud strategy," and quietly accepted what the market had been screaming: nobody wants to be locked in.
Each time, the pattern holds. The louder the marketing, the weaker the position.
The real threat isn't who you think
Here's the thing: Azure and Google aren't AWS's real problem.
The real problem is that developers don't choose AWS anymore (because honestly, given a choice between AWS and a vendor who thinks deeply about developer experience, who would?). They choose Vercel, or Netlify, or whatever their AI coding assistant suggests when they type "deploy this." Heck, I have Strong Opinions about this and even I get tired of wrestling with Claude Code's biases; just make the container work already.
[4]
These platforms run on AWS Lambda under the hood. The serverless functions, the edge compute, the CDN — they're AWS infrastructure, abstracted into invisibility. Users never see it, never need an AWS account, never learn what a VPC is. Why would they?
[5]
[6]
[7]One platform comparison put it bluntly: "In 2026, most developers no longer ask 'How do I deploy this?' They ask 'Which button do I click to ship to prod?'"
I don't put my infrastructure on AWS. I put it wherever the tool I'm using wants to put it. Cursor suggests Vercel? Fine. Claude wants Netlify? Sure. The LLM is making the infrastructure decision now, and it's not optimizing for "which cloud provider has the best managed Kafka offering." [8]Vercel charges a 15-20 percent premium over raw AWS pricing. Customers pay it gladly for faster deployments. The margin accrues to the abstraction layer. AWS gets the commodity revenue underneath.
The Lumen problem
Quick: name a Tier 1 internet backbone provider.
If you said Lumen, Cogent, Telia, or NTT, congratulations — you're in the tiny minority of people who think about this. These companies own the undersea cables connecting continents and operate the networks that carry 99 percent of international internet traffic. Nobody thinks about them.
[9]
Because all the interesting stuff rides on top of them. And the margins follow the interesting stuff.
AWS is on that path. Today they're running [10]33-39 percent operating margins — genuinely impressive, making AWS the engine that funds Amazon's everything else. But they're also [11]losing roughly 2 percent market share annually to Azure and Google, and that's just the competition they can see.
The competition they can't see is the abstraction layer that makes them irrelevant to the people actually building software. Every developer who ships via Vercel without knowing what region their Lambda runs in is a developer who will never care about AWS's 347 services. Every AI coding assistant that defaults to "deploy to Netlify" is training the next generation to treat cloud providers like plumbing.
[12]Amazon CEO Andy Jassy goes wobbly on AI bubble possibility
[13]A simple CodeBuild flaw put every AWS environment at risk – and pwned 'the central nervous system of the cloud'
[14]A single DNS race condition brought Amazon's cloud empire to its knees
[15]Amazon brain drain finally sent AWS down the spout
The memory hole
AWS spent 2024 promising an AI future that [16]95 percent of enterprises aren't seeing ROI on , according to MIT's "GenAI Divide" study. They spent 2025 quietly accepting the multi-cloud reality they'd fought for a decade. Now they're hoping you forgot.
Meanwhile, the [17]October 2025 us-east-1 outage took down Snapchat, Fortnite, Ring doorbells, McDonald's mobile ordering, and Britain's tax website for fifteen hours ( [18]six and a half million Downdetector reports ). It was the third major outage in five years from their oldest region. The institutional knowledge keeps [19]walking out the door , and it shows.
[20]
But none of that matters if developers stop choosing AWS directly. In a world where nobody consciously selects their cloud provider, the only time AWS will have any relevance is when they fail. Outages become the brand.
AWS's destiny isn't to lose to Azure or Google. It's to win the infrastructure war and lose the relevance war. To become the next Lumen — the backbone nobody knows they're using, while the companies on top capture the margins and the mindshare.
The cables matter. But nobody's writing blog posts about them. ®
Get our [21]Tech Resources
[1] https://www.businessinsider.com/amazon-eases-restrictions-words-microsoft-google-reinvent-2019-11
[2] https://www.techtarget.com/searchnetworking/opinion/AWS-reInvent-2025-signals-a-shift-in-multi-cloud-networking
[3] https://www.businessinsider.com/amazon-ai-capacity-crunch-pushed-customers-to-rivals-google-anthropic-2025-11
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_offprem/paasiaas&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aXhGe3_y7R55PK-AJ0a8sgAAAMQ&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_offprem/paasiaas&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aXhGe3_y7R55PK-AJ0a8sgAAAMQ&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_offprem/paasiaas&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aXhGe3_y7R55PK-AJ0a8sgAAAMQ&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[7] https://northflank.com/blog/vercel-vs-netlify-choosing-the-deployment-platform-in-2026
[8] https://sacra.com/c/vercel/
[9] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_offprem/paasiaas&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aXhGe3_y7R55PK-AJ0a8sgAAAMQ&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[10] https://www.geekwire.com/2025/amazon-web-services-profits-squeezed-as-ai-arms-race-drives-spending-surge/
[11] https://futurumgroup.com/insights/amazon-q1-fy-2025-earnings-reflect-cloud-momentum-operating-margin-gains/
[12] https://www.theregister.com/2026/01/20/amazon_ceo_andy_jassy_ai_bubble/
[13] https://www.theregister.com/2026/01/15/codebuild_flaw_aws/
[14] https://www.theregister.com/2025/10/23/amazon_outage_postmortem/
[15] https://www.theregister.com/2025/10/20/aws_outage_amazon_brain_drain_corey_quinn/
[16] https://www.theregister.com/2025/08/18/generative_ai_zero_return_95_percent/
[17] https://www.theregister.com/2025/10/23/amazon_outage_postmortem/
[18] https://statusgator.com/blog/aws-least-reliable-region-in-2025/
[19] https://www.theregister.com/2025/10/20/aws_outage_amazon_brain_drain_corey_quinn/
[20] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_offprem/paasiaas&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aXhGe3_y7R55PK-AJ0a8sgAAAMQ&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[21] https://whitepapers.theregister.com/
Multi-cloud meant a lowest-common-denominator architecture. Multi-cloud meant giving up the managed services, the security integrations, the features that justified the lock-in. Multi-cloud was for that indecisive tranche of companies that couldn't commit. And until 2019, [1]mentioning other clouds at all was forbidden at AWS conferences.
Then in November 2025, AWS launched AWS Interconnect with Google Cloud as a launch partner, with Microsoft Azure joining in 2026. Just a bit of jiggery-pokery in the console to connect your VPC to a competitor's network, and boom: the thing they spent years telling you was a mistake is now productized and fully managed. AWS didn't change its mind, so much as get told by [2]about 84 percent of its customers to shut up and get with the times.
The pattern
AWS has a tell: it markets hardest wherever it's losing.
2017: Machine learning. SageMaker launched with promises to "democratize ML" through "single-click model training." The keynotes were as breathless as they were incomprehensible. Meanwhile, Google's TensorFlow had already become the industry standard, and AWS was playing catch-up with a product most data scientists ignored. AWS trumpeted "250% user growth" the following year — impressive until you realize they were growing from approximately nobody.
2024-2025: Artificial intelligence. At re:Invent 2024, Bedrock would lead the generative AI revolution, touting a marketplace of one hundred models, multi-agent collaboration capabilities, and Nova foundation models with names that sound like car trim levels. The messaging was unambiguous: the AI future runs on AWS. But by mid-2025, the market had rendered its verdict. Internal documents obtained by [3]Business Insider showed Bedrock hit "critical capacity constraints," pushing customers to competitors. Epic Games took a $10 million Fortnite project to Google Cloud because AWS couldn't provide sufficient quota. Thomson Reuters chose Google for its CoCounsel AI after finding Bedrock 15-30 percent slower. When you're losing eight-figure deals because you can't keep the lights on for your flagship AI product, the marketing isn't the problem.
2025: Multi-cloud. After years of resistance, AWS launched interconnect products with competitors, published guidance on "succeeding with a multicloud strategy," and quietly accepted what the market had been screaming: nobody wants to be locked in.
Each time, the pattern holds. The louder the marketing, the weaker the position.
The real threat isn't who you think
Here's the thing: Azure and Google aren't AWS's real problem.
The real problem is that developers don't choose AWS anymore (because honestly, given a choice between AWS and a vendor who thinks deeply about developer experience, who would?). They choose Vercel, or Netlify, or whatever their AI coding assistant suggests when they type "deploy this." Heck, I have Strong Opinions about this and even I get tired of wrestling with Claude Code's biases; just make the container work already.
[4]
These platforms run on AWS Lambda under the hood. The serverless functions, the edge compute, the CDN — they're AWS infrastructure, abstracted into invisibility. Users never see it, never need an AWS account, never learn what a VPC is. Why would they?
[5]
[6]
[7]One platform comparison put it bluntly: "In 2026, most developers no longer ask 'How do I deploy this?' They ask 'Which button do I click to ship to prod?'"
I don't put my infrastructure on AWS. I put it wherever the tool I'm using wants to put it. Cursor suggests Vercel? Fine. Claude wants Netlify? Sure. The LLM is making the infrastructure decision now, and it's not optimizing for "which cloud provider has the best managed Kafka offering." [8]Vercel charges a 15-20 percent premium over raw AWS pricing. Customers pay it gladly for faster deployments. The margin accrues to the abstraction layer. AWS gets the commodity revenue underneath.
The Lumen problem
Quick: name a Tier 1 internet backbone provider.
If you said Lumen, Cogent, Telia, or NTT, congratulations — you're in the tiny minority of people who think about this. These companies own the undersea cables connecting continents and operate the networks that carry 99 percent of international internet traffic. Nobody thinks about them.
[9]
Because all the interesting stuff rides on top of them. And the margins follow the interesting stuff.
AWS is on that path. Today they're running [10]33-39 percent operating margins — genuinely impressive, making AWS the engine that funds Amazon's everything else. But they're also [11]losing roughly 2 percent market share annually to Azure and Google, and that's just the competition they can see.
The competition they can't see is the abstraction layer that makes them irrelevant to the people actually building software. Every developer who ships via Vercel without knowing what region their Lambda runs in is a developer who will never care about AWS's 347 services. Every AI coding assistant that defaults to "deploy to Netlify" is training the next generation to treat cloud providers like plumbing.
[12]Amazon CEO Andy Jassy goes wobbly on AI bubble possibility
[13]A simple CodeBuild flaw put every AWS environment at risk – and pwned 'the central nervous system of the cloud'
[14]A single DNS race condition brought Amazon's cloud empire to its knees
[15]Amazon brain drain finally sent AWS down the spout
The memory hole
AWS spent 2024 promising an AI future that [16]95 percent of enterprises aren't seeing ROI on , according to MIT's "GenAI Divide" study. They spent 2025 quietly accepting the multi-cloud reality they'd fought for a decade. Now they're hoping you forgot.
Meanwhile, the [17]October 2025 us-east-1 outage took down Snapchat, Fortnite, Ring doorbells, McDonald's mobile ordering, and Britain's tax website for fifteen hours ( [18]six and a half million Downdetector reports ). It was the third major outage in five years from their oldest region. The institutional knowledge keeps [19]walking out the door , and it shows.
[20]
But none of that matters if developers stop choosing AWS directly. In a world where nobody consciously selects their cloud provider, the only time AWS will have any relevance is when they fail. Outages become the brand.
AWS's destiny isn't to lose to Azure or Google. It's to win the infrastructure war and lose the relevance war. To become the next Lumen — the backbone nobody knows they're using, while the companies on top capture the margins and the mindshare.
The cables matter. But nobody's writing blog posts about them. ®
Get our [21]Tech Resources
[1] https://www.businessinsider.com/amazon-eases-restrictions-words-microsoft-google-reinvent-2019-11
[2] https://www.techtarget.com/searchnetworking/opinion/AWS-reInvent-2025-signals-a-shift-in-multi-cloud-networking
[3] https://www.businessinsider.com/amazon-ai-capacity-crunch-pushed-customers-to-rivals-google-anthropic-2025-11
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_offprem/paasiaas&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aXhGe3_y7R55PK-AJ0a8sgAAAMQ&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_offprem/paasiaas&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aXhGe3_y7R55PK-AJ0a8sgAAAMQ&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_offprem/paasiaas&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aXhGe3_y7R55PK-AJ0a8sgAAAMQ&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[7] https://northflank.com/blog/vercel-vs-netlify-choosing-the-deployment-platform-in-2026
[8] https://sacra.com/c/vercel/
[9] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_offprem/paasiaas&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aXhGe3_y7R55PK-AJ0a8sgAAAMQ&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[10] https://www.geekwire.com/2025/amazon-web-services-profits-squeezed-as-ai-arms-race-drives-spending-surge/
[11] https://futurumgroup.com/insights/amazon-q1-fy-2025-earnings-reflect-cloud-momentum-operating-margin-gains/
[12] https://www.theregister.com/2026/01/20/amazon_ceo_andy_jassy_ai_bubble/
[13] https://www.theregister.com/2026/01/15/codebuild_flaw_aws/
[14] https://www.theregister.com/2025/10/23/amazon_outage_postmortem/
[15] https://www.theregister.com/2025/10/20/aws_outage_amazon_brain_drain_corey_quinn/
[16] https://www.theregister.com/2025/08/18/generative_ai_zero_return_95_percent/
[17] https://www.theregister.com/2025/10/23/amazon_outage_postmortem/
[18] https://statusgator.com/blog/aws-least-reliable-region-in-2025/
[19] https://www.theregister.com/2025/10/20/aws_outage_amazon_brain_drain_corey_quinn/
[20] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_offprem/paasiaas&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aXhGe3_y7R55PK-AJ0a8sgAAAMQ&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[21] https://whitepapers.theregister.com/
disk iops
hmm, so thousands of low-talent, low-skill, low-IQ H1b taking away American jobs doesn't result in the best infra and software products? whodathunk!
(yes, I've rubbed shoulders with plenty of the H1b in the Seattle offices thanks, so my observation is first-friggin' hand, K?)
IMO Amazon should jettison all the "value-add" and focus on the essential services. Let someone else offer fancy-pants services that are transparently utilizing underlying SQS/EC2/Lambda et. al.
AWS ops/devs are spread too thin. They can't/won't kick customers out of us-east or close the damn door. Sorry, it's no longer a permitted region in which to deploy. We saw this being a giant problem over 10 years ago but Sales was more interested in not ruffling customer feathers then facing the damn music. Seriously, fk Sales people. They're 2nd in line behind the lawyers.
I ran a service that used Textract for financial forms. This year the powers decided to use Google's Gemini AI to do OCR. It's so bad at it, my chicken can do better. And yet, we're supposed to soldier on because it's 3x the cost of Textract. Huh?
GCP and Azure like to think they have parity with AWS on the essential building blocks, but they really don't, IMO. GCP is just fickle and obtuse. Azure well, their VPC networking is just atrocious to manage.
Late stage
elsergiovolador
This article is pointless. All three cloud services have the same owners. The market share doesn't matter.
You have illusion of choice.
A dumb pipe
IGnatius T Foobar !
All we've ever wanted, whether it's a telecom provider or a cloud provider, is a dumb pipe. Give us the infrastructure and then get out of the way.
Karma
retiredFool
I imagine Amazon makes more on the Chinese products they sell than the Chinese manufacturer makes. Justice if someone does it to Amazon.
runt row raggy
you can look the author up as QuinnyPig on x where the tagline is "Cloud Economics and AWS shitposting, which are the same thing
i am disappointed this "special to el reg" is not more clearly called out according to the author's self-professed intent.
They'll only care when the margins change
I believe this. I also think that if Amazon could make as much money being a Lumen without having to do the trade shows and Senate hearings I think they'd be fine with it. However the hype is what brings the premium rates, so I don't think they can turn away from this fight.
And it's a fight they're losing. My company is only choosing them for basic building blocks that we can get *very* cheaply. But any service that's more complicated than EC2 with some S3 is coming from other places. We'll migrate a bunch of small stupid boxes up there, but our future is elsewhere. We're not huge, but we're big enough for other vendors to court us and our business in ways that Amazon never has. And once that settles down and we're taken for granted we will still be on a platform that feels a need to hustle to stay in the game more than the moribund AWS.